ERDAS Image Compressor

User Guide

14.00.0001
May 2014

Copyright
Copyright 2014 Intergraph Corporation. All Rights Reserved.
Including software, file formats, and audiovisual displays; may be used pursuant to applicable software license agreement; contains
confidential and proprietary information of Intergraph and/or third parties which is protected by copyright law, trade secret law, and
international treaty, and may not be provided or otherwise made available without proper authorization from Intergraph Corporation.

U.S. Government Restricted Rights Legend

Use, duplication, or disclosure by the government is subject to restrictions as set forth below. For civilian agencies: This was developed
at private expense and is "restricted computer software" submitted with restricted rights in accordance with subparagraphs (a)
through (d) of the Commercial Computer Software - Restricted Rights clause at 52.227-19 of the Federal Acquisition Regulations
("FAR") and its successors, and is unpublished and all rights are reserved under the copyright laws of the United States. For units of the
Department of Defense ("DoD"): This is "commercial computer software" as defined at DFARS 252.227-7014 and the rights of the
Government are as specified at DFARS 227.7202-3.
Unpublished - rights reserved under the copyright laws of the United States.
Intergraph Corporation
P.O. Box 240000
Huntsville, AL 35813

Terms of Use
Use of this software product is subject to the End User License Agreement ("EULA") delivered with this software product unless the
licensee has a valid signed license for this software product with Intergraph Corporation. If the licensee has a valid signed license for
this software product with Intergraph Corporation, the valid signed license shall take precedence and govern the use of this software
product. Subject to the terms contained within the applicable license agreement, Intergraph Corporation gives licensee permission to
print a reasonable number of copies of the documentation as defined in the applicable license agreement and delivered with the
software product for licensee's internal, non-commercial use. The documentation may not be printed for resale or redistribution.

Warranties and Liabilities

All warranties given by Intergraph Corporation about equipment or software are set forth in the EULA provided with the software or
applicable license for the software product signed by Intergraph Corporation, and nothing stated in, or implied by, this document or its
contents shall be considered or deemed a modification or amendment of such warranties. Intergraph believes the information in this
publication is accurate as of its publication date.
The information and the software discussed in this document are subject to change without notice and are subject to applicable
technical product descriptions. Intergraph Corporation is not responsible for any error that may appear in this document.
The software discussed in this document is furnished under a license and may be used or copied only in accordance with the terms of
this license. No responsibility is assumed by Intergraph for the use or reliability of software on equipment that is not supplied by
Intergraph or its affiliated companies. THE USER OF THE SOFTWARE IS EXPECTED TO MAKE THE FINAL EVALUATION AS TO THE
USEFULNESS OF THE SOFTWARE IN HIS OWN ENVIRONMENT.
Intergraph is not responsible for the accuracy of delivered data including, but not limited to, catalog, reference and symbol data. Users
should verify for themselves that the data is accurate and suitable for their project work.

Trademarks
Intergraph, the Intergraph logo, and GeoMedia are registered trademarks of Intergraph Corporation. Microsoft and Windows are
registered trademarks of Microsoft Corporation. Bing is a trademark of Microsoft Corporation. Google Maps is a trademark of Google
Incorporated. Pictometry Intelligent Images is a registered trademark of Pictometry International Corporation. ERDAS, ERDAS
IMAGINE, Stereo Analyst, IMAGINE Essentials, IMAGINE Advantage, IMAGINE, Professional, IMAGINE VirtualGIS, Mapcomposer,
Viewfinder, Imagizer, LPS, and ERDAS APOLLO are registered trademarks and exclusive property of Intergraph Corporation.
Other brands and product names are trademarks of their respective owners.

Contents
Whats New .................................................................................................................. 4
Overview ...................................................................................................................... 5
Architecture............................................................................................................. 6
Supported Environments ........................................................................................... 7
Installation Guide ........................................................................................................ 8
Windows ................................................................................................................. 8
Uninstallation .................................................................................................. 11
Linux ..................................................................................................................... 12
Uninstallation .................................................................................................. 15
Licensing.................................................................................................................... 16
Overview ............................................................................................................... 16
Installation & Configuration ................................................................................... 17
Compressor Usage ................................................................................................... 18
Wizard mode......................................................................................................... 18
Command-line mode ............................................................................................ 23
Input ............................................................................................................... 24
Output............................................................................................................. 24
Options ........................................................................................................... 25
Region Options ........................................................................................ 31
Advanced Options ................................................................................... 33
Reporting example ......................................................................................... 35
Usage examples ............................................................................................ 36
XML Project File ................................................................................................... 37
Example mosaic.xml ...................................................................................... 37
Supported input formats ....................................................................................... 39
Null block analysis ................................................................................................ 41
Comparison .................................................................................................... 42
FAQ ................................................................................... Error! Bookmark not defined.
Support....................................................................................................................... 50

ERDAS Image Compressor User Guide

Whats New

Mosaicking now supports multiband input, multiband output. Previously only RGB/RGBA
output was supported in 14.0
Mosaicking now supports greater than 8bit output for ECW v3 and JPEG2000 files
Parsing of mosaic projects with thousands of datasets now 10 times faster
Overriding of SRS information now possible in Mosaic XML as a new output option
Added support for MBTile, NITF and Terrashare RBD file input
12-bit JPEG compressed file types now supported
Multi-part polygon regions now supported (e.g. polygons with holes)
Further optimizations and bug fixes

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Overview
ERDAS Image Compressor is a geospatial image compression application designed to simplify the
creation of ECW and JPEG2000 data. The application provides,

Cross-platform, 64-bit support for both Windows and Linux

Command-line and simple wizard based user interfaces
A stand-alone, decoupled tool to plugin to existing data processing workflows
A cost effective solution to simplify format compression for existing Intergraph Geospatial
customers

The products functional scope is intentionally narrow. The ERDAS Image Compressor does not,

Support 32-bit platforms

Offer advanced mosaicking functions
o You cannot feather, blend, color-balance, dodge, adjust seam-lines, preview
output or perform many other tasks typically performed in a mosaic process
Resample or resize image resolution or cell size
Reproject or warp imagery to different output coordinate systems

The Image Compressor can be seen therefore as a transcoder only of input data. For other image
processing tasks, ERDAS IMAGINE 2014 remains the recommended tool for the full processing and
compression workflow.

Figure 1 - Wizard compression interface

ERDAS Image Compressor User Guide

Architecture
The ERDAS Image Compressor can be broken into input readers, compressor logic and output
writers. The input readers ensure wide industry format support by leveraging the GDAL and
ERMapper libraries. For transcoding of existing ECW or JPEG2000 datasets, there is a direct rendering
path where the ECWJP2 SDK is used both as the input and output pathway to ensure the fastest
throughput.
The compressor tier has the thread, memory, region and processing logic to ensure the input data is
passed as quickly as possible from the input readers to the output writer. Its also responsible for the
spatial intersection tests when passing in vector regions to denote null block or opacity band areas.

Figure 2- High level component architecture

The application is thread-safe1 and has been deployed on processing workstations up to 32-cores
with good CPU scaling using the new parallel ECW encoder. Peak performance has been recorded at
120MB/sec encoding a 10 gigapixel image and far exceeds the throughput of other third-party
applications that implement the ERDAS ECWJP2 SDK due to optimizations implemented in the Image
Compressor architecture.
The provided user interface is a thin wrapper around the command-line Image Compressor
executable. For advanced users it is expected that most will interface with the command line tool to
give more control integrating with existing workflows.
Note: Compression throughput speeds vary and are heavily dependent on hardware and the input
data format type and internal structure.

Not all GDAL supported formats are Thread-safe. See FAQ.

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Supported Environments
= One of the operating systems is required.
= One of the database server engines is required.
= One of database client engines is required.
= One of the internet browsers is required.
= One of the third party products is required.
= One of the Intergraph products is required.
R = Required
O = Optional
U = User must install
A = Automatically installed by Intergraph setup
ERDAS Image Compressor
Operating Systems

Windows 7, 64-bit
Windows 8 & 8.1, 64-bit
Windows Server 2008 & 2008 R2 64-bit
Windows Server 2012 & 2012 R2 64-bit
Red Hat Enterprise Linux 6.x, 64-bit
CentOS 6.x, 64-bit

Note: Other Linux distributions such as Debian, Ubuntu, Mint, Fedora and OpenSuSE are considered
viable platforms and may require additional installation requirements.

ERDAS Image Compressor User Guide

Installation Guide
Windows
1. Download and execute the installation package from the product website

2. The installation wizard will then start,

3. Default target location is C:\Intergraph\Image Compressor 2014\" but can be changed. 170
MB is required

ERDAS Image Compressor User Guide

ERDAS Image Compressor

4. There is currently no optional components as part of the install

5. The Intergraph SG&I EULA terms must be accepted to continue

ERDAS Image Compressor User Guide

6. Start menu shortcuts will be created by default under the following group,

7. The installation will finally unpack and deploy the required files

10

ERDAS Image Compressor User Guide

ERDAS Image Compressor

8. The installation is complete.

The Image Compressor can executed via command line from
C:\Intergraph\Image Compressor 2014\imagecompressor-cli\bin\ImageCompressor.exe
or via the Start Menu shortcut to launch the wizard user interface

See Licensing and Usage Chapters for next steps.

Uninstallation
To remove or uninstall, use the standard Windows Add/Remove Programs process. For
convenience there is also a shortcut within the Start Menu folder.

ERDAS Image Compressor User Guide

11

Linux
1. Download and extract the installer giving executable rights to ERDAS-ImageCompressor_2014.bin
2. Execute the bin file,

3. The installation wizard will then open, mirroring the steps found in the Windows version

4. Default installation location is /usr/local/Intergraph/image-compressor/. Ensure the

installing user has appropriate rights to write to this location

12

ERDAS Image Compressor User Guide

ERDAS Image Compressor

ERDAS Image Compressor User Guide

13

5. The installation is complete.

The Image Compressor can executed via command line from
/usr/local/Intergraph/image-compressor/imagecompressor-cli/bin/ImageCompressor.sh
or via the Application shortcut in the Graphics category on KDE and Gnome environments.

14

ERDAS Image Compressor User Guide

ERDAS Image Compressor

See Licensing and Usage Chapters for next steps.

Uninstallation
To remove or uninstall, execute the UninstallImageCompressor.sh script located in
/usr/local/Intergraph/image-compressor/ and Remove all components when prompted.

ERDAS Image Compressor User Guide

15

Licensing
Overview
The ERDAS Image Compressor is designed to complement existing Intergraph Geospatial 2014
products. The tool uses host based licensing to validate license availability in accordance with the
following table,

Product Family

Required License Level

Version

ERDAS APOLLO

Essentials

14.0

Advantage

14.0

Professional

14.0

ERDAS IMAGINE

Professional

14.0

GeoMedia Smart Client

Essentials

14.0

Advantage

14.0

Professional

14.0

Professional

14.0

GeoMedia WebMap

In the event node-locked or concurrent licenses are unavailable for any of the above products and
versions the compressor will fail.
Host based licensing ensures that if any of the above feature codes are already acquired on the same
machine the Image Compressor will validate but not check out an additional license. For example,
the Image Compressor and APOLLO Essentials can coexist under one 2014 license when run on the
same machine at the same time.
The Image Compressor user interface and command line tools will display the licensing status to help
identify which license was acquired. Where multiple concurrent licenses are available from a central
licensing server, it is advised to borrow licenses explicitly to prevent situations where Image
Compressor may acquire a license required by other users.

Figure 3 - License status in UI and Command line

16

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Once a license is acquired Image Compressor allows unlimited compression of input files in terms of,
1.
2.
3.
4.

The physical input size in gigapixels

The type of data, uint8, uint16 etc.
The amount of data over time
The number of concurrent threads that can be executed per job

Installation & Configuration

Refer to the Intergraph SG&I Licensing Portal for installation instructions and troubleshooting
information for the Intergraph Licensing server. Windows and Linux licensing information is
available.
For most users with an existing product, no additional licensing steps should be required as the
dependant product license codes should already be configured.

ERDAS Image Compressor User Guide

17

Compressor Usage
Wizard mode
1. The initial page guides users through the 4 step compression process. It also provides license
status information and about dialog for versioning and license information.
Note: If a license was not found, the Next button will be disabled.

2. The I icon launches the About dialog below to review license, build information and other
relevant help links.

3. Returning to the main wizard page,

a. For single image or set of images the Image Compressor will output the same
number of input files and provides a very simple method of compressing thousands
of images in batch
b. For Mosaic a set of images, the Image Compressor will combine all selected images
into a single output image. This mode was designed for tiled, preferably adjoining
images.
18

ERDAS Image Compressor User Guide

ERDAS Image Compressor

4. Regardless of the wizard selected, the next step is to select the input images.

a. Add images allows individual file selection using the default image filter of
common input file types (*.ecw, *.jp2, *.tif, *.alg, *.ers, *.img, *.vrt) or All files
b. Add folder can be used in conjunction with the Recurse directories option to add
thousands of files within a directory.
c. If mosaic wizard was selected, the order in which the datasets are displayed in this
list box represents the display order of the output mosaic where datasets displayed
at the top will be rendered on top of the output image. You may use the
contextual right click menu or drag and drop to reorder the display order.

5. Standard compression options are now presented.

ERDAS Image Compressor User Guide

19

a. The output filename and directory can be defined using standard file locations or
with templates.
b. The options dialogs match to the command line equivalent. See Command Line
Options chapter below for more information.
6. Opacity and null block options is the final configuration step.

7. A summary page is shown to confirm the selected settings.

20

ERDAS Image Compressor User Guide

ERDAS Image Compressor

8. The progress dialog will then execute the jobs in sequential order. Executing jobs in parallel
is not supported by the UI but can be achieved through direct command line access.
9. Press the Start button to begin the compression jobs.

a. During compression the Status bar will progress relative to the current input image.
Additional metrics are also reported,
i. Null Blocks: The number of blocks written to an ECW v3 file tagged as null.
This number will always show 0 unless a region is passed in and null blocks
are enabled.
ii. Blocks: The number of regions that have been compressed and written to
disk. One block contains all compressed data for all bands for one region,
i.e. for three bands there will be three precincts per block.
iii. Precincts: The number of regions that have been processed by the wavelet
filter. Each precinct is per band.
On the completion of each input image, two actions are supported
iv. Log: to review the standard log output including any errors or performance
information
v. View: will open the output file in the registered application handler such as
ERDAS IMAGINE or ER Viewer to review the output result
ERDAS Image Compressor User Guide

21

When an error occurs, the UI will report the error and only allow the Log to be
reviewed.
At any time during batch compression, the Stop button will halt the current
processing job. Pressing Start will then restart the batch project from the next
image in the list

22

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Command-line mode
The command-line mode offers fine-grain control over the compression process. Where possible
defaults are set to ensure that in many cases only the <input> and <output> file is required however
careful consideration of all options is recommended. When the user interface is used, the output
compression log includes the command line options provided.
Note: The available options are identical for both Windows and Linux versions of Image Compressor.

Figure 4 Image Compress Command Line Usage information

ERDAS Image Compressor User Guide

23

Input
For supported input format types see Supported Input Formats chapter. To compress any of these
formats pass in the file location and the Image Compressor will attempt to load it via one of the
supported Input Readers. If reading fails the compressor will exit.
Note: File location should be enclosed in quotes if path may contain spaces.

Output
Image Compressor supports writing ECW and JPEG2000 files where the type is defined by the output
file extension. To write JPEG2000 the output file should be specified as C:\output\compressed.jp2 or
for ECW C:\output\compressed.ecw. Based on the file type different compression options will
become available according to the following format capabilities.

Capability

ECW v2

ECW v3

JPEG2000

Line compression
Tile compression
8-bit unsigned
16-bit unsigned
16-bit signed
Visually lossless
Numerically lossless
Null block support
Opacity band support
Data statistics, histogram
RPC storage
Custom metadata
Georeferencing

GDT

GeoTIFF Tags

GML in JP2, GeoJP2

Colour space support

Greyscale, RGB,
Multiband

Greyscale, RGB,
Multiband

Greyscale, RGB,
Multiband

Largest known image3

14 terapixels

14 terapixels

756 gigapixels

Partial. Custom metadata can be written to JP2 UUID boxes however clients will have to explicitly
understand the presence and structure of the contents.
3

24

Compressed using ERDAS Image Compressor, as at January 2014.

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Options
-method line|tile
The Image Compressor implements two types of compression algorithms.
Line represents a scan-line based reader that reads across each scan-line, compresses and
continues to the next scan-line until the end of file. This approach has been used since ECWs
inception.

Line is inherently single-threaded and cannot scale across multiple CPU cores however it benefits
from lower memory requirements to compress. Line is particularly suited for scan-line structured
input data such as Strip TIFF or file formats that do not perform well with multi-threaded readers.
Note: Line must be used for the compression of JPEG2000 files.
Tile is a new parallel algorithm introduced in the ERDAS ECWJP2 SDK v5 that reads input data in
discrete tiles across multiple reader threads specified by the threads option. Each thread processes
independently in a thread-pool across the width of the dataset and is then repeated down the
image.

ERDAS Image Compressor User Guide

25

The new algorithm results in scaling across CPU Cores however it requires more memory to
compress the same input as line. There is a further trade-off with Disk I/O as the concurrent
threads increase load and requires more data to be processed than line, increasing the likelihood of
reaching a disk bottleneck. Its also possible that the input Data Readers have not been optimized for
multi-threaded reading creating additional bottlenecks. See FAQ for formats that are not threadsafe.
Fast I/O is a requirement in order to feed data to the multiple worker threads otherwise CPU
Utilization will be low and performance may be slower than the line algorithm. Where Data I/O is
sufficient, the tile algorithm can be more than 400% faster than line depending on hardware and
input format used.
Both algorithms are suited for different situations. In order to determine the ideal method for your
situation, benchmarks should be performed particularly where compression speed is an important
measure.
Note: Output ECW files are binary identical irrespective of the compression method used.
-listinputformats true
Retrieves list of supported input formats across all data readers. See Supported Formats chapter for
a formatted version. This option is mutually exclusive from all others.
-targetrate 15
The target compression rate is expressed as a ratio, where 15 represent 15:1 or in other words a
94% size reduction. The compressor uses this target to maintain image compression quality which
can result in the Actual Compression Ratio being higher or lower than the provided value. This is
expected and is commonly seen on datasets with nodata areas or other areas with spectral
characteristics that are highly compressible without sacrificing quality. A 15:1 target compressed

26

ERDAS Image Compressor User Guide

ERDAS Image Compressor

image always produces comparable image quality to any other 15:1 target image even though actual
rate may vary.
For RGB or Multiband input, 15:1 to 20:1 target compression rates produce visually lossless output
results. For greyscale, 10:1 is recommended. See the Analyzing the ECW Lifecycle Whitepaper for
more information.
Note: To compress numerically lossless JPEG2000, specify a targetrate value of 1. Any other value
will revert to the lossy compression type. ECW does not support numerically lossless compression so
a targetrate of less than 5 is not recommended.
-opacityband 4
The compressor will by default attempt to retain any opacity or alpha band present in the input file.
This option can be used to force the detection of the opacity channel if it cannot be determined
automatically from the input reader. For example, where an input 4-band image is tagged as Bands #
0 to 3, where Band #3 is actually a mislabelled opacity band.
-version 2 | 3
The ECW file format has two versions available. ECW v2 is the legacy format with the widest industry
support and ensures interoperability with all existing ECW software. ECW v3 was introduced in 2012
and adds new capabilities such as null blocks, improved metadata storage and support for uint16 cell
types.
Defaults to ECW v2. This option is ignored for JPEG2000 output.
-tempdir C:\temp\
As part of every compression process the Image Compressor must maintain a certain amount of
intermediate data on disk before reassembling to form the final output format. These intermediate
files are all stored in the tempdir location that defaults to the system TEMP environment location.
For performance reasons its strongly recommended to write temporary data to a different physical
disk than the input or output drives to maintain throughput. Users should allocate the same amount
of storage at tempdir as at the final target destination. For example, if a 10 GB ECW output is
estimated then at least 10 GB is required at the tempdir location.

Figure 5 Example ECW temp files written to tempdir

Note: The Image Compressor manages the files written to this location. On completion of all
compression jobs the temp files will be deleted however if the compression process is interrupted
temporary files may be left on disk. If no compression is currently running it is recommended to
delete any orphaned temp files.
ERDAS Image Compressor User Guide

27

-threads 4
The number of threads used when compressing via the tile compression method. Defaults to the
detected CPU Core count but can be adjusted to determine ideal concurrency level. Like memcache
using all threads may not yield best performance particularly if Disk I/O is inadequate. On large
deployments with 32 cores or more, optimal throughput may be found setting threads to 8 or 16.
Testing is always recommended.
Value is ignored when method is set to line as it will always be one. See method above for more
information.
-memcache 0.25
Allocate a percentage of total System Memory to a memory pool for the compression process. The
image compressor will calculate the required memory for the Write Memory Cache based on the
input image dimensions and compression method specified (see table). Any remaining memory in
the pool will then be assigned to the Read Memory Cache.
On a machine with 32 GB System Memory for example, a memcache of 0.7 can be visualized as
follows,

If the same input image was then compressed on a machine with only 16GB System Memory, the
same memcache value of 0.7 would yield,

Which would result in an error since the fixed Write Memory Cache requirement of 15gb is larger
than the whole memory pool budget (16gb * 0.7 = 11.2 gb). To remedy, a memcache value of 0.95
(15.2 GB) is required which would require careful management to ensure the system does not begin
to swap to pagefile. Adding additional memory would be recommended in this situation or switching
to the line compression method to reduce the memory requirement but potentially sacrificing
throughput performance.
To compress a smaller hypothetical image on the 32 GB machine with the same memcache value of
0.7, the write cache is reduced and the read cache size increases dramatically to take the remainder
of the memory pool.

28

ERDAS Image Compressor User Guide

ERDAS Image Compressor

The smaller image will now also be able to be run on the 16gb machine as the Write requirements no
longer exceeds the memcache budget,

The size of the Read Memory Cache is secondary and increasing this value does not always yield an
increase in performance. In other words, if a memcache of 0.25 is sufficient to compress your input
image it is not always advised to specify memcache of 0.9. It is dependent on the type of input
format, how well that dataset driver responds to a higher read memory cache and disk I/O
performance. In some cases performance could drop given additional overheads managing a large
cache verse simply reading data from disk.
It is always recommended to leave a memory buffer to ensure long running compression jobs do not
reach 100% Memory usage. Image Compressor can fluctuate within a degree of tolerance of the
memory budget before purging, so assigning 99% memory is never advised. This buffer size also
depends on other applications that may be running on the machine.
The following table highlights the fixed write memory cache requirements for different input
characteristics.

Input image size

Bands

Gigapixels

Output
Format

Compression
Method

Write Memory
Cache (MB)

50,000 x 50,000 px

2.5

ECW

Tile

234

2.5

ECW

Line

143

10

ECW

Tile

416

10

ECW

Line

234

50

ECW

Tile

1,872

50

ECW

Line

962

50

ECW

Tile

4,906

50

ECW

Tile

416

50

ECW

Line

234

50

JPEG2000

Line

202

50

ECW

Tile

1,023

250

ECW

Tile

1,873

250

ECW

Line

963

250

JPEG2000

Line

801

250

ECW

Tile

4,909

250

JPEG2000

Line

2,048

1000

ECW

Tile

3,695

100,000 x 100,000 px

500,000 x 100,000 px

100,000 x 500,000 px

500,000 x 500,000 px

1,000,000 x 1,000,000 px

ERDAS Image Compressor User Guide

29

Input image size

Bands

Gigapixels

Output
Format

Compression
Method

Write Memory
Cache (MB)

1000

ECW

Line

1,873

1000

ECW

Tile

9,102

Note,

30

the drop in memory when using line over tile compression method
the drop in memory when the width of the input changes, despite identical gigapixel value
the large increase in memory requirements moving from 3 to 8 bands
the small memory drop using our default JPEG2000 profile over ECW in line mode
although not shown, write memory requirements do not change depending on the
targetrate. A target of 10:1 will require the same memory as 50:1.

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Region Options
-opacity 0 | 1 | 2
The opacity value defines the behaviour of the input region defined by datasetregionfile, -shapefile
and wktfile.

-opacity 0 : opacity band only

-opacity 1 : null blocks only (ECW v3 only)
-opacity 2 : opacity and null blocks (ECW v3 only)
-opacity 3 : disable both null blocks and opacity band

When writing ECW v3, a value of 2 is recommended to benefit from an opacity band as well as
reduced storage requirements and possible performance improvements using the null block
capability. Be aware that null blocks introduce additional calculations in compression, so where the
null area represents a small percentage of the total dataset area (Ratio to data), or the region
complexity is large (a high number of vertices) a value of 0 may compress faster. Testing is
recommended to quantify this difference for your situation. The output quality will be identical
between 0 and 2. ECW users will be unaware whether null blocks are present and behave like any
other ECW file.
When writing ECW v2 and JPEG2000 output, a value of 0 will ensure the output has an opacity band
to make the background transparent.
If no region and no opacity value are specified, Image Compressor will recreate the input opacity
information when detected however a region will override this automatic translation.
The following diagram visualizes each available option where the red polygon is the input region,

No Region and Opacity 1 have opaque backgrounds with no opacity band in the output. Opacity
1 will have reduced storage requirements due to null blocks however will still have a solid
background colour (defined by nullcolor). Only opacity values of 0 or 2 will create an opacity band
and allow underlying data to be shown in this area.
Default value is 2 for ECW v3 output and 0 for ECW v2 and JPEG2000

ERDAS Image Compressor User Guide

31

-datasetregionfile region.txt
A file on disk that represents a closed polygon in image coordinates (pixel space) defining the active
or data area of the image. All area outside of this polygon will be tagged as null and/or transparent
as per the opacity option.
Image Compressor will not clip the output image dimensions to the extents of the input region so it
should not be used to subset larger images.
Dataset coordinates

Output result

563,1642;
1071,1681;
1040,1408;
899,1423;
915,1486;
727,1509;
758,1337;
1181,1345;
1181,1775;
625,1760;
625,1939;
1321,1916;
1353,1165;
664,1181

Note: Each coordinate pair must be specified on a new line with a semi-colon separator.
-shapefile vector.shp, layername, fid
Provide an Esri Shapefile to define the non-Null or active area region defined by a closed polygon
feature. Requirements:

Shapefile projection must match the coordinate system of the input raster
FID identifier must map to a single, simple polygon feature

-wktfile region.txt
Identical to worldregionfile however in OGC WKT format. Coordinates must match the dataset
source projection. Multi-polygon type is currently not supported.
Dataset coordinates

Output result

POLYGON((112.65148514850258721 23.95247524751998469,135.48118811879865575 8.2336633663325216,154.94257425741173506 53.51881188118210986,112.65148514850258721 23.95247524751998469))

32

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Advanced Options
-bitdepth 8 | 16
Force the output cell type to the provided value. If undefined the Image Compressor will maintain
the input celltype so this option is only required in instances where the input type is incorrectly
detected. Int16 is supported in ECW v3 and JPEG2000 output.
Note: Setting bitdepth to 8 (uint8) when the input range is uint16 will not rescale the data. Image
Compressor will merely truncate the data values.
-signed true | false
Used in conjunction with bitdepth 16 to set signed or unsigned int16 output. Signed INT16 is only
supported in JPEG2000 output.
-colorspace greyscale | rgb | multiband
Both ECW and JPEG2000 formats support three colorspaces which are dependent on the band count.
When undefined, Image Compressor will maintain the colorspace of the input data. This option can
be used in conjunction with bandlist to subset multiband input to create RGB or Greyscale output.
-bandlist 0,1,2
Explicitly select which input bands are to be compressed, indexed from 0. Default option is to
compress all bands in original order.
This option allows bands to be reordered such as BGR to RGB (eg. 2,1,0) or to be targeted selectively
when used with the colorspace option. For example a 7 band Multiband input can be mapped to a
RGB 7,4,2 output by specifying colorspace RGB bandlist 7,4,2.
-logfile C:\log.txt
All information printed to the console will also be logged to file, useful for auditing or performance
purposes.
-genstats true | false
When true the Image Compressor will keep count of histogram bins and data statistics such as
mean, mode, standard deviation for writing to the output file. This is calculated during the
compression process and is only available when writing ECW v3 files (with embedded metadata
support). For performance reasons the calculated statistics is based on the input pixels rather than
the output compressed pixels. This tradeoff will be improved in a future release.
Default value true for ECW v3.
-nullvalue min | max
The colour definition to assign to null blocks when opacity is set to 1 or 2. The min/max represents
the smallest or largest pixel value for the specified bitdepth range. For uint8, min will be 0 (black)
and max, 255 (white). Default value min.
-interactive true | false
Compressor will prompt for user input when the following conditions are met,
ERDAS Image Compressor User Guide

33

Insufficient memory cache

Output file already exists

When using batch compression mode via the GUI, interactive defaults to false. Command line
defaults to true.
-srs EPSG:2180
Override the input projection definition and use the specified EPSG code when writing the
georeferencing information in the output file. This parameter should only be used where the Image
Compressor is unable to detect the source projection correctly. For example, the following IMG input
lists a projection code of WGS84 / LOCAL indicating an offset was detected but EPSG lookup failed.

Where input files are known to have an EPSG code however the code was not detected correctly, the
srs value (-srs EPSG:31984) can be passed into the compressor to set this value explicitly and a
relevant warning will be shown in the log.

Note: The Image Compressor will not perform any reprojection of the source to target projection
systems, nor does it modify in anyway the origin or cell size values. The option should not be used
when the compressor is correctly listing the input projection code. For reprojection needs, ERDAS
IMAGINE is recommended.
-xml (C:\data\mosaic.xml)
Define all compression and mosaic parameters within the XML project file and pass into the
compressor. This option is exclusive of all others, including <input>, <output> and other options.
Example usage: ImageCompressor.exe xml c:\data\mosaic.xml

34

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Reporting example
The Image Compressor records a variety of metrics of the input, output, hardware and throughput
for each compression job. When reporting issues or concerns regarding processing speed, always
include the full report as part of the Intergraph Support ticket.
The information is also written to a logfile when used with the logfile option

Figure 6 - Command line processing report

Key reporting elements include

The Data Reader used (GDAL, ERMLib, ECWJP2 SDK)

The size and structure of the input data
o The internal block size (512x512) for tile, (10000x1) for strip

ERDAS Image Compressor User Guide

35

o Compression used, if any (RLE, LZW, JPEG, Packbits)

o Band count, cell type, opacity and projection information
Calculated memory cache settings
Output compression results in terms of throughput, time and actual compression rate
achieved

All relevant information on the input and output results is recorded in this report. Warnings and
Errors will also be reported in this area.

Usage examples
GeoTIFF to ECW v2 using the line encoder
ImageCompressor c:\data\test.tif e:\data\test.ecw method line
ArcInfo Binary Grid (AIG) to lossless JPEG2000 INT16
ImageCompressor c:\dem\w001000.adf e:\data\dem.jp2 targetrate 1 signed true
4 band BGRN UINT16 IMG HFA to RGB ECW v3 with target rate of 20:1 and band stats created
ImageCompressor c:\multi\bgrn.img e:\data\rgb.ecw version 3 bandlist 2,1,0
targetrate 20 -genstats true
3 band ERMapper ALG to ECW v3 with null blocks and opacity
ImageCompressor c:\alg\mosaic.alg e:\data\mosaic.ecw version 3 opacity 2 shapefile
c:\alg\region.shp,region,0
Mosaic XML project to ECW v2 at 20:1 target ratio (defined within the XML file)
ImageCompressor -xml project.xml
Panchromatic IMG to lossy JPEG2000 with opacity defined by WKT region
ImageCompressor c:\jp2\data.img e:\data\output.jp2 wktfile c:\jp2\region.txt opacity 0
Multiband 7 band TIF to RGB ECW with band selection
ImageCompressor multiband.tif rgb.ecw bandlist 7,4,2 colorspace rgb

36

ERDAS Image Compressor User Guide

ERDAS Image Compressor

XML Project File

The Image Compressor uses its own XML Project file to define all compression parameters and is the
only way to define a mosaic input task. This allows for hundreds or thousands of input datasets to be
mosaicked together in the one process by passing in the XML file to the compressor with multiple
input files. The compressor reads the XML and uses the ER Mapper library internally for processing
the mosaic, so performance will be comparable with an ALG equivalent file.
The XML can be generated programmatically or via the Mosaic Wizard User Interface. The XML can
describe only 1 input file so can also be used for batch processing. Mosaic functionality is used
when multiple file elements are present.
Currently the mosaic XML input only supports RGB output. If multiband or greyscale mosaicking is
required, ALG or VRT file input should be used. This limitation will be addressed in a future release.
Note: Due to the dependency on the ER Mapper library, not all input datasets from GDAL are
supported in mosaic mode.

Example mosaic.xml
<?xml version="1.0" ?>
<imagecompressor version="1.0">
<!--compression task: attributes describe the options for the process-->
<compresstask interactive="true|false" logfile="C:/temp/compressor.log" loglevel="Info" method="tile"
tempdir="C:/temp" threads="4" memcache="0.25">
<!--inputs: multiple files determine mosaic mode-->
<inputs>
<file bandlist="0,1,2" path="C:/dir/file_rgb1.ecw" zindex="0"/>
<file bandlist="0,1,2" path="C:/dir/file_rgb2.ecw" zindex="1"/>
<file bandlist="7,4,2" path="C:/dir/file_landsat5.ecw" zindex="2"/>
</inputs>
<!--output options-->
<output path="C:/data/output/compressed.ecw" version="2">
<!--output bandlist-->
<bandlist>
<band description="red" id="0"/>
<band description="green" id="1"/>
<band description="blue" id="2"/>
</bandlist>
<!--region definition from shapefile-->
<region layerid="1" layername="file" name="region1" path="file.shp" type="shape"/>
<!--region definition from WKT-->
<region name="region2" path="file.wkt" type="wkt"/>
<!--metadata: ecwv3 only-->
<metadata>
<item name="classification" value="raster image"/>
<item name="acquistiondate" value="2013-09-12"/>
<item name="acquisitionsensorname" value="Landsat 7"/>
<item name="author" value="John Smith"/>
<item name="copyright" value="Intergraph"/>
<item name="company" value="Intergraph"/>
<item name="email" value="[email protected]"/>
<item name="address" value="2 Abbotsford St, West Leederville WA 6007"/>
<item name="telephone" value="(08) 9388 2900"/>
</metadata>

ERDAS Image Compressor User Guide

37

<!Embedded RPC: ECW v3 only-->

<rpcdata>
<errorbias>12.23</errorbias>
<errorrandom>0.48</errorrandom>
<lineoffset>3522</lineoffset>
<sampleoffset>4406</sampleoffset>
<latitudeoffset>35.2298</latitudeoffset>
<longitudeoffset>-80.8601</longitudeoffset>
<heightoffset>186</heightoffset>
<linescale>3639</linescale>
<samplescale>4421</samplescale>
<latitudescale>0.0780</latitudescale>
<longitudescale>0.1020</longitudescale>
<heightscale>501</heightscale>
<!---the below four coefficient need to have 20 parameters separated by , for each-->
<linenumeratorcoefficients>
-2.082755E-03,-2.790828E-02,...-5.147002E-07
</linenumeratorcoefficients>
<linedenominatorcoefficients>
1.000000E+00,-3.544814E-05,...+2.449699E-07
</linedenominatorcoefficients>
<samplenumeratorcoefficients>
-7.918007E-05,+1.000871E+00,...+6.943991E-08
</samplenumeratorcoefficients>
<sampledenominatorcoefficients>
+1.000000E+00,-2.142645E-04,...+0.000000E+00
</sampledenominatorcoefficients>
</rpcdata>
<!--options-->
<options>
<option name="bitdepth" value="8"/>
<option name="targetrate" value="30"/>
<option name="colorspace" value="rgb|greyscale|multiband"/>
<option name="genstats" value="true"/>
<option name="blocksizex" value="64"/>
<option name="blocksizey" value="64"/>
<option name="huffman" value="1"/>
<option name="opacity" value="0|1|2|3"/>
<option name="nullblocks" value="region2"/>
<option name="nullvalue" value="min"/>
<option name="qualitylayers" value="50"/>
<option name="srs" value="EPSG:4326"/>
</options>
</output>
</compresstask>
</imagecompressor>

38

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Supported input formats

ECWJP2 SDK Input Reader:

GSBG: Golden Software Binary Grid (.grd)

Intergraph ECW (.ecw) version 2 and 3

GS7BG: Golden Software 7 Binary Grid (.grd)

JPEG2000 (.jp2, .jpc, .j2k, .jpf, .j2c, or .jpx) files

COSAR: COSAR Annotated Binary Matrix

(TerraSAR-X)

ERMapper Algorithm: .alg

TSX: TerraSAR-X Product

ERMapper Virtual Raster: .ers

COASP: DRDC COASP SAR Processor Raster

ERMapper Input Reader:

R: R Object Data Store

VRT: Virtual Raster

MAP: OziExplorer .MAP

GTiff: GeoTIFF

PNM: Portable Pixmap Format (netpbm)

NITF: National Imagery Transmission Format

DOQ1: USGS DOQ (Old Style)

RPFTOC: Raster Product Format TOC format

DOQ2: USGS DOQ (New Style)

ECRGTOC: ECRG TOC format

ENVI: ENVI .hdr Labelled

HFA: Erdas Imagine Images (.img)

EHdr: ESRI .hdr Labelled

SAR_CEOS: CEOS SAR Image

GenBin: Generic Binary (.hdr Labelled)

CEOS: CEOS Image

PAux: PCI .aux Labelled

JAXAPALSAR: JAXA PALSAR Product Reader (Level

1.1/1.5)

MFF: Vexcel MFF Raster

MFF2: Vexcel MFF2 (HKV) Raster

GFF: Ground-based SAR Applications Testbed File

Format (.gff)

FujiBAS: Fuji BAS Scanner Image

GSC: GSC Geogrid

ELAS: ELAS

FAST: EOSAT FAST Format

AIG: Arc/Info Binary Grid

BT: VTP .bt (Binary Terrain) 1.3 Format

AAIGrid: Arc/Info ASCII Grid

LAN: Erdas .LAN/.GIS

GRASSASCIIGrid: GRASS ASCII Grid

CPG: Convair PolGASP

SDTS: SDTS Raster

IDA: Image Data and Analysis

DTED: DTED Elevation Raster

NDF: NLAPS Data Format

PNG: Portable Network Graphics

EIR: Erdas Imagine Raw

JPEG: JPEG JFIF

DIPEx: DIPEx

MEM: In Memory Raster

LCP: FARSITE v.4 Landscape File (.lcp)

JDEM: Japanese DEM (.mem)

GTX: NOAA Vertical Datum .GTX

GIF: Graphics Interchange Format (.gif)

LOSLAS: NADCON .los/.las Datum Grid Shift

BIGGIF: Graphics Interchange Format (.gif)

NTv2: NTv2 Datum Grid Shift

ESAT: Envisat Image Format

CTable2: CTable2 Datum Grid Shift

BSB: Maptech BSB Nautical Charts

ACE2: ACE2

XPM: X11 PixMap Format

SNODAS: Snow Data Assimilation System

BMP: MS Windows Device Independent Bitmap

ARG: Azavea Raster Grid format

DIMAP: SPOT DIMAP

RIK: Swedish Grid RIK (.rik)

AirSAR: AirSAR Polarimetric Image

USGSDEM: USGS Optional ASCII DEM (and CDED)

RS2: RadarSat 2 XML Product

GXF: GeoSoft Grid Exchange Format

PCIDSK: PCIDSK Database File

PCRaster: PCRaster Raster File

NWT_GRD: Northwood Numeric Grid Format

.grd/.tab

ILWIS: ILWIS Raster Map

SGI: SGI Image File Format 1.0

NWT_GRC: Northwood Classified Grid Format

.grc/.tab

SRTMHGT: SRTMHGT File Format

ADRG: ARC Digitized Raster Graphics

Leveller: Leveller heightfield

SRP: Standard Raster Product (ASRP/USRP)

Terragen: Terragen heightfield

BLX: Magellan topo (.blx)

ISIS3: USGS Astrogeology ISIS cube (Version 3)

SAGA: SAGA GIS Binary Grid (.sdat)

ISIS2: USGS Astrogeology ISIS cube (Version 2)

KMLSUPEROVERLAY: Kml Super Overlay

PDS: NASA Planetary Data System

XYZ: ASCII Gridded XYZ

TIL: EarthWatch .TIL

HF2: HF2/HFZ heightfield raster

GDAL Input Reader:

ERDAS Image Compressor User Guide

39

ERS: ERMapper .ers Labelled

PDF: Geospatial PDF

L1B: NOAA Polar Orbiter Level 1b Data Set

OZI: OziExplorer Image File

FIT: FIT Image

CTG: USGS LULC Composite Theme Grid

GRIB: GRIdded Binary (.grb)

E00GRID: Arc/Info Export E00 GRID

RMF: Raster Matrix Format

ZMap: ZMap Plus Grid

MSGN: EUMETSAT Archive native (.nat)

NGSGEOID: NOAA NGS Geoid Height Grids

RST: Idrisi Raster A.1

IRIS: IRIS data (.PPI, .CAPPi etc)

INGR: Intergraph Raster

GSAG: Golden Software ASCII Grid (.grd)

Note: When mosaicking input data, not all GDAL supported formats listed above are available due to nonthread safe operations. The list above is supported for batch processing or single input only.

40

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Null block analysis

A major improvement with ECW v3 is the introduction of null blocks that can offer further file
storage savings and compression performance compared with ECW v2 or JPEG2000 without
sacrificing image quality. The key criteria as to whether null blocks should be enabled is the
relationship of the input data extent to the amount of null or no-data areas and the size of the input
image. Generally speaking the higher amount of null area defined by the input region with increasing
image input size, the greater the gains that enabling null blocks will provide.
The compressor output lists two important region metrics to help identify suitability,
1. Number of vertices which indicates the complexity of the input region. The greater the
vertices, the more expensive spatial intersection tests will be. The compressor has
implemented a variety of optimizations but where possible a simplified region will
ensure the fastest throughput. The compressor has been tested with polygons with tens
of thousands of vertices however its expected most use-cases will on average only
require vertices in the hundreds if not less.
2. Ratio of null area to data which can be visualized in the following diagrams where the
red area is the region passed into the compressor and the white is the remaining area to
be tagged as null blocks

An important observation in these examples are the first and third use-cases. Both have 4 vertices
however clearly the percentage ratio to data is a lot higher in the third example at 95%. Therefore
null blocks will provide the greatest benefit to the third image both in terms of additional file storage
savings and compression speed. The first example will still benefit and is still a good candidate for
null blocks however will not see as significant gains. Null blocks in this way will always provide
varying levels of optimizations depending on input as highlighted in the following examples.

ERDAS Image Compressor User Guide

41

Comparison
Sydney Landsat scene

Input image
Dimensions:

15,221 x 14,661 px
(0.223 gigapixel)

Structure:

3 Band, RGB UINT8

Opacity band:

false

Projection:

EPSG:32656

Shapefile region (yellow)

Null Vertices:

Ratio to data:

30.234%

Null blocks enabled

Hardware
Platform:

Hardware
Windows 7 / Server 2008 R2

Platform:

Windows 7 / Server 2008 R2

CPU Model:
Intel(R) Core(TM) i7 CPU
Q 740 @ 1.73GHz

CPU Model:
Intel(R) Core(TM) i7 CPU
Q 740 @ 1.73GHz

CPU Cores:

CPU Cores:

Memory:

8
8,128.00 MB

Memory cache

42

Null blocks disabled

Memory:

8
8,128.00 MB

Memory cache

System:

512.00 MB

System:

512.00 MB

Read:

1,911.85 MB

Read:

1,911.85 MB

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Write:

120.15 MB

Write:

120.15 MB

Threads:

Threads:

Precincts:
Total Blocks:

73376
18344

Precincts:
Total Blocks:

73376
18344

Data Blocks:

13251

Data Blocks:

18344

Null Blocks:

5093

Null Blocks:

Duration:

0 hours 0 mins 55 seconds

Duration:

0 hours 1 mins 20 seconds

Target Ratio:

15:1

Target Ratio:

15:1

Actual Ratio:

31.3:1

Actual Ratio:

30.7:1

Throughput:

11.5 MB / sec

Output Data
File Name:

Throughput:

8.0 MB / sec

Output Data
f:\landsat-null.ecw

File Name:

f:\landsat-no-null.ecw

File Type:

ECW v3

File Type:

ECW v3

Data Writer:

ECW JPEG2000 SDK v5.1

Data Writer:

ECW JPEG2000 SDK v5.1

Dimensions:
Structure:
File Size:

15,221 x 14,661 px
4 Band, RGB UINT8
20.39 MB

Dimensions:
Structure:
File Size:

15,221 x 14,661 px
4 Band, RGB UINT8
20.79 MB

1. File savings: 400kb ( ~ 2% smaller )

2. Time savings: 25 seconds ( ~ 30% faster )
The small file size difference is expected in this example despite the 30% ratio to data because the
input image is only small at 0.2 gigapixels. This means there are fewer resolution levels within the
ECW file which in turn means there are fewer null blocks in the output. Irrespective of the small file
size improvement, enabling null blocks increases compression speed by 30% which can be significant
depending on use-case, for example compressing thousands of images in batch.
Note: It is a coincidence that the compression speed improvement of 30% matches the input region
ratio to data of 30%. The exact performance throughput gains will vary.

ERDAS Image Compressor User Guide

43

Corridor mapping example

Input image
Dimensions:

372,535 x 477,806 px
(177.999 gigapixel)

Structure:

4 Band, RGB UINT8

Opacity band:

true

Projection:

EPSG:28350

Shapefile region (yellow)

Null Vertices:

33

Ratio to data:

89.731%

Null blocks enabled

Hardware
Platform:

Hardware
Windows 7 / Server 2008 R2

Platform:

Windows 7 / Server 2008 R2

CPU Model:
Intel(R) Xeon(R) CPU
E5410 @ 2.33GHz

CPU Model:
Intel(R) Xeon(R) CPU
E5410 @ 2.33GHz

CPU Cores:

CPU Cores:

Memory:

44

Null blocks disabled

8
16,380.00 MB

Memory:

8
16,380.00 MB

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Memory cache

Memory cache

System:

512.00 MB

System:

512.00 MB

Read:

2,319.47 MB

Read:

2,319.47 MB

Write:

1,775.53 MB

Write:

1,775.53 MB

Threads:

Threads:

Precincts:

57967752

Precincts:

57967752

Total Blocks:

14491938

Total Blocks:

Data Blocks:

1506776

Data Blocks:

14491938

Null Blocks:

12985162

Null Blocks:

Duration:

0 hours 53 mins 45 seconds

Duration:

14491938

3 hours 41 mins 19 seconds

Target Ratio:

15:1

Target Ratio:

15:1

Actual Ratio:

209.4:1

Actual Ratio:

169.3:1

Throughput:

210.6 MB / sec

Output Data
File Name:

Throughput:

51.2 MB / sec

Output Data
g:\corridor-null.ecw

File Name:

g:\corridor-no-null.ecw

File Type:

ECW v3

File Type:

ECW v3

Data Writer:

ECW JPEG2000 SDK v5.1

Data Writer:

ECW JPEG2000 SDK v5.1

Dimensions:

372,535 x 477,806 px

Dimensions:

372,535 x 477,806 px

Structure:
File Size:

4 Band, RGB UINT8

3,242.55 MB

Structure:
File Size:

4 Band, RGB UINT8

4,011.66 MB

1. File savings: 769mb ( ~ 19% smaller )

2. Time savings: 2 hours 48 minutes ( ~ 410% faster )
This example shows the strengths of enabling null blocks. It has a relatively simple input region and a
high level of null data of 89%. Unlike the previous example, we can now observe significant gains to
both compression speed and file savings with no degradation to image quality.

ERDAS Image Compressor User Guide

45

FAQ
Can I purchase the Image Compressor application separately?
Not at this time. Please contact your local Intergraph representative to express interest in an
individual licensing model.
Should I use Image Compressor or IMAGINE Professional?
You are entitled to use both. Depending on your workflow you may find Image Compressor too
limiting or conversely perfectly suitable.
Are files created by Image Compressor any different to those from IMAGINE?
No. There may be slight variations to the metadata stored in the file headers, but otherwise an ECW
or JPEG2000 compressed image will be identical regardless what application is used. Both
applications use the same underlying ECWJP2 SDK version that performs the compression.
What relationship does ERDAS Image Compressor have with the legacy ER Mapper Image
Compressor application?
None. The ERDAS Image Compressor is a completely redesigned and rebuilt compression application.
Despite the similar product naming, the capabilities are not equivalent and is not a direct upgrade
from the legacy product.
What advantages does Image Compressor provide over using the ECWJP2 SDK within GDAL?
Although GDAL has excellent support for the ECWJP2 SDK, a key feature it lacks is the multi-threaded
tile encoder and region management for null block and opacity generation. There are also other
performance implications that limit throughput if GDAL is used for both reading and writing. Image
Compressor will remain quicker at compressing equivalent input even if the GDAL reader is used.
What mosaic file format gives the best performance?
Image Compressor supports both ER Mapper ALG and GDAL VRT file formats. Both support a wide
range of dynamic capabilities including the ability to mosaic separate files together to present them
as a single virtual file or mosaic. The Image Compressor mosaicking process uses the ER Mapper ALG
engine internally for this capability and was chosen for performance reasons and because its threadsafe. VRT is supported however is known to perform slower than an equivalent ALG especially for
large input sizes. VRT files can also only be compressed using the line encoder.
Why cant I define a multi-polygon for opacity/null definition?
This limitation will be removed from a future release of Image Compressor.
I use complex ER Mapper Algorithms that link to Virtual ERS rasters, kernels and so on, will Image
Compressor compress these files?
Yes. Image Compressor uses the full ER Mapper Library thats included in ER Mapper Professional
2014, however in a 64-bit environment. Image Compressor can be used to compress large datasets

46

ERDAS Image Compressor User Guide

ERDAS Image Compressor

where memory requirements exceed 32-bit limits but still benefit from the powerful visualization
tools that ER Mapper provides.
Is Image Compressor faster than Product X?
Image Compressor compresses faster than the legacy ER Mapper 7.2 (64-bit), ER Mapper Mosaic
Balance Compressor (MBC), Safe FME and Global Mapper products all of which are used heavily
throughout the industry for compressing to ECW and JPEG2000. These products use the legacy
ECWJP2 SDK v3.x which was released over 7 years ago. In comparison ERDAS IMAGINE 2014 and
Image Compressor 2014 are built on the latest v5.1 ECWJP2 SDK and benefits from years of further
optimizations, bug fixes, ECW v3 and other improvements. Image Compressor is also maintained by
the same development team as the ECWJP2 SDK ensuring the product is the reference, best-of-breed
implementation of the underlying compression engine.
Why does the compressor prevent some input formats from being compressed in tile mode?
Using the tile compression method forces multi-threaded accesses on the input data. For a variety of
reasons many input formats are not thread-safe. Where we are aware of formats that are not
thread-safe we default to the line compression method and will return an error if tile is used.
The GDAL VRT format is one such example of an input format that compresses without issue with a
single thread however causes significant problems when called from multiple threads. This problem
is a general GDAL Data Reader problem and is not specific to Image Compressor, see GDAL
documentation. As improvements are made to the Data Readers like GDAL, Image Compressor will
re-enable these formats.
Can the compressor be used on a headless Linux machine?
Yes, however a known limitation is that we provide only a GUI based installation (BIN) package. This
will be remedied in a future release. Files can be transferred easily to the headless machine after
installation and the command line compressor will function as expected.
What window managers are supported on Linux?
The Image Compressor GUI is built using the Qt cross-platform library and has been tested across
Gnome, KDE and Unity platforms.
Why are some Linux platforms listed as viable?
Due to time constraints, platforms which are viable are deemed to be usable and in some cases
tested however not to the same degree as fully supported platforms and may require additional
libraries to function.
Why is georeferencing sometimes lost in the output files?
The Image Compressor relies on the Data Readers to interpret georeferencing information and
determine an EPSG code. For a variety of reasons this process is not always successful in which case
the Image Compressor will create output files in a WGS84/LOCAL. This definition can be altered to
the correct EPSG code by updating the file header after the fact; a recompression is not required. Or
alternatively the srs command line parameter can be defined prior to compression to force the
output definition.
If no georeferencing is detected at all and the input data is raw, RAW/RAW projection/datum
pairing will be written to the output.
ERDAS Image Compressor User Guide

47

To confirm, check the Compressor log and look for the Projection value set within the Input Data
metadata area.
What settings should I use to get best throughput?
This question is dependent on too many variables and can only be answered after benchmarks and
comparisons conducted on customer hardware with customer input data. Contact Intergraph
Support for assistance.
Are pyramids or overviews required for compression?
No. The Image Compressor only reads the input data at native resolution. If files such as RRD, OVR or
internally defined overviews are present they will be ignored. Generating these files is a redundant
processing step for compression workflows; they do not speed up the compression of ECW or
JPEG2000 files.
Why is the output actual compression ratio significantly different to the target when using ECW v3
null blocks?
Depending on the region supplied and the amount of area defined as null, it is not uncommon to
specify 20:1 target and obtain an actual compression rate of 50:1 or more. When null blocks are
enabled the amount of data written to the output file can be substantially reduced which is reflected
in the actual compression ratio since this divides the uncompressed input size by the output size. It
makes no allowances between null or data areas.
This behaviour can be quantified by customers using the APOLLO Essentials image quality utility or
by your own visual quality assessment by overlaying a sample image with and without null blocks
enabled. The image quality of the data areas will remain identical to the same image compressed
to 20:1 without null blocks even though its actual target rate may be closer to the target rate.
When should null blocks be enabled?
Null blocks were designed to shrink the output file size and speed up compression speed, but these
benefits are directly related to the characteristics of the input. To help identify when null blocks
should be enabled, the compression output will report the Ratio to data as well as the number of
vertices that make up the selected region. Where the ratio to data is low enabling null blocks will not
yield much change to the output file size and in many cases will increase compression time due to
the additional spatial intersection checking. Similarly if the number of vertices is in the thousands,
performing spatial intersection tests against very complex polygons could adversely impact the
compression time negating the compression speed improvements.
The degree to which these trade-offs affect throughput is heavily dependent on the input data but
also the hardware that the compression is performed on. The only reliable way to answer this
question is to perform two identical compression tests, one with opacity 0 (or with no region
specified) and one with opacity 2. This will allow direct comparison of the benefits that null blocks
provide. See the Null Block Analysis chapter for more information and comparisons.
Why does the compressor need to calculate min/max values?
In order to compress signed or unsigned 16-bit output the compressor needs to know the data range
to ensure high quality output when the actual data range is a smaller sample set, for example 11-bit
stored in a 16-bit range. This step is not required for 8-bit output.

48

ERDAS Image Compressor User Guide

ERDAS Image Compressor

Depending on the size and format type this range calculation may take a few minutes before the
actual compression begins. This calculation uses an approximate algorithm by taking an overview
image of approximately 2,500 pixel samples to determine the range. If the range reported appears
wildly inaccurate it is recommend to calculate full statistics prior to compression and the Image
Compressor will read existing statistics when detected.
Image Compressor supports reading statistics stored in *.aux.xml, ECW v3, ERS and ALG files.
When compressing large inputs why is there a delay between reaching 100% complete and the
compressor exiting?
For large inputs 100gigapixels or more, there can be a delay of a few minutes between when the
compressor reaches 100% complete and the output statistics are printed and the compressor exits.
This is expected as the compressor needs to reassemble the temporary files into the final output and
delete these files from disk before the compression is deemed complete. You will observe a high
level of disk activity within this period.

ERDAS Image Compressor User Guide

49

Support
The Image Compressor product support is available to all customers with active Software
Maintenance on any of the dependant Intergraph Geospatial 2014 products. See the Intergraph SG&I
Support webpage for more information on how to raise support requests.

When reporting problems, please include the example Image Compressor output which will include
all relevant information. Input data may be required.

50

ERDAS Image Compressor User Guide