Scribd
Upload
10 views

Basic Image Processing

Uploaded by

Nguyễn Mạnh Quỳnh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
10 views

Basic Image Processing

Uploaded by

Nguyễn Mạnh Quỳnh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 32

Lecturer: Bui Ha Duc, PhD

Email: [email protected]
▪ Image Representation
▪ Types of image
▪ Image Features

▪ Introduction to OpenCV
▪ Load Image
▪ Access pixel value
▪ Set pixel value
▪ Digital image is presented by pixel matrix
▪ Image processing operation in a computer may be observed as a
matrix operation
O x

Image(3,4) = ?
▪ Represented by 3 matrices
▪ Colors are seen as variable
combination of primary colors Red
(R), Green (G), and blue (B)
▪ Each element are integer number
range from 0 to 255
▪ Intensity of the pixel with respect to the
color
▪ In RGB system, it’s possible to
(R,G,B)=(146,185,216)
create 2563=16777216 different
colors
Color Image
RGB components
R/G/B histogram
Color image processing is generally challenging
▪ Three separate, “independent” channels
▪ Time consuming
▪ Require nonlinear approaches
▪ Represented by 1 matrix
▪ Each element of the matrix is the
intensity of the corresponding pixel
▪ Range from 0 (black) to 255
(white)
▪ Covert from RGB image:
𝑮 = 𝟎. 𝟐𝟗𝟗𝑹 + 𝟎. 𝟓𝟖𝟕𝑮 + 𝟎. 𝟏𝟏𝟒𝑩

intensity=177
▪ Also called Boolean images
▪ Represented by a matrix
▪ All elements are 0 and 1
▪ 0 is black
▪ 1 is white

▪ Result of thresholding operations


▪ Important in segmentation
Basic features Advanced features
▪ Histogram ▪ Regions
▪ Centroid, size, shape
▪ Color
▪ Lines
▪ Edge
▪ Equations
▪ Corner ▪ Start, end points,
▪ orientation

▪ Key points
▪ Location
Edge
▪ Direction
▪ scale Key points
▪ OpenCV: Open Source Computer Vision & Machine
Learning software library
▪ Created in 1999 by Intel
▪ Supported from 2008 by Willow Garage
▪ Willow Garage also supported the Robotic Operating system (ROS)
and Point Cloud Library (PCL)
▪ OpenCV is a cross-plaform
▪ Available in Windows, Linux, Android, MacOS …

▪ OpenCV support a wide range of programming languages:


▪ OpenCV is written in C/C++
▪ Stable source code: opencv.org/release
▪ Developments
▪ Source code github.com/opencv
▪ Online documentation docs.opencv.org

▪ Current stable version is 4.4.0


▪ Written for C++ 11
▪ Better video object detection and tracking
▪ More support for Deep learning
Include headers of
openCV modules

All OpenCV classes and


functions are in the cv
namespace

Without using name space cv Multi-channel array


class for image data

BGR instead of RGB


▪ Mat is a dense multi-channel array class used to store image data
▪ Mat class is the epic center of the OpenCV library
▪ Majority of functions in OpenCV are
▪ member of Mat class,
▪ take Mat as an argument
▪ or return Mat as a return value

▪ Mat contains 2 data part:


▪ Header: containing information such as the size of the matrix, the method used
for storing, at which address is the matrix stored
▪ Pointer: containing the pixel values
▪ Create a Mat constructor
Mat img(int rows, int cols, int type)
Example
img is a 2-D array with 3 rows, 10 columns,
Mat img(3, 10, CV_32FC3) 3 channels, data type is Float 32 bit
▪ Mat data type:
CV_<bit_depth><data_type>C<nb_channels>
▪ Examples

▪ Mat M1 (2, 3, CV_8UC1)


(2x3) single channel array with 8-bit unsigned char data

▪ Mat A (10, 10, CV_16SC3)

(10x10) 3-channel array with 16-bit signed short integer data


▪ Point and Point3 are classes used to store Cartesian coordinates of
2D/3D points
▪ Class for short numerical vector specified by its Cartesian coordinates
Most common Mat Types:
▪ CV_8UC1 : 8-bit, 1-channel for gray scale images
▪ CV_32FC1 : for 32-bit 1 channel gray images
▪ CV_8UC3 : 8-bit 3 channel for color images
▪ CV_32FC3 : for 32-bit 1 channel gray images

The default setting with imread function is CV_8UC3


Access individual pixels value
▪ Use Mat::at<type>(r,c) method Notes:
▪ E.g. For CV_8UC1 Mat: uchar a = img.at<uchar>(r,c) r: row →y
For CV_8UC3 Mat: Vec3b a = img.at<Vec3b>(r,c) C: column → x

▪ Use Mat::at<type>(Point2i(x,y)) method

▪ Use Mat::ptr<type>(int r) to get pointer of the beginning of


a row → ptr[c] to get pixel value at row r, column c
Access individual pixels value
▪ Use Mat::ptr<type>(int r) to get pointer of the beginning of
a row → ptr[c] to get pixel value at row r, column c
e.g.
// Iterate over all pixels of the image
for(int r = 0; r < img.rows; r++) {
// Obtain a pointer to the beginning of row r
Vec3b* ptr = img.ptr<Vec3b>(r);

for(int c = 0; c < img.cols; c++) {


// Invert the blue and red values of the pixel
ptr[c] = Vec3b(ptr[c][2], ptr[c][1], ptr[c][0]);
}
}
Access block pixels value
Set value for a pixel
▪ Use Mat::at<type>(r,c) = value
▪ E.g. For CV_8UC1 Mat: img.at<uchar>(r,c) = uchar a
For CV_8UC3 Mat: img.at<Vec3b>(r,c) = Vec3b b
for (int x = 0; x<img.rows; x++) {
for (int y = 0; y<img.cols; y++) {
// Accesssing values of each pixel
if (img.at<uchar>(x, y) >= threshhold) {
output.at<uchar>(x, y) = 254; // Setting the pixel values to 255 if it's above the threshold
}
else if (img.at<uchar>(x, y) < threshhold) {
output.at<uchar>(x, y) = 0; // Setting the pixel values to 255 if it's below the threshold
}
}
}
Set value for a block of pixels
▪ Use cv::addWeighted() → page 104 “Learning OpenCV 3”
Set value for a block of pixels
Mat img1 = imread(argv[1],1);
Mat img2 = imread(argv[2],1);

Mat roi1(img1,Rect(50,50,100,100));
Mat roi2(img2,Rect(0,0,100,100));

addWeighted(roi1, 0.0, roi2, 1.0, 0.0, roi1);

You might also like