imreadmulti() in Python Opencv

OpenCV (Open Source Computer Vision) library primarily focuses on image and video processing. To load a single image through OpenCV, imread() function is used. However, to load multiple images in a single object, imreadmulti() function is used. Images within a file are considered as a page. Although this method can handle single-page files, it frequently works with multi-page or multi-frame image files.

Some of the file types supported by imreadmulti() are BMP (Windows bitmap), DIB (Device-independent bitmap), JPEG or JPG (Joint Photographic Experts Group), JP2 (JPEG 2000), PNG (Portable Network Graphics), PBM (Portable bitmap), PGM (Portable gray map), PPM (Portable Pixmap), TIFF (Tagged Image File Format) or TIF (Tagged Image Format), PFM (Portable float map), WEBP (Web Picture). Some file types such as EXR (OpenEXR) require codecs.

The result is loaded in a vector of Mat objects. In OpenCV, the cv::Mat class is used to represent a multi-dimensional array. It can be used to store image data in many different formats, including 32-bit floating-point numbers (CV 32F) 2-channel floating-point array, 8-bit unsigned integers (CV 8U) 8-bit single-channel array, and 16-bit signed integers (CV 16S).

Syntax: cv2.imreadmulti()

Syntax: 

cv2.imreadmulti( filename, mats, flags, start(optional), count(optional))

Parameters:

1. filename : The path of the .tiff files
2. mats : The list/vector where loaded images will be stored
3. flags : Default is cv2.IMREAD_ANYCOLOR . Other possible flags such as cv2.IMREAD_GRAYSCALE etc can be found here.
4. start (optional) : To set the start index of the image frame to be loaded from the file. Default is 0.
5. count (optional): The count of images to be added in mats.

Returns: 

• ret : Boolean value for successful execution.
• mats : The loaded images list/vector.

Example 1:

• In the mentioned code, the Python OpenCV Library is imported. 
• The path variable stores the image file path of the .tiff image file from the local machine.
• The images list variable is declared to store the loaded images.
• The Boolean value and Mat object returned by cv2.imreadmulti() function are stored in ret and images variable.
• The loaded images are displayed on the screen by cv2.imshow() function.
• The windows are then destroyed after waitKey is pressed.

# Importing Opencv
import cv2

# Path to the tiff file
path = "gfg.tiff"

# List to store the loaded image
images = []

ret, images = cv2.imreadmulti(mats=images,
                              filename=path,
                              start=0,
                              count=2,
                              flags=cv2.IMREAD_ANYCOLOR)

# Show the images
if len(images) > 1:
    for i in range(len(images)):

        # Dynamic name
        name = 'Image'+str(i)

        # Shape of Image
        print(name, 'Shape :', images[i].shape)

        # Displaying the image
        cv2.imshow(name, images[i])

        # Save the images
        cv2.imwrite(name+'.jpg', images[i])


# Waiting for user to press any key to stop displaying
cv2.waitKey()

# Destroying all windows
cv2.destroyAllWindows()

Output:

Image0 Shape : (512, 512, 3)
Image1 Shape : (515, 570, 3)

Image0:

Image1:

Example 2:

The imreadmulti() function offers optional parameters that let users specify the starting frame and the number of loaded images.

The second and third images from a 3-page tiff file are shown in this example. You can set the count option to 1 to only show one image.

# Importing Opencv
import cv2

# Path to the tiff file
path = "gfg_2.tiff"

# List to store the loaded image
images = []

ret, images = cv2.imreadmulti(mats=images,filename=path,start=1, count=2, 
                              flags=cv2.IMREAD_COLOR)

# Show the images
if len(images) > 1:
    for i in range(len(images)):
        
        # Dynamic name
        name = 'Image'+str(i+1)
        
        # Shape of Image
        print(name,'Shape :',images[i].shape)
        
        # Displaying the image
        cv2.imshow(name, images[i]) 
        
        # Save the images
        cv2.imwrite(name+'.jpg', images[i])


# Waiting for user to press any key
cv2.waitKey()

# Destroying all windows
cv2.destroyAllWindows()

Output:

Image1 Shape : (515, 570, 3)
Image2 Shape : (176, 176, 3)

Output Images:

Image1: 

Image2:

Example 3:

If the image has only one page. Then the output will be also one image. To avoid issues while using this example code, the parameters' order must align with what is given.

Note: When loading single-page images, the imread() and imreadmulti() functions differ as the image is directly stored as a Mat object in the first while the latter stores it as a list of Mat objects that must be accessed using an index in square brackets (here- images[0], images2[0]).

Alternatively, the flags can also be specified as numerical value (here- cv::IMREAD_GRAYSCALE = 0, cv::IMREAD_COLOR = 1).

# Importing Opencv
import cv2

# Path to the file
path = "Ganesh.jpg"

images = []
ret, images  = cv2.imreadmulti(mats=images,
                               filename = path,
                               flags=0)

print('Number of images :',len(images))

# With different flags
ret2,images2 = cv2.imreadmulti(filename = path,
                               flags=cv2.IMREAD_COLOR)



# Displaying image in grayscale
cv2.imshow("Gray Image", images[0]) 
 # Save the images
cv2.imwrite('Gray Image.jpg', images[0])

# Displaying image in color
cv2.imshow("Color Image", images2[0]) 
 # Save the images
cv2.imwrite('Color Image.jpg', images2[0])

# Waiting for user to press any key
cv2.waitKey()

# Destroying all windows
cv2.destroyAllWindows()

Output:

Number of images : 1

Output Images:

Gray Image:

Color Image: