about .mat file

[chensming]

Hi, I am interested in your nice work. But I don't know how to convert png to .mat (For example, sometimes I want to see what the original image looks like). I would appreciate if you could answer this question.

[nagejacob]

Does the 'original image' means the raw camera output? If so, the png can not revert to 'original image' due to lack of ISP model and parameters. However, the original image of SIDD dataset could be downloaded from their website (http://130.63.97.225/share/sidd_dng/) as DNG files.

[chensming]

Does the 'original image' means the raw camera output? If so, the png can not revert to 'original image' due to lack of ISP model and parameters. However, the original image of SIDD dataset could be downloaded from their website (http://130.63.97.225/share/sidd_dng/) as DNG files.

Oh, I just want to know how to prepare the dataset like these structure:

[nagejacob]

Please download SIDD Medium Srgb dataset from https://www.eecs.yorku.ca/~kamel/sidd/dataset.php, and download SIDD Validation and Benchmark '.mat' files from https://www.eecs.yorku.ca/~kamel/sidd/benchmark.php.

[chensming]

Thanks! You are so kind! And I would like to visualize the '.mat' files(eg. ValidationNoisyBlocksSrgb.mat), could you give me some suggestion? (I 'm sorry that I am new to this field) ^_^

[nagejacob]

Try this:

import imageio
import numpy as np
import os
import scipy.io as sio

if __name__ == '__main__':
    SIDD_path = '/Users/nagejacob/Documents/datasets/SIDD'
    mat = sio.loadmat(os.path.join(SIDD_path, 'SIDD_Validation/ValidationNoisyBlocksSrgb.mat'))
    noisy_block = mat['ValidationNoisyBlocksSrgb']
    n = noisy_block.shape[0]
    k = noisy_block.shape[1]
    for i in range(n):
        for j in range(k):
            img = noisy_block[i, j]
            print('extracting %d_%d.png' % (i, j))
            imageio.imwrite('%d_%d.png' % (i, j), img)            

And I advice you to search on google first to solve these common problems.

[chensming]

Try this:

import imageio
import numpy as np
import os
import scipy.io as sio

if __name__ == '__main__':
    SIDD_path = '/Users/nagejacob/Documents/datasets/SIDD'
    mat = sio.loadmat(os.path.join(SIDD_path, 'SIDD_Validation/ValidationNoisyBlocksSrgb.mat'))
    noisy_block = mat['ValidationNoisyBlocksSrgb']
    n = noisy_block.shape[0]
    k = noisy_block.shape[1]
    for i in range(n):
        for j in range(k):
            img = noisy_block[i, j]
            print('extracting %d_%d.png' % (i, j))
            imageio.imwrite('%d_%d.png' % (i, j), img)            

And I advice you to search on google first to solve these common problems.

Oh， thanks a lot!