Hi @sturkmen72, I have added the handlers as discussed. In addition to this I noticed that color resize was not taking place in case of imread, I fixed that as well.

i believe your addition is not correct ( or unnecessary)

also see https://github.com/opencv/opencv/blob/master/modules/imgcodecs/src/loadsave.cpp#L494 should be added in imdecode

Hi @sturkmen72 thanks for the response, so the addition to imread is not needed? Or even imdecode is not necessary? As mentioned in last comment, do I add that to imdecode?

imread manually resizes other formats than jpeg see https://github.com/opencv/opencv/blob/master/modules/imgcodecs/src/loadsave.cpp#L494-L497

@sturkmen72, thanks for that, shall I add that to imdecode as well as you showed earlier?

@ganesh-k13 Please fix warnings from Win32 builder (add static_cast<size_t>())

@alalek let me do some tests before merge

imread manually resizes other formats than jpeg see https://github.com/opencv/opencv/blob/master/modules/imgcodecs/src/loadsave.cpp#L494-L497

@sturkmen72, do we make this in a different PR, or shall I add it to imdecode now?

Hey @sturkmen72 / @alalek, I'm getting cores if we add that if conditions and try other formats, if that if is not there then resize does not happen but no crash. It involves some work, can I raise a different PR for that, might involve more work. Please advise.

you can add

    if( decoder->setScale( scale_denom ) > 1 ) // if decoder is JpegDecoder then decoder->setScale always returns 1
    {
        resize( mat, mat, Size( size.width / scale_denom, size.height / scale_denom ), 0, 0, INTER_LINEAR_EXACT);
    }
   decoder.release();

end of imdecode_
( delete decoder.release(); which located a few lines upper )

Ohh thanks, I didn't see it was getting freed. I'll add that and test.

@sturkmen72 thanks for the fix, it works. @alalek I have added a case for the same by resizing a PNG to hit that code path.

@sturkmen72 / @alalek, any updates on this one? any changes needed?

Hello @ganesh-k13 The patch looks good to me. I can propose some test code enhancement: OpenCV TS module like Google Test supports parameterized tests (https://github.com/google/googletest/blob/master/googletest/docs/advanced.md#value-parameterized-tests). It means that you can get rid of duplicated blocks of code and run the same test body for JPEG and other image formats. You can look on example in test_grfmt.cpp.

Thanks for the link @asmorkalov, I'll make those changes

Parametrized tests look like overkill here. Duplicated code and checks may be moved into common function, like this (important note, that return value should be 'void').

Hi @alalek , @sturkmen72 , @asmorkalov, we can see that imread and imdecode have duplicate code in both test and src. I would suggest we make it a different PR as we have digressed from the initial issue. I can raise a new issue and PR for that.

Hi @alalek , @asmorkalov , I went ahead with refactor of tests anyways. I went with Value-Parameterized Tests as I wanted to try something new and it seems very well suited for our case. Please merge this PR if changes are satisfactory.

@ganesh-k13 Well done! Please squash your commits to have clean history and we can merge the patch. @alalek What do you think?