WSQ decoding

[Mlaras]

Hi !

I was testing your project since I have to decode a wsq file into a png image, but I'm having some errors.

At first I tried with a wsq file i'm currently working with (which I'll attach) and I got the following output.

File "D:\BachelorThesis\decoder.py", line 321, in
main()
File "D:\BachelorThesis\decoder.py", line 317, in main
WSQFileReader().decode(WSQcompressed_file,showDetails).tofile(WSQuncompressed_file)
subbands[i] = np.asarray((self.blocks[0] + self.blocks[1] + self.blocks[2])[prev: prev + size]).reshape((M,N))
ValueError: cannot reshape array of size 8797 into shape (98,90)

Then I tried with an image from the NIST sample dataset where It worked fine. I got the png image successfully.

Then the error also showed up when I used an image from the sample images from the NIST dataset. Specifically with:
D:\reference_images_v2.0_raw\decode\225\sample_01.wsq (Which I'll also attach)

"File "D:\WSQ\BachelorThesis\decoder.py", line 127, in decode
subbands[i] = np.asarray((self.blocks[0] + self.blocks[1] + self.blocks[2])[prev: prev + size]).reshape((M,N))
ValueError: cannot reshape array of size 149900 into shape (375,400)"

Do you mind checking this out?. I suspect it might be a dependency problem. I installed all of the requeriments you specified on a virtual environment.

sample_wsq_files.zip

PS: Awesome work.

[Alessio-jpg]

Hello, sorry for the late response but I'm quite busy with exams at the moment.

I've investigated the issue and there seems to be a size mismatch between subbands during the decoding process.

I have attached a quick solution to the problem. Although the fix does not work in the expected way (the decoded image might have some artifacts), it does provide a way to continue the decoding process without halting the program.
At the moment this is the best I can do, but I will definitely look into it at a later time.

Sorry for the inconvenience

quickfix.zip