Question

If we compare image procesing of the losslessly compressed images with the image processing of the lossy compressed images, does the latter provide the results comparable to the former one.

I am asking this question because the images prodiced by lossless compression are ok for human eye but they vary at minute details which may effect the processing of images by the computer. But I can’t tell how much.

Answer 1

I don’t see much of a question here, but you are right. It is especially visible if processing a JPG image with a medium compression ratio — the 8×8 squares of which JPG’s are built of tend to get more visible after filtering.

This is comparable to the rising of computational error when operating on computer-based floating point numbers.

Your best bet for image processing is using lossless formats for image processing — PNG’s are a good choice, cause they both provide lossless compression, as well as a decent support for bitdepths, transparency and are browser-compatible.

Another format, more often used in the professional world are TIFF’s (Targa).

However, note that if your source image is already in a loss-based format, converting it to a lossless one will only prevent adding additional artifact’s, not spreading and enhancing the old one. You can however reduce the extent of error by converting it to a lossless format and running it through a small seed gaussian blur.