用于12位灰度JPEG压缩的霍夫曼表

Question

I want to compress 12-bit grayscale images. I already developed an 8-bit grayscale compressor using the example Huffman Tabels (for Luminance) from the ITU T.81 in C++.

Are there any example Huffman Tables for 12-bit grayscale images? (Converting from 12-bit to 8-bit before compression is not an option)

Answer 1

Clearly there is no such animal in the standard. The code here is 8-bit only

http://www.colosseumbuilders.com/sourcecode/imagelib403.zip

but it shows how to calculate optimal huffman tables using pointers to functions. You could try something like that in your encoder. If you did not want to make two passes over the quantized DCT values, you could stick do this as a temporary measure, run a number of sample images, and create your own huffman table for reuse.