encoding the diffs - huffman

[suman1209]

look into huffman encoding

encode the diffs with lower number of bits(for smaller values) with anomalies (position encoding)

[billy0705]

Huffman Coding with Python

source 1 \ source 2

[suman1209]

next items:

• [ ] save the encoded huffman encoding as a binary file and check the os size and comapare with manual calculation
• [ ] implement the decoder
• [ ] RLE before huffman (experiment different traversing directions e.g zigzga)
• [x] Huffman code for RGB instances and compare current flattened Huffman encode @billy0705

[billy0705]

The frequency for pixels, there are 2422 different kinds of diff pixels. The most frequent one is the [0,0,0], it appears 5139 times.

The result from the first frame of this approach is as follows:

[billy0705]

The result for separating the channels, and applying Huffman encoding on each channel. The result is a bit worse than only doing one Huffman encoding with the whole frame.