Low Bit-Rate Compression for Still Images using Wavelet

Amir Averbuch
Department of Computer Science
School of Mathematical Sciences
Tel Aviv University, Tel Aviv 69978
 
amir@math.tau.ac.il

The still compression algorithm is based on consecutive application of four different techniques:

1. Geometric multiresolution wavelet decomposition [4, 5]. It contributes a speedup factor of 4 in comparison to the regular 1-D tensor (separable) application of the wavelet transform[4]. It has a prefect reconstruction property which does not degrade or affect the quality.
2. Fast tree arrangement of the wavelet coefficients for efficient quantization classification. This includes efficient multiresolution data representation which enables fast traversing of the multiresolution wavelet tree[4, 7].
3. Quantization of the wavelet coefficients and the tree structure using bit-plane arrangement [1, 5, 6]
4. Entropy coding using adaptive arithmetic coding to pack the quantized wavelet coefficients and tree structure in a lossless manner. Each piece of information, the tree structure and the bit-plane description of the quantized wavelet coefficients, is packed into a single byte. The packing is forced to be on a byte boundary so we can reconstruct an image from every byte or bytes of the compressed image an additive mode. If we reconstruct using any byte or bytes from the compressed image we get the best reconstructed image out of these bytes. This enable to have a full flexible adaptive transmission.

The fast compression algorithm has an option of manual allocation of compression bits to specific area which enables a better reconstruction quality at a pre-selected zoom area[4].

• References
• About this document ...