Project: Image and Video Compression

PEOPLE

Zhihai He, Canhui Cai, Tian-Hu Yu,S.K.Mitra, 

OBJECTIVE

Data compression is an important component in many signal processing applications to save bandwidth and storage requirements. Over the last few years, the wavelet transform has become a very effective method of transform coding of images. We have developed several improvements to wavelet transform based compression algorithms.

In wavelet image compression, all information loss occurs during quantization of the wavelet coefficients. Based on the property of bi-orthogonal wavelets, optimal 2-D quantization error feedback filters are designed to reduce the reconstruction error. With very low complexity with regard to computation and implementation, the error feedback system improves the PSNR of reconstructed image by about 0.25 dB. In addition, due to its similar structure to the dithered quantizer, it also improves the subjective quality of the reconstructed image by reducing the contouring and ranging effects. Results of this investigation have been presented at an international conference [1].

Two basic approaches have been followed for developing image and video compression algorithms. One approach is based on the optimization of the quantization, data representation, and entropy coding to improve the coding performance. The improvement in the coding performance, which in most cases have not been that significant, has come at the expense of increased computational complexity and implementation cost. In the second approach, the objective is to develop coding algorithms with very low computational complexity and implementation cost but with competitive coding efficiency. We have investigated the second approach and have developed four different algorithms for wavelet-based image coding.

One of the coding schemes developed by us employs the multi-level dyadic wavelet decomposition, linear quantization with a proper dead zone, 1-D addressing complexity by raster scanning within subbands, variable length block coding, small alphabet representation of 1-D integer sequences, and adaptive arithmetic entropy coding. Despite the simplicity of the proposed coding scheme, the rate-distortion (R-D) performance of the proposed image compression algorithm is competitive with the best image coders in the literature. A paper describing this algorithm has appeared in the proceedings of an international conference [2].

In a variation of the above method, to explore the non-stationary nature of the subband data, we partition the decomposed image into 2 2 blocks after uniform threshold quantization. Each block is called an insignificant block and marked by a '0' if all the coefficients inside are zeros. Otherwise, it is called a significant block and marked by a '1'. The block marks are efficiently coded by the quadtree representation scheme. Each coefficient from the significant blocks is coded by the variable length integer (VLI) coding scheme. The integer size is compressed by a first-order arithmetic coder. The residue bits are directly sent out to the decoder. Despite its low complexity, the proposed coding algorithm has comparable performance with the SPIHT algorithm at very low bit rates, such as rates below 0.25 bpp. However, for higher coding bit rates it outperforms the SPIHT algorithm, especially for images with significant high-frequency components. This modified approach was reported at a conference [3].

Another coding algorithm of low addressing and implementation complexity has also been developed. It is based on partitioning uniformly quantized wavelet coefficients into multiscale blocks with each block classified as either an all-zero block or a non-zero block. The non-zero blocks are coded by a new method called zero mapping whose outputs are further compressed by a first-order arithmetic coder. The performance of the proposed coding algorithm compares favorably with that of some well-known coding algorithms, particularly for images with considerable high-frequency components. A paper describing this algorithm has been presented at an international conference [4].

We have found that with blockwise binary classification and data partitioning, the image subbands can be converted into the type of source data for which the trellis-coded quantization (TCQ) has the best quantization performance. Compared to the arithmetic coded TCQ (ACTCQ) and other TCQ-based coding schemes, this new algorithm significantly reduces the computational complexity, while performing competitively with the best available coding algorithms reported in the literature with regard to the rate-distortion performance. This algorithm has been reported at a recent conference [5]. A new image-coding scheme, called the stack X-tree coding, has also been developed. The new method replaces the zero runs in the stack-run coding by the X-tree. The X-tree is defined as a spatial hierarchical tree without significant descendants. By taking advantage of the inter-band dependencies, the X-tree can represent more efficiently insignificant coefficients in a wavelet image than the zero runs. Experimental results have shown that the performance of the proposed algorithm is better than that of the stack-run method, and is highly competitive with that of the SPIHT algorithm. This work has been presented at an international conference [6].

PUBLICATIONS

These materials are presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each authors copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.

1. C. Cai, T-H Yu, and S.K. Mitra, "Stack X-tree im­age coding," Proc. 2000 IEEE Asia Pacific Conference on Circuits & Systems , Tianjin, China, December 2000  - to be published.

2. Zhihai He, Tian-Hu Yu, and S. K. Mitra, "Fast and accurate rate prediction and picture quality control for wavelet image coding," Proceedings of 34th Asilomar Conference on Signals, Systems, and Computers, October 2000. [ PDF file ] 

3. Zhihai He, Tian-Hu Yu, and S. K. Mitra, "Fast image coding using blockwise binary classification and trellis coded quantization ," Proceedings of 34th Asilomar Conference on Signals, Systems, and Computers, October 2000. 

4. Tian-Hu Yu, Zhihai He, and S. K. Mitra, "A novel coding scheme for wavelet image compression," Proceedings of 34th Asilomar Conference on Signals, Systems, and Computers, October 2000. 

5. Zhihai He, Tian-Hu Yu, and S. K. Mitra, "Blockwise zero mapping image coding," in Proc. IEEE International Conference on Image Processing (ICIP 2000) , Vancouver, Canada,  September 2000.

6. Zhihai He and S. K. Mitra, "Optimum quantization error feedback filters for wavelet image coding, " in Proc. IEEE International Conference on Image Processing (ICIP 2000) , Vancouver, Canada, September 2000.

7. Tian-Hu Yu, Zhihai He, and S. K. Mitra, Simple and efficient wavelet image compression, in Proc. IEEE International Conference on Image Processing (ICIP 2000) , Vancouver, Canada, September 2000.