Wavelet-based Video Encoder

Wavelet transform is a kind of transform encoding method in image processing. By decomposing original image to multi-resolutions in frequency domain, it provides ideal transform domain for image compression. Set Partitioning in Hierarchical Trees (SPIHT) method is a more effective wavelet coefficients scanning method originated from zero-tree method. The software and hardware designs of video encoder based on wavelet transform are studied in this paper. The key problems such as the selection of hardware source, work flow, algorithm, methods of code optimization and the high-speed board design, etc. are discussed in detail. Firstly, based on the research of wavelet transform and SPIHT coding compression method, the algorithm is programmed in C language and tested on PC. Secondly, a general video encoder hardware system is designed, where TI's TMS320C6201 DSP chip is chosen as its key processor. Lastly, the SPIHT algorithm is migrated on DSP. At last, the real time performance of the system is analyzed. The experimental results show that SPIHT algorithm achieves competitive performance with both extremely fast encoding and decoding speed and excellent reconstructed image quality at very low bit rates. The video encoder based on the general high performance DSP chip has especial advantages and is becoming more and more popular due to its high reliability, small-size, fast data processing, low power consumption and flexibility.