Study Of Rate Control, Quantization, And Encoding Technologies In Image Compression

The image compression technology is extremely important for transmission, process, storage, and display of image signal. This paper researches the rate control, quantization, and encoding technologies in image compression system. The main content is as follows:1. An efficient rate control method, which is designed for lossless mode of JPEG2000, is proposed. This method uses a fractional upward shift of bitplane based on 5-3 wavelet filter to simulate more precise subband weighting. Compared to the previous integer upward shift of the bitplane, it achieves superior performance at most bit-rates without the cost of increased computational complexity.2. A modified embedded Trellis Coded Quantization (TCQ) for JPEG2000 is presented. The modified embedded TCQ formulation is more reasonable than before. For example, the reconstruction value falls closer to the middle of the corresponding quantization interval in default condition. Experimental results show that this minor modification yields significant performance improvement compared to the original one in JPEG2000.3. A context-based inverse quantization method is proposed and its application in wavelet image compression is presented. The proposed method breaks the traditional one-to-one mapping of the quantization index to reconstruction value in inverse quantization while maps an index to several different reconstruction values according to the corresponding contexts of the index. This method can reduce the quantization distortion significantly. The reconstruction values of contexts need to be sent to decoder as side information.4. Three modified zig-zag scan modes are proposed, which are based on the energy concentrating property of DCT and the directional character of three DWT details subbands HL, LH, and HH. These scan modes make the energy of a 2-D block concentrating on the front-end of the scanned 1-D sequence. This benefits entropy coding a lot.5. An adaptive coding method of recording the locations of non-zero coefficients is proposed. There are many modes to encode the quantized transform coefficients of image/video data. For example, sample by sample, quartation, point coordinates, and so on. The proposed method adopts there all modes as the optional modes to encode a square while the final one indeed used to encode this square is selected by encoder.6. An improved variable length coding method is proposed. The code word of variable length code comprises of prefix and suffix. No code word is a prefix of any other code word in the set. This paper finds the uneven distributions of "0", "1" symbols in prefix are the main redundancy within the variable length codes. Therefore, this paper proposes to group the prefixes of a number of variable length codes together after traditional variable length coding to further compress.