A QUANTIZATION PARAMETER SELECTION ALGORITHM FOR SELF-ADAPTIVE IMAGE CODING

As a key part of image coding, quantization can improve compression efficiency and eliminate information redundancy. Considering the optimization trade-off of bit rate and the complexity of algorithm in image coding, based on the DZ+UTQ quantizer, a quantization parameter selection algorithm for self-adaptive image coding is designed. According to the transformed DCT location in the frequency domain, the image was divided into 64 independent sources, and each source was assumed to follow the Laplacian distribution. Under the guidance of the anti-water injection algorithm, it allocated equal budget distortion to each source to be quantified and optimized the quantization step size for each source. On this basis, the rounding offset of DZ+UTQ was introduced to adjust the dead zone adaptively, so as to achieve the purpose of optimizing the coding bit rate. Compared with the quantization algorithm with fixed rounding parameters, the results show that the image reconstruction quality of the proposed algorithm is higher, and almost no extra complexity is introduced.