Research On Key Technologies Of Distributed Video Coding

This paper focuses on the research and improvement on the correlation noise modeling and quantification these two key technologies based on the in-depth analysis of DVC.Firstly, this paper proposes a hybrid distribution correlation noise model(HDCNM) based on K-Mediods and the algorithm to estimate its parameters. The HDCNM uses K-Medoids clustering algorithm to cluster the center part and the tail of the AC sub-band residuals, while using Laplacian distribution to describe the distribution of residual coefficients of the center part and using Cauchy distribution to describe the tail parts. The experimental results show that the proposed hybrid model can describe accurately correlation noise distribution, so as to improve the performance of the distortion rate of the transform domain DVC system effectively.Transform domain distributed video coding uses a 4×4 quantization matrix to quantize the DCT coefficients, but such scheme has fixed quantization matrix, which can’t adjust the quantization matrix according to the characteristic of video. For this shortcoming, this paper proposes an adaptive quantization(Adaptive Quantization, AQ) scheme. The encoder will use the simple correlation noise model to calculate threshold to achieve adaptive quantization effect. The AQ can improve the rate distortion performance and guarantee the distributed video coding a simple enough encoder.Finally, on the basis of the existing DISCOVER video coding framework, this paper will designs a advanced distributed video coding system by using the proposed HDCNM and the proposed adaptive quantization algorithm. The experimental results prove that the the overall system rate distortion performance was improved, the quality of the reconstructed image is also improved. So the proposed system can improve performance for monitoring system, record the occurrence of incident better.