Research On Reconstruction Algorithm In Distributed Video Coding

With the high development of wireless network, more and more mobile videoterminals are used in multimedia communications. The encoder should below-complexity and error resilient. The traditional video compression standards such asMPEG and H.26x could not be used in the above-mentioned application environmentbecause of the complicated encoder and the encoded video data is highly sensitive toinformation loss. Therefore, the scholars around the world proposed a novel videocompression method, DVC (distributed video coding) technology, which has recentlybecome a research focus of distributed video coding.In the unidirectional DVC system, the reconstruction technique at the decodergreatly affected Wyner-Ziv (WZ) frame decoding quality. Based on the theoretical basisof the DVC, the thesis firstly proposed reconstruction algorithm based on the quantifiedcoefficient of WZ frame, which is able to adaptively adjust the contribution of thequantized side information (SI) and the decoded quantized coefficient to thereconstruction，with the help of the correlation obtained from the encoder and thequantized WZ, thus completing the reconstruction. Secondly, the reconstructionalgorithm based on SI confidence applied in bit error channel is proposed, according tothe decoding distortion analysis, correlation noise model based on the quantized SI isproposed, and the confidence of quantized SI is calculated to modify the decodedquantized WZ， thereby improving the quality of decoding.Experimental results show that compared with the existing minimum mean squareerror (MMSE) reconstruction algorithm, the proposed reconstruction algorithm caneffectively improve objective performance PSNR(Peak Signal to Noise Ratio) by0.3dB to2dB; Compared with H.263coding method, the reconstruction video quality ofthe proposed Wyner-Ziv video coding is obviously enhanced, besides, the proposedencoder is simple and easy to realize，then it can be applied to changeful network andthe video communication systems for low-complexity encoding terminals.