Skip to Main Content
In previous years, error recovery on image compression methods, especially on VQ, has been an intensive research field. This paper presents a new pixel-based image reconstruction method (PIR) for recovering the lost blocks caused by transmission errors over noisy channels. Since there are high statistical correlations between the neighboring pixels of an image, PIR reconstructs the lost pixels using codebook search according to their adjacent pixels. In addition, an accelerating technique to shorten the execution time of the reconstruction is also proposed. The experimental results have shown that our PIR algorithm can efficiently and effectively recover the lost pixels. Moreover, if the image is complex or the size of the lost block is large, PIR is superior to the previous work, side-match vector quantization (SMVQ). Besides, unlike SMVQ, which is only suitable for the VQ encoding, PIR could be applied for any image compression algorithms such as JPEG, DCT and so on.