Subimage error concealment techniques

Images transmitted via ATM networks suffer from quality degradation due to buffer overflow or cell header errors which cause ATM cells to be lost. This paper presents a new approach to conceal the errors in the received images by the application of novel error recovery techniques to the decomposed DCT-coefficient subimages of the corrupted image. These techniques were developed to recover images corrupted by impulsive noise. Since decomposing the corrupted image into the DCT-coefficient subimages generates low resolution images corrupted by impulsive noise, all the techniques used to recover images corrupted by impulsive noise can be used to recover the subimages and hence the corrupted image. In this paper, we study the performance of different iterative and non-linear techniques to recover the corrupted subimages. The quality of the recovered image using these techniques is better than the quality obtained by many classical error concealment techniques