Compression Adapting To The Channel And Error-Resilient Coding Based On DCT For Image Transmission

In this paper, I present a compressing adapting to the channel and error-resilient coding based on DCT image transmission technique based on researching the congestion control and the error control in the fields of network and wireless multimedia. In the congestion, I illustrate an image compressing technique adapting to the channel by actively discarding the low-frequency part of some image blocks. In this technique, the sender discards the low-frequency part of some image blocks and adjusts the step size of quantization to meet the need of bandwidth, the receiver retrieves the discarded low-frequency to recover the original image using the original image using the received high-frequency part of some image blocks and adjacent complete blocks. This technique will reduce the deviation of low-frequency part by adding the step size of quantization for some blocks when the low-frequency part is retrieved; it will make the quality of the recovered image even. So, an image with global good visual effect will be gained.In error control, I first introduce a error-resilient image coding with smart-IDCT error concealment technique. Then, I integrate it and above image compressing technique adapting to the channel by actively discarding the low-frequency part of some image blocks by changing it. It consists of error-resilient image coding and post-processing error concealment. The former prevents errors from propagating across image block boundaries with little overhead. The latter uses frequency interpolation and discards suspicious ac values to retrieve low-resolution image rather than a totally corrupted one. I change it in some degree. In the partial discardable image layer, it does nothing about dc values dependence between image blocks since these blocks don't contain dc values. This further reduces the data redundancy. When the dc value of the basic image layer is erroneous, it is replaced by the average dc value of the neighboring blocks in four corners. When the dc value of the partial discardable image layer is erroneous, it is replaced by the average dc value of the neighboring blocks in four borders.Experimental results show this technique works better than the digital image transmission technique with actively discarded image blocks adapting to the channel and the adjusting step size of quantization technique. It efficiently prevents the errors from propagating across image block boundaries and realizes error correction and concealment.