Research On Coding Of Compound Images

With the techniques of multimedia and network have been rapidly developed and widelyapplied in our society, it is becoming more and more convenient and quick for us to store,process and transmit digital multimedia signals. On the one hand, it is possible for theinformation makers to design their real time video codec systems based on common ordedicated processors under the capability development of those kinds of processors, on theother hand, the appearance of new video standards can improve the compression ratio andPSNR. But different picture compression standards have different backgrounds;singlecompression standard often achieves poor compression ratios for those compound images.The still picture compression standards can't get rid of temporal redundancy informationin moving pictures. While video compression standards will regard some details in still picturesas redundancy and throw them away. It will make the edge of some still pictures misty.Furthermore, video compression standards are always too complex to suit real-time systems.Therefore, it is necessary to adopt both compression standards for compound images inreal-time systems.The key point for this scheme is how to distinguish between still pictures and movingpictures. In this paper, we analyzed two kinds of methods first. The first is based on thedifference between two frames. Its main weakness lies in its high rate of misjudgments, slowresponses. The other is based on the statistic information in a frame. It has low rate ofmisjudgments and quick responses, but it can't distinction between still pictures and movingpictures.In this paper, we proposed a new method which is based on both the difference betweentwo frames and the statistic information in a frame. Firstly, we use the statistic information inthe frame to detect close regions in the frame through edge detection. Then, we make use of thedifference between two frames to estimate whether the regions are moving pictures or not. Weshowed how to carry out this method.In the end, we simulate this method. In our simulations, it can make quick responses to thechange of regions.