A new method for boundary artifact reduction in real-time high-speed image compression system

Image produced by CCD camera has very high resolution, and its output data rate is usually very large. Therefore, there are high requirements for image compression and its storage. CCD pictures are required to divided into several blocks for compression because image data can be processed with high speed in internal memory of DSP. But, in this way, there is to be severe boundary artifact in encoding procession. In this paper we present a new method to reduce boundary artifact based on symmetric-extend wavelet transform and overlapped data. A concrete application to image "lena" shows that this method can improve peak signal-to-noise ratio (PSNR), and acquire satisfactory visual quality of the reconstructive image. Comparing with the image without boundary artifact reduction, the reconstructed image has smaller MSE by using the new method.