摘要
从ATM网络出发,提出了一种基于子带的图象分层编码技术。通过子带分解,将图象信息分为低频带和高频带两层,对低频带采用标准化的运动补偿帧间预测、DCT和变字长编码(VLC),以保持兼容性;对高频带,则根据其特点,用视觉模式进行边缘编码。与ATM信元的优先级对应,低频带信息通过高优先级信元传输,而高频带信息有丢失的可能,通过低优先级信元传输。模拟实验说明,这种方案可以减小信元丢失对图象质量的影响。最后,我们提出了一种新的重建丢失信元的边缘定位补偿算法,对丢失的高频子带从视觉意义上给予恢复。
A larered image coding scheme based on subband decomposition is presented for ATM networks. It divides image information into two classes according to spatial frequencies of subbands. The Low-frequency band is assigned a high priority, and coded by means of standard modules such as MC/DPCM, DCT and VLC, to keep its compatibility. The high-frequency bands, which are liable to be discarded during channel congestion, are assigned a low priority and coded by visual pattern. Computer simulations show that image quality degradation caused by cell loss could be reduced through this way. Furthermore, A new alogrithm, edge location Compensation (ELC), is presented to reconstruct the would-be lost cells of high-frequency subbands via edge matching.
出处
《北京工业大学学报》
CAS
CSCD
1996年第4期60-73,共14页
Journal of Beijing University of Technology
基金
国家"八五"攻关课题
北京市自然科学基金