机构地区: 中国人民解放军第一军医大学生物医学工程系医学图像处理全军重点实验室
出 处: 《计算机学报》 2002年第11期1170-1174,共5页
摘 要: 该文提出一种图像编码的新算法(EBOHT),该算法先将图像进行小波变换,然后按空间位置对应关系将小波系数分成若干块,每一块数据采用一种新等级树编码,比传统的SPITH方法增加了预测过程,进一步利用了相邻小波系树的相关性.块间进行基于率失真曲线的优化方法,使每一块图像有合理的比特分配.同时由于数据是分块编码的,因此该算法较SPITH有更好的容错能力,且易于实现感兴趣区(ROI)编码. A new image compression algorithm(EBOHT) is proposed in this paper. Firstly, a wavelet transform is used to an image, then the transform coefficients are divided into many blocks according to spatial position. For each block coding, authors adopt a improved SPIHT algorithm which adds a predictive procedure from traditional SPIHT to exploits the correlation a-mong the adjacent coefficients. At last, a optimization method based on rate-distortion curve is proposed to make reasonable bit-allocation for all blocks. Due to the arranging the datum into blocks, so this algorithm is better than SPIHT about error resilience and it is easy to achieve coding of images with Regions of Interest (ROD.
关 键 词: 零树 内嵌块优化等级树算法 图像编码 小波变换 图像处理
领 域: [电子电信] [电子电信] [自动化与计算机技术]