基于SPIHT算法的改进ROI图像编码

将分层树集合划分算法(Set Partitioning in Hierarchcal Trees, SPIHT)引入到感兴趣区域(Region of Interest, ROI)编码中。对前n个高比特平面进行编码时，将ROI区域与背景无区分进行编码。对n个高比特平面之后的平面进行编码时，将背景区域进行清零操作，ROI区域继续用SPIHT算法进行编码。此算法的优点：位平面数减少；整个图像的基本信息能够优先被恢复出来。在解码端，依靠原有的SPIHT解码程序就可以将其解码出来。可用于现有的SPIHT算法应用中，并且能够很好地恢复感兴趣的图像。

Improved ROI image coding based on SPIHT

There are two ROI(region of interest) coding methods in JPEG2000: max shift and scaling-based shift method, which can reconstruct ROI at frist. SPIHT, a very efficient embedded zerotree wavelet coding, was adapted in ROI coding. The frist n high bit planes were encoded in no difference between ROI and backgroud. The rest bit planes were encoded with zero for backgroud, but SPIHT for ROI. The merit of this method is that the number of bit planes is reduced and the base information of whole image is rebuilt at first. The image can be decoded applying original SPIHT decoding program. This method can be used in the existing SPIHT method and rebuilt ROI.