摘要
提出了一种基于优化截取内嵌块编码的矩形感兴趣区域编码的干涉多光谱卫星图像压缩方法 ,不需要对小波域的系数进行提升 ,而是在码流组织时通过对多光谱区域自适应失真跟踪实现。解码器不需要进行小波域系数的逆提升 ,即可达到感兴趣区域的恢复质量要求。改变感兴趣区域的失真指标 ,只需要再次进行码流组织 ,而不需要重复进行优化截取内嵌块编码的T1编码 ,因此更为有效、灵活。同时提出了一种整型率失真斜率估计值快速算法 ,既可以保证率失真优化截取的精度 ,又避免了率失真斜率的浮点除法计算和浮点表示 ,降低了系统的复杂度。
A method is proposed for multi spectral image compression based on rectangle region of interest (ROI) coding of EBCOT, which does not lift the coefficients of wavelet domain and is complemented by code stream organization. When the code stream is organized, the distortion of ROI can be calculated approximately, which is used to adjust the code stream organization to meet the need of the peak signal to noise ratio (PSNR) of ROI. This procedure is adaptive. The decoder needs not to lift reversely the coefficients and works normally. If the distortion requirement is changed, the organization can be done again, but the T 1 coding of EBCOT needs not be done again, so the algorithm is more efficient and more flexible. A fast computation of the slopes is proposed, by which not only the precision of the slopes is enough but also the floating point division and floating point expression of the slopes are avoided, so the system complexity is decreased. This method is suitable for multi spectral image compression.
出处
《光学学报》
EI
CAS
CSCD
北大核心
2004年第3期299-303,共5页
Acta Optica Sinica
基金
"十五"军事通信技术预研项目(410 0 10 302)
国防科技预研基金(J2 0 12 1 DZ0 1)
高交时高光谱遥感图像压缩技术研究 (70 8 2 3 3)资助课题
关键词
信息光学
图像处理
图像压缩
优化截取内嵌块编码
多光谱图像
率失真
感兴趣区域
失真指标
information optics
image processing
image compression
embeded block coding with optimized truncation (EBCOT)
region of interest (ROI)
multi spectral image
rate distortion
