摘要
归一化灰度组合法是基于灰度组合矩阵的一种灰度相关法 ,主要利用相似图像间像素灰度组合应该最少的原理进行图像匹配。它可以统一地表示其它灰度相关法 ,解决了相关法对噪声和灰度变化敏感的问题。算法中主要涉及整数运算 ,与其它相关法相比 ,运算速度更快 ,并且可以通过灰度量化使归一化灰度组合法速度更快 ,匹配度更高 ,匹配唯一性更好。实验表明减少灰度量化级数不影响图像匹配值的大小。
Normalized intensity combination method is based on intensity combination matrix extracted from the images. It is based on the fact that the intensity combination should be minimal between two similar images. It can represent other intensity correlation methods in a uniform way. It solves the problems of sensitive to noise and illumination change, which most correlation methods suffer. Most calculations in this algorithm are integer computations, so it is quicker than other correlation algorithms. Moreover intensity quantization can make it run more quickly and have higher matching quality. The experiment sho ws that the decreasing of intensity quantization's level will not make similar matching metric value lower. \;
出处
《红外与激光工程》
EI
CSCD
2000年第5期5-9,共5页
Infrared and Laser Engineering
关键词
图像匹配
灰度组合矩阵
计算机
图像处理
Image matching
\ Intensity combination matrix
\ Intensity correlation
\ Intensity quantization
