摘要
针对可见光与红外图像的特点和难点,提出了可见光与红外图像配准与融合中的关键技术,即:使用新型的基于一维最大类间方差和最大连通性测量的图像分割方法对源图像进行分割来更好地实行图像粗配准;使用新型的特征点提取方法,特征点的匹配及误匹配的消除来更好地实行图像精配准;采用新型的基于区域的树状小波活性测度计算来实现树状小波图像融合;利用自生成神经网络来实现模糊图像融合。
Aiming to the characteristics and difficulties of the visible and infrared,the key technologies and the problem-solving schemes are proposed in their registration and fusion,i.e.,the novel image segmentation algorithm is used based on the 1-D largest intra-class variance and the largest shape connectivity measurement to segment the source images for the better coarse registration,the novel feature point extraction method,feature point matching and mismatching removal methods are employed for the better fine registration,the novel region-based tree wavelet activity measurement calculation is applied to realize tree wavelet image fusion,the self-generating neural network is adopted to realize the fuzzy image fusion.
出处
《红外与激光工程》
EI
CSCD
北大核心
2006年第z4期7-12,共6页
Infrared and Laser Engineering
关键词
一维最大类间方差
最大连通性测量
特征点的匹配及误匹配的消除
树状小波图像融合
模糊图像融合
1-D largest intra-class variance
Largest shape connectivity measurement
Feature point matching and mismatching removal
Tree wavelet image fusion
Fuzzy image fusion
