摘要
分析了对比度变化对Hu矩的影响,构造了一种新的不变矩形式,该矩不但具有通常的旋转、尺度缩放以及平移不变性,同时还具有对比度变化不变性。以三类具有旋转、尺度缩放、平移以及对比度变化(RSTC)的目标图像为例进行了识别仿真实验,结果表明新的不变矩形式消除了对比度变化带来的影响,增强了三类图像的类内内聚性和类间可分性,实验结果证明能够对具有RSTC变化的目标图像进行有效识别。
The influence of contrast change on Hu moments and is analyzed and a new kind of invariant moments is proposed. The moments have the accustomed invariance of rotation, scaling and translation, and contrast change. A simulating experiment is carried out using three kinds of target images with rotation, scaling, translation and contrast change （RSTC）. Results show that the new invariant moments have eliminated the influences of contrast change and strengthened the centrality within a class and the separability betweenclasses. They can also be used to recognize the target images with RSTC change effectively.
出处
《数据采集与处理》
CSCD
北大核心
2006年第2期225-229,共5页
Journal of Data Acquisition and Processing
关键词
模式识别
畸变不变识别
不变矩
对比度
pattern recognition
distortion invariant recognition
moment invariant
contrast
