摘要
由于Harris角点理论上具有不随光照条件改变而改变及旋转不变性等优点,被广泛应用于计算机视觉系统中。但实际应用中拍照光线的变化往往会引起图像的灰度失真,本文对Harris角点提取和SAD、SSD、NCC三种匹配算法在灰度变换下的性能进行了分析。分析结果表明,Harris角点提取具有较好的光照条件鲁棒性;而NCC是三种匹配算中灰度变换影响最小的。
When the images are taken with the variance in light rays and the angle transformation, The Harris corners of these images are not change. So The Harris corners are extensively applied in Computer vision system. However, the gray of images is usually not true if the images are photoed with the variance in light rays. The performance of Harris corner, SAD, SSD and NCC is evaluated. The evaluation result shows that the Harris corners are robust to variance in light rays and the NCC is least affected by gray transformation.
出处
《软件》
2016年第11期19-22,共4页
Software
基金
国家自然科学基金资助项目(61274133)
关键词
图像匹配
特征点
角点
灰度变换
Image matching
Feature points
Corner
Gray transformation
