摘要
为了减少图像处理中特征匹配所需的时间和提高匹配的准确性,研究了以特征点组成形状特征且以此形状特征在被搜索的ROI区域内进行特征匹配的方法。从影响匹配精度和速度两个方面着手,利用双阈值法解决特征提取对噪声的敏感性,引入旋转和比例缩放系列模板解决模板与检测区域之间可能存在的一定角度旋转和大小比例不一致等问题。在低分辨率模板和被搜索区域进行特征粗定位,在高分辨率模板和被搜索区域上用粗定位的结果进行快速匹配和定位。
To decrease the time of feature-matching and improve the veracity during image processing, the method of constructing a shape-feature with character segments and performing the feature-matching process in the searching ROI area with this constructed shapefeature is studied. Processed from matching accuracy and speed, the sensitivity issue is treated with hi-threshold method. The rotating and scaling discord problems between sample template and monitoring area is resolved by a series of rotation and scaling template. The rough location in low resolution template and monitoring area is performed, and the fast precise matching and location with this roughlocating result in high resolution area is performed.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第5期1096-1099,共4页
Computer Engineering and Design
