CAJ | 학술논문

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.
위료감소도상처리중특정필배소수적시간화제고필배적준학성,연구료이특정점조성형상특정차이차형상특정재피수색적ROI구역내진행특정필배적방법.종영향필배정도화속도량개방면착수,이용쌍역치법해결특정제취대조성적민감성,인입선전화비례축방계렬모판해결모판여검측구역지간가능존재적일정각도선전화대소비례불일치등문제.재저분변솔모판화피수색구역진행특정조정위,재고분변솔모판화피수색구역상용조정위적결과진행쾌속필배화정위.
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 shape-feature is studied. Processed from matching accuracy and speed, the sensitivity issue is treated with bi-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 rough-locating result in high resolution area is performed.