一种子像素级图像测量直线提取算法  

Image Lines Feature Extraction with Sub-pixel

作　　者：张春森[1] 姜楠[1] 

机构地区：[1]西安科技大学测绘科学与技术学院,陕西西安710054

出　　处：《中国矿业大学学报》2009年第6期903-908,共6页Journal of China University of Mining & Technology

基　　金：陕西省教育厅科研计划项目(06KJ239)

摘　　要：分析了图像量测中线特征目标高精度的提取,提出了一种子像素级图像测量直线提取算法.采用改进的Hough变换对线状特征进行提取,获得具有确定长度和端点的直线段;以此直线段的端点为初值,自适应生成直线模板,利用最小二乘方法进行模板与图像线特征匹配,实现图像线特征的子像素提取.对仿真及真实图像的验证结果表明:该混合算法稳定可靠,达到了子像素级的提取精度,在图像测量中能有效地提高测量精度.The method of image line feature extract with high precision was studied based on the significance of line feature extraction in image measurement. A new mixed algorithm was proposed and the procedure that was taken in the algorithm can be divided into two steps. First the lines were extracted with improved Hough transform, the lines segment and its end were obtained. Then the self-adaptation line template was produced based upon initial value of lines segment＇s end. Matching template with the line feature of image by the method of least squares image matching, the line feature with sub-pixel was realized. The results with simula- tive and real data show that the algorithm is robust and its precision can achieve to sub-pixel level. It can improve the precision of the image measurement effectively.

关 键 词：线特征提取 最小二乘法 图像量测 HOUGH变换 子像素 

分 类 号：TN911.73[电子电信—通信与信息系统]