【作者】 陈洪波； 郭国强；

【Author】 Chen Hong-bo, Guo Guo-qiang(Department of Compute Science and Technology, Hunan University of Arts and Science, Changde, Hunan, China, 415000)

【机构】 湖南文理学院计算机科学与技术系；

【摘要】 提出了一种有效的角点提取的方法。在该方法中,角点将被定义为二条(或二条以上)直线的交点,并利用Hough变换来检测直线。若存在二条(或二条以上)直线经过某点,则该点就被认为是角点(角点候选点)。还对Hough变换中的映射方式作了改进,在传统的映射中采用的是点一线的映射方式,而改进方法所使用的是点-多线的映射方式。在这种映射方式下,Hough变换的参数空间不会发生峰值扩散的现象,这样会更有利于峰值的检测(直线检测),从而更有利于角点的检测。实验证明,该方法能精确地提取出图像中的角点。更多还原

【Abstract】 A new and effective method of corner detection is proposed. The corner is defined as the point of intersection of two (or more) lines, and the Hough transform is used to extract the lines. The feature point, by which there are two or more lines passed, would be regarded as corner (or candidate corner). The mapping scheme in Hough transform is improved. The point-line mapping scheme, where a pixel in image is mapped to a line (or a curve), is used in traditional Hough transform, but the point-multiline scheme is used in the new method. By the new mapping scheme, the peak value is not spread around in the Hough parameter space, which is beneficial to detect peak value point. Experiments show, using the new method based on improved Hough transform, the corner in the image can be detected accurately.更多还原