摘要： 在图像量测及物体精确定位中，Hough 变换是一种应用非常广泛的图像边缘检测技术，该文在提取空间立方体上分布的规则格网线，进而确定其格网点空间坐标的研究中，针对传统Hough 变换直线检测中易产生过连接线（点）、不易确定直线端点和长度以及定位精度低等问题，提出了一种适用于该类问题，基于Hough 变换空间直线（点）位置自动检测的新算法。实验结果表明改进的Hough 变换算法有效，空间交点坐标量测中误差达到子像素级。

关键词: 图像量测；Hough 变换；点位检测

Abstract: In images measurement and precision of position, the Hough transformation is a general method to extract the edge characters of the image. Based on the Hough transformation, a new automatic detection approach, which is better at the studies on regular grid extraction in spatial cubes and further on spatial coordinate extraction of grid points, is proposed in this paper. By using improved Hough transform, the lines in the image and their parameters can be steadily determined. Point of intersection caused by intersect of two neighboring lines then can be obtained more accurately and with insensitive to noise. The results indicate that the algorithm to calculate the locations of point of intersection given in this paper is valid and mean-root-square error of point of intersection achieves sub-pixel.

Key words: Image measurement; Hough transform; Point location detection