【摘要】 提出一种基于计算机视觉任务的图像直线提取技术 ,首先用边界检测算子提取边缘 ,然后用链码将边缘表示出来 ,并记录链码的端点 ,即直线的端点 ,通过逐步删除对计算机视觉任务无意义的链码 ,得到图像直线的链码表示 ,最后用最小二乘法拟合图像直线的链码。该技术能尽可能多地从图像中提取直线信息 ,既可以较准确地提取直线的端点信息 ,又能得到直线的准确表示更多还原

【Abstract】 In this paper,a new method was presented to extract straight line from image based on computer vision task. Canny edge detector is used in this method to get edge, link the edge using Freeman chain code, and gradually delete no meaningful chain code to computer vision task. We use least square method to fit the straight line. The presented method can extract as much information as possible from original images. It excels both in getting the information about end points and in representing the straight line accurately.更多还原