基于边缘检测的编组法直线提取

Phase-grouping Lines Extraction Method Based on Edges Detection

对直线相位编组法进行改进,提出了一种在边缘检测基础上进行相位编组提取直线的方法。该方法充分利用了图像的梯度和幅值信息,能够比较准确地提出物体的边缘直线,克服了相位编组法中边缘灰度幅值信息利用不够、容易生成多边缘、不利于突出物体主边缘直线的缺点,在大比例尺城市房屋的提取中能发挥比较好的辅助作用。

The paper presents a new phase-grouping lines extraction method based on edges detection, which improves the traditional phase-grouping lines extraction method greatly. The method can make full use of the gradient orientation information and the magnitude information in the image, so it can extract the objects' straight edges accurately. While the traditional phase-grouping lines extraction method cann't utilize gradient and magnitude information efficiently, so it often extracts lines inaccurately and incorrectly and even produces multi-edges. Using the method presented in this paper, we can overcome such shortcomings and help the task of extracting buildings from large scale urban images.