基于相对最值点的最小二乘直线检测方法

为了解决传统目标检测中遇到的检测速度慢、精度低等问题，本文提出一种新的图像特征——相对最值点，并依据相对最值点提出一种新的最小二乘直线检测的方法。该方法首先搜索Canny边缘的闭合包络，并寻找闭合包络的相对最值点。包络上相邻相对最值点之间长度满足阈值要求的即为疑似直线的两端点，进而利用最小二乘拟合方法获得该疑似直线的拟合方差，最后利用拟合方差与拟合直线长度之比作为直线判定的标准用来检测出图像中的直线。实验表明，本方法与传统方法如PPHT直线检测、LSD直线检测方法等相比，检测直线的精度更高，检测速度更快。

Least Square Line Detection Method Based on Relative Maximum Point

In order to solve the problems of slow detection speed and low accuracy in traditional target detection, This paper presents a new image feature pair maximum point, and proposes a new method of least squares line detection based on the relative maximum point. The method obtains the closed envelope of the Canny edge and looks for the relative maximum point of the closed envelope. If the length between the adjacent relative maximum points on the envelope satisfies the threshold requirement, the fitting variance of the suspected line is obtained by using the least square fitting method, and then the least square fitting method is used to obtain the fitting variance of the suspected straight line, Finally, the ratio of the fitted variance to the fitted straight line length is used as the criterion for the line determination to detect the straight line in the image. The experiments show that compared with traditional methods such as PPHT and LSD, the method has higher precision and faster detection speed.