一种基于模糊阈值的直线拟合策略在铁路沿线柱体检测及距离估计中的应用

首先提出了一种单幅图像中由边缘检测自动估计铁路沿线柱体到铁轨大致距离的方法，该方法主要通过对铁轨及其沿线柱体的检测、分类识别和距离估计来完成。因为如何在有畸变和复杂背景的图像中准确和有效检测边缘在图像处理和模式识别中一直是一个关键而困难的问题，为此提出了一种基于模糊阈值的直线连接拟合策略，该策略主要包括3个步骤：边缘提取、角点检测和基于模糊阈值的直线递归拟合。该策略可以有效地在有图像畸变和复杂背景的实际拍摄图片中通过参数控制获取感兴趣目标的直线边界。实验结果证明本文的直线拟合策略是精确的和具有鲁棒性的，距离估计方法是有效的。

A Line Extraction Algorithm in Poles Detection Based on Fuzzy Threshold Applied and Distance Estimation

A novel method is presented to estimate the distance between the railway and the poles alone the railway in one image based on the edge extraction of the poles. It has three main parts which are poles and the railway detection, poles classification and distance estimation. In the image with distortion and complex background, how to effectively and accurately detect the edges of the target is a crucial and difficult issue for image understanding and pattern recognition. This paper presents a robust straight-line extraction algorithm to solve this problem. This algorithm can extract the exact straight line edges of interesting target in the image with distortion and the complexity of scenes based on a fuzzy threshold and line fitting. The three main steps of the algorithm are edge detection, corner detection and the straight-line fitting based on fuzzy threshold method. The experimental results on real images demonstrate that the straight-line extraction algorithm we proposed is superiorly accurate and effective, and the distance estimation method is effective.