顾及建筑物框架线的直线提取方法性能分析

Performance Analysis of Line Extraction Methods Considering Building Frame Lines

图像是人类认识世界的基础数据之一,对人类社会的发展有着至关重要的作用,而直线是构建图像最基本的元素,精确的建筑物框架线检测对室内外自主导航领域的发展具有重要意义。目前主流的直线检测方法有LSD、Hough以及基于卷积神经网络的提取方法,为对采用三种方法对建筑物框架线提取性能进行分析,文章采用了多种场景下的图像数据进行对比实验,并分析直线检测结果及影响因素。实验结果表明,Hough变换效果在三中算法中效果最差,在Wireframe数据集测试中,与LSD相比,在两个指标上分别提高了31%和20.2%。

Image is one of the basic data for humans to understand the world,and plays a crucial role in the development of human society.Straight lines are the most basic unit for constructing images.Accurate line detection is of great significance for the development of fields such as road recognition,building structure extraction,camera calibration,and so on.Currently,the mainstream line detection methods include LSD,Hough,and convolutional neural network based extraction methods.Multiple sets of comparative experiments are conducted using image data from multiple scenes to analyze line detection results and influencing factors.The experimental results show that Hough transform has the worst effect among the three algorithms,and in the Wireframe dataset test,compared with LSD,it has improved by 31%and 20.2%on the two indicators,respectively.