基于Hough变换的线段检测算法  

Line segment detection algorithm based on Hough Transform

作　　者：杨全银[1] 魏雅杰[1] 王欣[1] 

机构地区：[1]山东大学信息科学与工程学院,山东济南250100

出　　处：《红外与激光工程》2009年第3期525-529,共5页Infrared and Laser Engineering

基　　金：国家自然科学基金资助项目(60872119)

摘　　要：为了克服Hough变换需要大量存储资源的缺点,在分析传统Hough变换定义方式的基础上,提出了一种新的改进算法。该算法采用多点对应一点的匹配映射规则,对参数空间中的线段形参数单元进行逐个单独处理,寻找图像空间中与之对应的模板,只有那些符合模板匹配条件的图像特征点才对该单元进行投票,通过动态寻找局域峰值点以及记录该峰值点所对应图像空间中线段的端点位置完成图像中的线段检测。该算法不需要预先存储整个参数空间,可以节约大量的存储资源,保持了与传统Hough变换相同的精度,且易于并行处理。In order to overcome the shortcoming of the Hough Transform, which needs a great deal of memory resource, a new improved Hough Transform algorithm was presented based on the analysis of the conventional Hough Transform. Line-segment parameter cells were processed independently in the parameter domain to search corresponding template in the image domain using many-to-one matching mapping rule. Only the feature points in the image that accorded with the template＇s matching conditions could vote for the corresponding accumulator cells. The line segments detection was achieved by seeking the local peak dynamically and noting the coordinates of＇ the line ends related to the local peaks.The algorithm needn＇t prestore the whole parameter domain so that lots of memory resource can be saved. It preserves the same precision as the conventional Hough Transform and can be processed in parallel.

关 键 词：HOUGH变换 线段检测 特征点匹配 参数单元 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]