基于自适应阈值区间的广义Hough变换图形识别算法

Pattern identification algorithm with adaptive threshold interval based extended Hough transform

针对Hough变换检测算法计算量大、参数空间中峰值点被次峰值点包围易导致漏检或误检、阈值难以设定等问题,在分析Hough变换原理以及直线与圆弧检测特性的基础上,首先将图像边缘像素点根据邻接关系聚集为不同的类别,顺序存放到数组中,其次提出采用阈值区间的概念检测直线与圆弧,对数组中动态采样得到的像素点按照一定规则进行阈值区间的自适应估算,并依据阈值区间对存储的像素点进行直线和圆弧检测。最后使用GREC2003标准集和人工构造的图形集进行测试,通过对比相关改进算法的实验结果,验证该算法在减小计算量的同时,提高检测的效率。

Aiming at the problems that Hough transform detection method requires large amount of calculation,causes false or missing detection when the peak is surrounded by the sub peak in the parameter space,and it is difficult to set the threshold,this paper proposes a new algorithm and suggests the following improvements based on analyzing the principle of Hough transform and the detection characteristics of straight line and arc.Firstly,the image edge pixels are identified into different classes according to their adjacent relation and sequentially stored in the data array.Secondly,the concept of threshold interval is adopted to detect straight line and arc,and then the dynamically sampled pixel points in the array are used to estimate the adaptive threshold interval according to certain rules.Finally,the straight line and arc are detected from the stored pixels using the obtained threshold interval.The GREC2003 standard set and artificially constructed graphical set were used to conduct experiment tests.The result comparison between the relevant algorithm and the improved algorithm illustrates that the proposed algorithm can reduce the computational task and improve the detection efficiency at the same time.