基于机器学习算法的图像边缘识别  

Image Edge Recognition based on machine learning algorithms

作　　者：朱建洲 韩清宇 林佳涛 徐小雅 Zhu Jianzhou;Han Qingyu;Lin Jiatao;Xu Xiaoya(Guangdong University of Finance and Economics,Guangzhou 510320,China)

机构地区：[1]广东财经大学,广东广州510320

出　　处：《无线互联科技》2023年第4期125-128,132,共5页Wireless Internet Technology

摘　　要：文章首先改进边缘检测算法对图像边缘进行定位,使用双边滤波代替高斯滤波进行去噪处理,利用灰度相似性与几何相似性的权值分配边缘特征,提高精度;其次,在提取边缘坐标工作中利用多项式插值的方式提取出亚像素级精度的图像边缘坐标表示;最后,利用边缘轮廓追踪分析法实现了边缘线段分割,并将直线型、圆弧型以及椭圆弧型边缘进行曲线拟合,实现边缘几何属性提取。This paper firstly improves the edge detection algorithm to locate the image edges,and uses bilateral filtering instead of Gaussian filtering for denoising processing.The edge features by gray similarity and geometric similarity are used to improve the accuracy of result.Secondly the edge coordinate representation with subpixel accuracy is extracted by polynomial interpolation.Finally,this paper uses edge contour tracing analysis.And straight line type,circular arc type and elliptic arc type edge curve fitting,achieves the edge geometric attribute extraction.

关 键 词：亚像素边缘提取 边缘拟合 边缘提取算法 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]