摘要
总结了目前二维灰度图像的亚像素边缘检测算法,针对它们存在的原理误差、计算复杂、用时长及不能通用等问题,提出了一种新的亚像素边缘检测算法.分析了3种基本边缘(阶跃型边缘、脉冲型边缘、屋脊型边缘)的特点,利用一维质心算法对这3种边缘特征计算上的通用性及简易性进行了二维推广,得到了一种具有快速通用性的亚像素边缘检测算法.并在此基础上针对提高离散化过程中的精度问题,引入了高斯卷积平滑的预处理方法;引入了Sobel算法对图像像素进行了筛选,进一步提高了计算速度.通过实验验证了此算法的有效性,并分析了误差产生的原因.
Based on the summary of subpixel edge detection algorithms now mainly used on two-dimensional image,a novel universal and fast algorithm has been put forward to solve problems that exist in old algorithms such as principle error,calculation complexity,consuming long time and not universal,etc.Based on the analysis on the characteristics of three kinds of basic edge（step edge,pulse edge,roof edge）,the algorithm was extended to two-dimensional based on the universal and simple characteristics of one-dimensional mass center algorithm,meanwhile Sobel algorithm and convolution fuzzy algorithm were introduced to shorten operation time and improve precision.The efficiency of the algorithm was proved by experiment and the reason for errors was analyzed.
出处
《应用科技》
CAS
2010年第8期35-39,共5页
Applied Science and Technology
关键词
快速通用算法
卷积平滑
亚像素边缘检测
fast and universal algorithm
convolution smoothing
subpixel edge detection
