基于Otsu阈值的MSI不连续破损边缘提取  

Edge Extraction of Discontinuous Damage in Multi SceneImage Based on Otsu Threshold

作　　者：李德新[1] LI De-xin(Science and Technology College Gannan Normal University,Ganzhou Jiangxi 341000,China)

机构地区：[1]赣南师范大学科技学院,江西赣州341000

出　　处：《计算机仿真》2020年第9期358-362,共5页Computer Simulation

基　　金：江西省教育厅科技项目(GJJ181541)。

摘　　要：由于多场景图像环境复杂,边缘易出现破损问题。当前方法图像边缘提取结果错分概率较高。为此提出基于Otsu阈值的多场景图像不连续破损边缘提取方法。对Otsu阈值图像预处理增强图像对比度,通过中值滤波去除图像内噪声。对图像采取二值化处理,运算类间方差最大值,获得最优阈值。将像素邻域滑动窗口划分为多数不重叠的子区域,去除区域亮度对边缘强度产生的影响和虚景边缘。通过对边缘像素点运算,确定边缘精度,实现对多场景图像不连续破损边缘的有效提取。实验结果表明:所提方法能够提取到较高精度的破损边缘,且运算耗时远远低于现有方法,具有较高应用价值。At present,the multi-scenario image environment is complex,and its edge is easy to be damaged.Due to high misclassification,this paper presented a method to extract discontinuous broken edges in multi-scenario image based on Otsu threshold.At first,we need to enhance the image contrast by preprocessing Otsu threshold image,and removed the inner noise by median filter.Based on the binarization of image,we calculated the maximum variance be⁃tween classes to find out the optimal threshold.Then,we divided the sliding window of pixel neighborhood into some non-overlapping sub-regions to remove the influence of regional brightness on edge strength and the edge of virtual scene.After calculating the edge pixels,we determined the edge precision.Finally,the discontinuous broken edges in multi-scene image were extracted.Simulation results show that the proposed method can extract the damaged edge with high accuracy,and the computation time is much less than that of the existing method.

关 键 词：阈值 多场景图像 边缘提取 边缘算子 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]