基于视觉特性的激光光斑图像分割方法研究  

Research on laser spot image segmentation method based on visual characteristics

作　　者：张可人 李夕雯 冯祥[1] ZHANG Keren;LI Xiwen;FANG Xiang(Chengdu College of University of Electronic Science and Technology of China,Chengdu 611731,China)

机构地区：[1]电子科技大学成都学院,成都611731

出　　处：《激光杂志》2023年第9期114-119,共6页Laser Journal

基　　金：四川省科技厅项目(No.2021JDKP0077)。

摘　　要：由于激光光斑图像的对比度不足,难以观察图像背景与光斑之间的差别,影响分割效果,因此,提出基于视觉特性的激光光斑图像分割方法。该方法将待分割的激光光斑图像做网格划分处理,计算出网格图像的平均灰度熵值,根据平均灰度熵值选取多个样本块,采用SSD距离准则合成最佳样本块。根据样本块背景和光斑梯度之间的视觉特性区别,增强处理激光光斑图像,提高图像的对比度。完成对比度增强后,通过像素点的灰度分布视觉特性确定神经元的相关参数,根据最大二维Renyi熵准则分割激光光斑图像。实验结果表明,所提方法的定位评价函数为0.0001,均方根误差为0.29,分割位置与光斑位置基本一致,分割精度接近100%。因此,说明所提方法的光斑定位误差小、均方根误差小、分割精度高。Because the contrast of the laser spot image is insufficient,it is difficult to observe the difference between the image background and the spot,which affects the segmentation effect.Therefore,a laser spot image segmentation method based on visual characteristics is proposed.This method divides the laser spot image to be segmented into grids,the average gray entropy value of the grid image is calculated,multiple sample blocks are selected according to the average gray entropy value,and the best sample block is synthesized by SSD distance criterion.According to the difference of visual characteristics between the background of the sample block and the spot gradient,the laser spot image is enhanced to improve the contrast of the image.After the contrast enhancement is completed,the relevant parameters of neurons are determined through the gray distribution visual characteristics of pixel points,and the laser spot image is segmented according to the maximum two-dimensional Renyi entropy criterion.The experimental results show that the evaluation function of the proposed method is 0.0001,the root mean square error is 0.29,the segmentation position is basically the same as the spot position,and the segmentation accuracy is close to 100%.Therefore,the proposed method has small spot positioning error,small root mean square error and high segmentation accuracy.

关 键 词：视觉特性 激光光斑图像分割 平均灰度熵值 对比度增强 光斑梯度 

分 类 号：TN911[电子电信—通信与信息系统]