基于超像素和图论相结合的激光图像分割方法

Laser image segmentation method based on the combination of superpixel and graph theory

为了提升激光图像分割精度,研究基于超像素和图论相结合的激光图像分割方法。首先增强处理激光图像,并对增强后激光图像进行超像素块分割,提取全部超像素块;然后以超像素信息为基础,以图论绘制激光图像的无向带权图,设计分割判断函数,利用该函数判断无向带权图中近邻超像素块的合并与分割情况,最后把分割后无向带权图映射至原图,实现激光图像分割。实验结果表明:本方法边界召回率高达0.93,欠分割错误率仅有0.09、可达分割准确率高达0.95,可以有效提升激光图像分割精度。

In order to improve the accuracy of laser image segmentation,a laser image segmentation method based on superpixel and graph theory is studied.Firstly,the laser image is enhanced,and the superpixel blocks are segmented to extract all the superpixel blocks;Then,based on the superpixel information,the undirected weighted graph of laser image is drawn by graph theory,and the segmentation judgment function is designed to judge the merging and segmentation of adjacent super-pixel blocks in the undirected weighted graph.Finally,the segmented undirected weighted graph is mapped to the original image to realize the excitation image segmentation.The experimental results show that the boundary recall rate of this method is as high as 0.93,the under segmentation error rate is only 0.09,and the segmentation accuracy is as high as 0.95,which can effectively improve the segmentation accuracy of laser image.