基于CS算法的结构光图像融合技术研究  

Research on the structured light image fusion algorithm based on CS algorithm

作　　者：任建[1] 丁旭 王福强 刘斌[1] 张圣才 REN Jian;DING Xu;WANG Fuqiang;LIU Bin;ZHANG Shengcai(School of Information Science and Engineering,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]沈阳工业大学信息科学与工程学院,沈阳110870

出　　处：《黑龙江大学自然科学学报》2020年第6期735-742,共8页Journal of Natural Science of Heilongjiang University

基　　金：国家自然科学基金资助项目(6187011861);辽宁省教育厅青年育苗项目(LQGD202009)。

摘　　要：结构光图像广泛应用于工业自动化机器人的机器视觉。由于光学传感器的性能限制,测量精度与检测范围存在矛盾,光线的对焦不同、存在遮挡物会导致成像困难。本文运用基于压缩感知理论(Compressed sensing,CS)的多源图像融合技术解决了数据量大和工件成像难的问题。构建了最优小波稀疏基和最优小波融合基,对比了基于正交匹配追踪(Orthogonal matching pursuit,OMP)和经典的梯度投影(Gradient projection for sparse reconstruction,GPSR)的双小波变换压缩感知算法,成功进行超稀疏的正交匹配追踪工件图像融合与准确定位。实验表明,本文算法在结构光视觉成像时,可去除遮挡物并避免传统融合的信息损失和计算繁琐现象,提高了目标图像的空间分辨率和清晰度。这有利于机器视觉对工件位置、形状和类型的判断,成功进行精准定位放置等自动化机械操作。The structured light image has been widely used in machine vision applications of industrial automation robots.Due to the performance limitation of optical sensor,there is a contradiction between the measurement accuracy and the detection range.Different focusing of light and the presence of occlusion will lead to difficulties in imaging.The multi-source image fusion technique based on Compressed sensing(CS)is applied to solve the problems of large amount of data and for workpiece imaging.The optimal wavelet sparse basis and the optimal wavelet fusion basis are constructed,and the dual wavelet transform compressive sensing algorithm is compared,based on Orthogonal matching pursuit(OMP)and classical gradient projection for sparse reconstruction(GPSR).The ultra-sparse orthogonal matching pursuit workpiece image fusion and accurate positioning are successfully carried out.Experimental results show that the proposed algorithm can remove occlusions and avoid the information loss as well as computational complexity of traditional fusion in the structured light visual imaging,so as to improve the spatial resolution and clarity of the target image.It is helpful for machine vision to judge the position,shape and type of workpiece,and successfully carry out automatic mechanical operation such as precise positioning and placement.

关 键 词：结构光 双目视觉 图像融合 双小波变换 压缩感知 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]