Research On Fast Depth Segmentation Technology Based On Phase Singular Features

Depth segmentation technology has been widely concerned with and studied in the fields of computer vision and image processing.The purpose of depth segmentation is to divide targets at different depths into independent regions with continuous depth,so as to conduct in-depth analysis and understanding of three-dimensional(3D)data.This technology plays an important role in the fields of robot grabbing,face recognition,and camouflage detection.However,with the rapid development of computer technology,depth segmentation technology presents the characteristics of complex algorithm framework,high computational consumption,and high cost,which to some extent limits the development and application of traditional depth segmentation technology in various areas.The fringe projection technology is a widely used three-dimensional profile measurement technology,which has the properties of non-contact,high resolution,and high robustness.This paper takes the phase singular feature as the bridging factor,innovatively merges the two technologies,and proposes a fast depth segmentation technology based on the phase singular feature,trying to study some solutions to improve the speed of the existing deep segmentation technology.Meanwhile,its related applications are explored based on the proposed depth segmentation technology.The main content of this paper is reflected in four aspects:(i)The mathematical relationship between the phase singular feature and the object singular feature is deduced,and it is proved that the phase singular feature can be used as a sign for depth segmentation.By analyzing the key factors affecting the performance of depth segmentation,the research direction of this article is clarified,and a fast depth segmentation framework and a depth segmentation line optimization algorithm are designed.(ii)Based on the idea of low amplitude fringe measurement,this paper proposes two fast depth segmentation techniques based on orthogonal fringe patterns(OFP).These two technologies only need five fringe patterns for depth segmentation without reconstruction and pattern sequence changing,and achieve effective segmentation of dynamic objects.(iii)On the basis of orthogonal fringe patterns segmentation,this paper proposes a fast depth segmentation technology based on composite orthogonal fringe patterns(COFP).This technology only needs three stripes to achieve depth segmentation;With cross-stripe segmentation technology,the segmentation speed of this technology is increased to 166.7%.In addition,robustness experiments prove that this technology can still achieve segmentation under dark light conditions.(iv)Based on the core idea of fast depth segmentation,to explore the application of fast depth segmentation,this paper proposes a camouflage object detection technology.The camouflage detection technology(CDT)uses OFP for effective segmentation of camouflage objects in invading complex scenes,in which the location,texture,and shadow parameters of the camouflage objects can be determined.