Based On The Research Of Daisy Algorithm Of Stereo Matching

According to relevant statistics,human Obtaining information outside world mainly through vision to complete,this ratio is around 80%.This is enough to explain the visual for people to obtain information is very important.However,some of the production environment and the risk of human vision can not meet the production requirements of the occasion,such as industrial inspection,testing and other industries,it is often necessary recruited replacement-machine vision to resolve.Therefore vision technology has become a key research direction of the robot research and computer fields.Wherein the binocular stereo vision is a popular visual technology research direction.Binocular stereo vision by cameras located in two different locations of objects in three-dimensional space to shoot,in order to gain an object at different angles into two images.Then calculated using the geometric principle position deviation of two images corresponding image points,and then get to the three-dimensional information of the target object.This article is to study Daisy binocular vision stereo matching algorithm based on the main content of the work as follows:(1)Research on the binocular stereo vision system.Analysis of the camera to capture images of linear and nonlinear models,and then explores the binocular stereo vision imaging model,and explores determined by camera calibration method parameters in the model.Finally Gets analyzed depth information of stereoscopic vision.(2)The extraction of the corner feature images.To Moravec,Harris,FAST isocenter detection algorithm to achieve analyzed,and gives their comparative performance.Ability and image corner point extraction method is improved to enhance the resistance to noise extraction corners and it has a scale-invariant features.(3)It analyzes how the image of the extracted feature points will be described.The main characteristics of the Daisy descriptor algorithms are analyzed.(4)Research on the stereo matching how about the view.Proposed by improved corner feature extraction methods and Daisy algorithm,Euclidean distance similarity measure to achieve matching feature about the view point.Then use the matching point spread way to get around dense disparity map view.Finally,the work of this paper is summarized,and points out the stereo matching design proposed the existence of shortcomings,as pointed out the direction of future research.