Research On 3D Reconstruction And Target Detection Based On Binocular Vision

Under the current situation where the level of science and technology has significantly improved,the related disciplines such as electrical and electronics,physical optics,information science,and image processing continue to develop,expand,and improve,and important researches such as military,aerospace,medical,and agricultural machinery have been carried out for 3D reconstruction technology.The field has played an increasingly important role and has received people's attention and enthusiastic research boom.Every industry pays more attention to the degree of machine automation.By automating the machine to replace the manpower to achieve high-efficiency production and operation,and using high-precision threedimensional reconstruction standards to improve automation,the threedimensional reconstruction technology has become an urgent need to solve the current field of computer vision research.One of the important issues.The use of binocular cameras instead of the human eye to perform threedimensional reconstruction is a hot project in the field of computer vision research and has received the attention of researchers from various science departments.This article aims to establish a binocular vision system model,fully understand the research progress of binocular vision technology at home and abroad,and the camera calibration technology in the three-dimensional reconstruction process based on binocular vision,image preprocessing technology in binocular vision,Stereo matching algorithm optimization,three-dimensional reconstruction process and a series of key issues in-depth study.The main research contents of this article are as follows:(1)Introduce the camera model by introducing the principle of binocular imaging,and describe the linear model(hole imaging)and nonlinear model(radial distortion and tangential distortion)of the camera.Through the establishment of camera parameters,the world coordinate system,camera coordinate system,image physical coordinate system and image pixel coordinate system were established and the transformation relationship between them was described in detail.For the camera calibration algorithm used in this paper,a detailed analysis and formula derivation are made,and the error factors caused by radial distortion and tangential distortion are revised.(2)Mainly research the image preprocessing technology and feature extraction method in binocular vision.The otsu adaptive binarization segmentation is performed on the images captured by the left and right cameras that contain black water glasses.This image segmentation method is more adaptable to the environment than general image segmentation.The effect of adaptive image segmentation is improved by using image filtering and image enhancement methods.Harris point detection,Shi-Tomasi corner detection and SIFT algorithm were used to detect feature points in the black water cup.(3)Three types of stereo matching algorithms are analyzed: regionconstrained local matching algorithm,feature-based stereo matching algorithm,and global constraint-based optimized matching algorithm,and feature-based stereo matching algorithm and region-based stereo matching.The algorithm was compared.The four steps of the stereo matching algorithm are described: matching cost calculation,cost aggregation,computed parallax and parallax refinement.(4)This paper proposes an adaptive stereo matching algorithm which combines optimized region matching and feature point matching.It adopts otsu adaptive binarization method to perform binarization processing before feature point matching,and then passes edge detection.The mode box selects the smallest rectangular area of the outline of the black cup,so that the rectangular area is more suitable for image segmentation.This optimized stereo matching algorithm improves the detection efficiency of black water cup feature points,reduces the time required for stereo matching,and achieves a significant matching effect.At the same time,it also obtains a more dense matching point pair,greatly improving the three-dimensional The effect of reconstruction.