The3D Reconstruction Methods Research Based On Binocular Stereo Vision

The traditional operation is easy to cause medical accidents due to rely onmanual operation. There are a few problems such as inadequate preoperative planning,high pressure of surgeon and patients don’t understand the operation process. Thesurgical robots control the scalpel with mechanical arm. The surgeon can intervenethrough relevant system which could show the location and shape of nidus andachieve3D visualization. The3D reconstruction is the core technology of surgicalrobot. Considering the high surgery accuracy of skull, this paper study on the3Dreconstruction of the skull part.A3D reconstruction system has been designed in this article which used codedstructure-light to achieve non-contact measurement and used two cameras to imitatethe visual perception of human eyes by capture photos. A m-array based experimentmethod has been designed to ensure the accuracy and real time.A adaptive dynamicwindow has been designed to locate the head of the picture and provide gist for imagecrop. This article study on the retinex theory based image enhancement to change thebrightness. The paper proposed a clustering image enhancement method based oneuclidean distance to convenient for encoding. The paper uses k-means clusteringsegmentation to separate points array and background. The article study on centroidsextracting method and extracted the centroids of points array as featured points,discuss relevant problems of coding and ifnally achieved feature matching.The articleestablished a camera imaging model which uses parallax measuring principle.Thispaper study on DLT method to ifnished camera calibration. This paper use the Lawsonalgorithm of delaunay triangulation subdivision to connect spatial points cloud andthen achieved3D reconstruction. In the part of experiment, the paper gave all experimental images, data,reconstruction effect and relevant analysis. Experiment show that the method of thispaper is very good.