Crankshaft Omnidirection Image Information Collection And There-dimensional Reconstruction

The central hole of the workpiece shaft plays a role of positioning and datum to process procedures.So the positioning of the central hole impacts on product quality directly.Particularly for the crankshaft and the shaft with defective surface,the position of central hole is more important.It relates whether the workpiece has an appropriate allowance.In the mass production,it has more economic signality to deal with this issue.The project team proposed a new method of positioning and drilling central hole in crankshaft and shaft.Three-dimensional model of actual crankshaft,which is constructed in computer based on machine vision technology,and three-dimensional model which is designed by three-dimensional CAD match,in order to access to the best positioning point,and then drill here.Using this method to position and drill can reduce the working time and improve productivity and processing quality.And this method is more flexible,which is applied to a variety of shaft and other parts of the processing positioning.The reconstruction of actual crankshaft three- dimensional model in computer is the one of key technologies to achieve the above central hole positioning and drilling method.The paper will study crankshaft all-round image information collection based on pinhole imaging model and proposal and arithmetic three -dimensional reconstructed in computer based on image information collection,in order to lay the foundation for the further study.This paper,by comparing a variety of measurement method,uses binocular stereo vision measurement method,and binocular vision system based on PC-control of three pairs of camera,to collect image information from three different directions, and on the basis of accuracy and experiment environment determine the hardware, collecting the full range of crankshaft image information.The template box used as the calibration board respectively calibrates left and right camera.And from a variety of methods Tasi calibration method is selected,in order to get the internal and external camera parameters. The image getting from camera is processed by VC++,including the gray-scale processing,smoothing,image enhancement,threshold segmentation,candy edge detection operator,and then point extraction by improved Harris.Based upon camera parameters calibration and feature point matching of left and fight image,stereo matching t,calculating the depth information of feature point collection,gets the spatial coordinates of characteristic point.And the necessary interpolation calculated,texture mapping the three-dimensional reconstruction model makes the model realistic.