Research On A Vision Measurement System Of Micro Platform Three-dimensional Motion Error

With the continuous progress of human society and the manufacturing industry, People have increasingly high precision with Micro-mechanical manufacturing and testing. However, due to micro-mechanical small size and poor of load capacity commonly used in contact detection tools difficult to achieve the movement of micro-mechanical measurements and error detection. In this paper we use digital cameras for micro-mechanical two-dimensional image to restore their three-dimensional infor- mation. This method is the reverse process which is the use of cameras to obtain three-dimensional objects in two-dimensional image, and it is the bionic eye research. This is the three-dimensional reconstruction of objects.This article makes small size of the processing of all objects into a tiny rectangle abstract platform. And we called micro platform, then carries on the visual survey to this micro platform the research. This study is based on Marr's idea of theoretical models based on computer vision, and then study the entire process of the three-dimensional reconstruction. These are the focus of this study. Including image pre- processing,edge detection,feature extraction (match),three-dimensional recon- struction and so on.The main purpose of image preprocessing is to eliminate the influence of external factors, in this section we use of grayscale and binary to processing the original image,then we use a improve methods of Sobel operator-Eight template algorithm to determine the points on the edge of the high accuracy to get the edge;Image feature extraction is to obtain accurate feature points, It is also the basis of three-dimensional reconstruction model, This paper selects an improved Harris algorithm to obtain the feature point, then we get the reliability of the improved Harris algorithm by comparison it with Traditional Harris algorithms; In the reconstruction process, we introduced the two images without prior knowledge of the reconstruction and with prior knowledge of image reconstruction: Reconstruction process in the two images feature points in the first calculation of the extracted feature points of two images in gray scale correlation between the NCC, the initial matching by fuzzy matching set, and then use relaxation to get the angle of the previous fuzzy match points into the elimination of one to one match, the last application of the minimum value method to eliminate errors in the match; In one image reconstruction, as with prior knowledge so it can get the world coordinates of each point.In the last we use of Visual C++ and Matlab as a tool to achieve a vision measurement system. Experimental results show that the measured data with very small error, and the expected results achieved. For further research provides a solid foundation in theory and practice...