Study On The Quantification Of Micro Stereovision With A SLM And 3D Data Reconstruction

Micro stereovision based on stereo light microscope (SLM) is used in some micro-domains such as micromanipulation, microassembly, etc al. There are mainly two purposes of micro stereovision: the first is that it is used to auto position in micromanipulation and microassembly by the vision feedback and helps the mechanical hand to complete the expected operation; the second is that it contains the 3D depth information which is used in the measurement of 3D attributes. In this paper, some important issues correspondent with micro stereovision based on SLM are studied for solving problems of micro positioning and micro measurement, such as the mathematical description of the imaging process of SLM, the calibration of stereoscopic vision model, the correction of image distortion, the stereo matching algorithm of color images with high precision, the 3D measurements based on images, etc al . Results derived from this paper can be used to 3D measurements of objects with a small scale and 3D positioning in micromanipulation, and microassembly.First, object recognition and edge extraction in micro color images are studied, and the algorithms of object recognition and edge extraction are developed for micro color images based on improving the old algorithms. A new method that proceeds firstly by the objects recognition, and secondly edge extraction of objects in color images is applied to reduce the time of computation and improve the accuracy of edge extraction process. Results will server the need of test, parameter calibration and data reconstruction.Second, the modeling problem of micro stereovision with SLM is discussed, and a vision model that describes the projection between 2D image space and 3D object space is derived. A pair of optical paths of SLM is denoted by a group of basic points and basic planes, and then two models of micro stereovision with SLM are derived via the geometrical projection of virtual incident rays and the weak nonlinear phenomenon. The symmetry of a pair of optical paths of SLM is considered in the first model, and the weak difference is considered in the second model. The positioning experiments with a calibration standard as a test sample show that the weak disparity model of stereovision has a reconstruction precision 5% in parallel direction and 10% in vertical direction.Third, the parameter calibration of the weak disparity model is studied, and then a new calibration method is developed to resolve the calibration with approximate conditions. The parameters describing the stereoscopic model are divided into two types: the main parameters and the adjustive parameters, and these two parameters are calibrated respectively. The symmetry and weak difference between the two optical paths of SLM are considered for driveling the constraint conditions of parameter calibration. The association between the parameters and additive noises, the robustness of parameter estimates, parameter correlations and repeatability are discussed by experiments.Fourth, the distortions in micro stereovision system are analyzed, and three kinds ofdistortions, the rotation of CCD image planes, the depth difference in object space of SLM, the distortions of an optical system, are considered mainly. And the effects of distortion parameters are tested by the significance test and correspondence test methods.Fifth, the stereo matching problem of color images is discussed. An algorithm based on multi similarity measurements and multi-matching areas is proposed for stereo matching of color images. In this algorithm, the samples of matching output are acquired by the statistics of multi similarity measurements and multi-matching areas. On the other hand, the false matching results and the wrong matching results are partially filtered by performing the mean-square error statistics of matching output samples. Further more, a sub-pixel matching accuracy is expected by calculating the average values of two new samples derived from the mean-square error statistics. The experiment results show that the stability and a...