Research On Three Dimensional Reconstruction Of Binocular Vision Combined With Structure Light

Computer vision technology is a warm research topic in modern times, three dimensional reconstruction is a significant branch of this technology. It concerns some image processing and vision processing knowledge and is widely applied on industrial automation、reverse engineering,preservation of culture relics,virtual reality,augumented reality and robotic application. Low cost, high precision, fast acquiring, flexibility are required for this technology, some focus and difficulty lie on this.This thesis design a three dimensional reconstruction system combined stereo vision and structure light technology. We deeply analysis all the principle and algorithm within these two technologies. This thesis first show the principle of camera imaging、binocular and monocular system’s calibration theory experiments、principle and actualization of stereo rectification;then compare and analyse block matching、semi-global blocking matching、global matching algorithm, give a novel dense disparity estimation algorithm that using superpixels;introduce normal structure light coding method,design a gray-code with phase moving coding technology.Based on these principle and theory, we invent some algorithms and construct stereo vison and projector system. Using the hardware, finish the design,coding,experiment of stereo images obtaining,stereo calibration,stereo matching,stereo rectification,region extraction,image decoding,disparity generation and optimization,point cloud generation and optimization. Some experiments are taken to prove the flexibility and efficiency of our system.