Projection Optics Surface Shape Detection Algorithm Optimization

In the detection methods of the optical element surface characterization, applying the projection method to the stereo vision detection technology is a hotspot in current research. The image processing plays an important role in the process of the projection detection methods and is also an important factor which affects the accuracy and speed of the detection. The paper bases on the theory of the projection-based stereo detection technology, through the research on the related image processing algorithms of the lens surface characterization detecting. The paper mainly complete the following tasks:Firstly, through referring to correlated papers, the optical element surface characterization detecting principle and several methods are understood. The paper analyzes the evaluation methods and indicators of the optical element surface characterization and determined the surface characterization evaluation criteria of the detecting.Secondly, based on the detecting principle of structured light binocular vision inspection systems, the paper analyzes the mathematical model of the imaging system, and establishes an algorithm flow of the digital image processing.Thirdly, the parallax calculation and the model analyses methods of the image processing are studying by analyzing and comparing different feature matching algorithms. The paper combines with the image matching characteristics and algorithm design requirements, and then selected the SURF matching algorithm, and proposed an optimization scheme focusing on its shortcomings.Fourthly, through the research of the SURF algorithm and combines with the image detection characteristics in this paper, in consideration of shortcomings of the extraction and matching feature points for the SURF algorithm, the SURF algorithm is optimized according to the optimization scheme. Then the optimized algorithm are simulates with MATLAB verified the feasibility of the algorithm theory.Fifthly, according to the experiment principles and scheme, the paper establishes the experimental platform and collected images from experiment. The experiment is processed by using MATLAB and VC++software, and then completes the camera calibration, image matching and3D reconstruction. According to the surface characterization evaluation criteria, analysis of the results of the optimized matching algorithm, and compared with the matching results of the SURF algorithm, the experiment shows that the optimized algorithm has an great feasibility in the optical element surface characterization detecting.Sixthly, summarized the papers and discussed this subject. The paper points out the parts of the improved algorithm still needs to improve and perfect, and the direction needed to further study.