Research On Calibration Of Structure Light Projector Based On OPENCV

As a member of the non-contact in computer vision,structured light vision measurement is becoming more and more important in machine vision. Calibration of the optical transmitter is the first step and the most important step in three-dimensional structured light vision measurement; it can directly affect the accuracy of the entire system. The working principle of optical transmitter is that some images are projected by Core chip onto the place where we want to see, we think it’s contrary to the principle of the camera. In order to ensure the robustness and reliability of measurement, calibration of the optical transmitter is significant.Existing calibration techniques are mainly divided into two categories, one is the phase analysis method, and another is the perspective projection method. Based on the research of binocular vision, we do an in- depth research on Single projector- single camera system using the perspective projection method. The main researches as follows:(1) In this paper, we establish a single projector- single camera mathematical model standing in the perspective of the human eye, building the camera and the projector model into a binocular vision model and improve the model. We establish the thinking of a virtual projection plane, which can only rely on the projector, a camera and image processing and analysis equipment to complete the calibration. Finally, our target is reconstructing the virtual projection plane.(2) In this paper, we have difficulty in matching key points of image and image matching. Because of the existence of location transformation, angular transformation and size conversion between images we need to match, so we introduce the concept of scale space into image matching to complete the image matching. Gauss Blur Images and detect the key points of images by SIFT algorithm, and then improve matching accuracy by using the optimization algorithm proposed in this paper. Experiments show that through the optimization we solve the problem on traditional image matching.(3) Improvement suggestions are given to solve the problems in camera calibration method based on planer target, extract feature points on the plane on the premise of camera without distortion. Assume the center of the image coordinates is the optical center coordinates of camera, and then we get the solutions of matrix by using the linear model. After considering the nonlinear distortion model parameters, improve the accuracy of the parameters With LM algorithm.