Binocular Stereo Vision And Research Of Pore Vision Measurement System

With the advantages of fast measuring speed, high accuracy, non-contact, easy operation and automatic measuring, etc, the binocular stereo vision inspection technique play a significant role in many area as a new measuring method. Taking the application of the binocular stereo vision technique in industrial inspection as study object, we discussed some key problems about 3D circle fitting. The major content is as follows.1,The fundamental principles, components and main functions of the binocular stereo vision inspection system are elucidated in details. The main algorithms of the binocular vision inspection are introduced.2,The mathematical model of three-dimensional (3D) measurement for the binocular vision sensor is discussed. The effects of the baseline distance (the distance between projection center of left camera and right camera) of the binocular vision sensor on the volume of sensor structure and measuring accuracy are analyzed in details. The structure configuration of the binocular vision sensor is also introduced.3,The real camera projection model is analyzed in details. Based on classic camera calibration methods, the camera parameters calibration method to be suited for vision inspection with non-coplanar calibration reference object is analyzed and one calibration method to determine the structure parameters of the binocular vision sensor with a rigid distance is proposed. The initial values of the camera parameters and the structure parameters can be estimated automatically in the calibrating process. The proposed calibration method can attain high calibration accuracy and increase the calibrating efficiency, so it can be applied to the on-line calibrating for the parameters of the binocular vision sensor in the real industrial environments.4,The feature point forms of the reference object for camera calibration are introduced. One automatic feature detection method with sub-pixel accuracy for the image coordinates of the center of the circle and the corner point of the chessboard is proposed.5,Based on the epipolar constraint, one correspondence point matching approach which uses the feature point with sub-pixel accuracy and the cross correlation is analyzed in details.6,The perspective projection model of the circle in the 3D space is analyzed in details. When the plane of the circle is not parallel to the image plane, the projection of the space circle should be an ellipse and the projection of the circle center in the 3D space is not to equal to the center of the center of the projection ellipse.7,Two feasible vision inspection solutions to measure the dimensional of the circle and the 3D coordinates of the circle center is proposed. Firstly, the coordinates of the circle edge in the 3D space are measured by the binocular stereo vision inspection, then the dimensional parameters of the circle in 3D space can be got based on 3D circle fitting with those the coordinates data of the circle edge. The proposed approach can eliminate the perspective projection shape distorted of the circle features in the image plane, so it improves the measuring accuracy of the circle features in the image plane. At the same time, another method of measuring 3D circle based quadric curves is proposed. So both of the methods improve the measuring accuracy of the circle with the binocular stereo vision inspection method.