Reparability measurement of vision sensor in active stereo visual system

Abstract

The active visual inspection has made giant strides in development. Due to the complexity of field conditions, the binocular stereo vision sensor in active stereo visual inspection system is difficult to acquire the ideal image pair under the poor illumination environment, or cannot work normally because one of two cameras is damaged. To solve the practical problem, a reparability method of the vision sensor in active stereo visual inspection system is proposed. Based on the traditional measurement platform with a structured stripe projector and two industrial cameras, a reparability mathematical model is established, which merges the measurement mode of the stereo vision and the structured light vision to inspect feature points based on the reparability measurement model. Meanwhile, a mathematical model and calibration method of the structured light vision are developed based on the homography between the image and light plane. The method greatly improved the stability and reliability of measurement data. It is proven by experiment that the method is valid, which can accomplish reparability tasks commendably and achieve uninterrupted inspection of feature points in the actual industrial applications.