The Research Of Vision System Basic Method In Identifying For Apple Picking Robot

In recent years, picking robot has become a focus of research in many universities,and the core problem of research picking robot is in the vision system. When a picking robot is working, the binocular lens distortion and the multiple overlapping of ripe fruit will affect the volatility of the visual system collecting data, result in the system error bigger, influence the efficiency of picking indirectly. In order to searching methods to solve the problems, this paper expands basic technology research to identify the robot to pick.First of all, for multiple overlapping ripe fruit extracting, this paper proposes an algorithm adapting to a set of images of the process: firstly, with the theory of statistical analysis of the fruits and leaves of color component distribution, the target fruit is extracted. Secondly, segmentation based on Watershed Algorithm is used to get the foreground target. The last, fruit contour and the apple center coordinate is extracted by using the least squares fitting circle.Secondly, to the visual system of lens distortion problem, In the dual target timing,this article add the distortion model to calibration toolbox and correct it to improve the accuracy of the binocular calibration.At last, because of the high error of Z coordinate of the target the fruit, this paper carried out the analysis of the influence factors of error and error compensation in order to get higher precision of spatial coordinates.Based on the research of the binocular vision system, creating a strong adaptability and high stability of visual identification system maybe can improve the accuracy and working efficiency of picking robot.