The Recognition And Detection Of Precision Injection Parts Based On Machine Vision

The research of target recognition and detection based on machine vision is the hotspot study in recent years, and machine vision is applied to industrialized recognition and detection, which can not only improve production speed and efficiency, but also reduce costs and promote the degree of automation in industrial production. On the basis of the foundation of machine vision technology, the improvement in the flexibility of the industrial robot is regarded as its aim, and according to the characteristics of injection parts and industrial robot, studied the system of precision injection parts of the recognition and detection is done. The main research contents are as follows:According to the function and the requirements of the system, the overall scheme is selected and the system is designed. And from that starting point, the design of identification and detection system, in which the system structure design, hardware configuration and selection, relevant software selection are included. In order to facilitate the follow-up work, the image processing method such as image segmentation, image de-noising, edge detection, and the image morphological was studied, on the basis of the concrete characteristics of image already obtained. These methods provide convenience for the subsequent treatment. According to the characteristics of the system, the thesis focuses on the visual system calibration method and experimental calculating process. With 2D target, the camera calibration with the internal and external parameters and hand-eye calibration of the robot vision system are finished with the help of HALCON software. Practice has proved this process is simple and precise. In the aspect of visual recognition, the extraction of some image characteristics and their realization ways are analyzed. According to their respective characteristics, the visual identification and classification are researched and realized based on the image characteristics and the neural network method. At the same time, the extraction of workpiece position is carried out on the basis of coordinate transformation, and the grabbing experiment is done. In the visual inspection, taking small module plastic gears as an example, carried on the measurement of gear surface defects, and achieved the expected testing purpose.On the basis of the above comprehensive process, the modularization design is made in the visual recognition and the measuring system with the help of Visual C++ 2008 and HALCON software. There for, a complete visual system with the function of identification and testing is realized, which lays a good foundation to improve the efficiency and automation of the precise injection production line.