Research And Realize On Image Recognition Technology For Multi Targets On-the-Fly System

Placement machine is the symbol of the modern electronic manufacturing equipment. It crosses many disciplines, such as the electronics machinery, automation, optics and computer. And it has involved many core technologies, such as the precise visual inspection, high speed and high precision control, precision machining and computer integrated manufacturing. Placement machine is a typical mechatronics and optics integration field. The speed and accuracy are the keys to measure placement machine performance, which are decided by the placement machine’s visual system.This dissertation focuses on the research and implementation of the vision system, which is based on OpenCV. The research platform is placement machine specially used for LED, the vision system need to meet the characteristics and requirements of its structure. The main research contents are illustrated as below.Firstly, the general scheme of the vision system which meets the on-the-fly visual and multi targets recognition are proposed and the function of the key modules is analyzed. Meanwhile, the system performance indices are put forward on the much analysis.Secondly, the image acquisition device is designed and constructed to identify MARK points and detecting electronic components. It will introduce how to design and select the LED lighting system and optical imaging system. The LED lighting source can supply the most appropriate lighting for the electronic components. So In conjunction with the light source, the optical imaging system can capture high quality images in the motion state.Thirdly, some fundamental algorithms based on OpenCV of image processing are studied. Such as image smoothing filtering, image threshold segmentation, image morphology processing and edge detection, sub-pixel edge detection. They are the foundation of electronic components and MARK points positioning and detection.Fourthly, LED components are positioned. The geometric characteristics of LED components are introduced and the algorithm for detecting and locating the LED components is studied. Experiments show that the algorithm locates precisely, satisfies the requirement of the LED placement machine. And the method of multi targets recognition is good for improving speed.At last, MARK points are sub-pixel positioned. The using principle of MARK points is introduced and the algorithm for detecting and sub-pixel locating the PCB circular MARK points is studied. Through the experiments, the algorithm locates precisely and has strong antijamming capability for the noise. And the algorithm satisfies the requirement of the LED placement machine.