Development On Semiconductor Automatic Sorting System Based On Machine Vision Technology

Automatic sorting with large quantities of mixed and used semiconductor components plays a significant role in follow-up regeneration dispose, mainly through manual visual inspection at present, low efficiency and less precision. Based on this, using the semiconductor triode as the main research object, this paper constructs a machine vision system for automatic sorting of semiconductor element, and has a systematic research on the automatic sorting method of the used semiconductor element, and mainly launches the research from two aspects of image processing and character recognition, the ultimate realization of real-time automatic sorting of semiconductor electronic components based on machine vision technology.First of all, the whole design of the system is analysed, and the hardware system of real-time automatic sorting for waste semiconductor component is built. The electromagnetic vibration technology is adopted to realize automatic feeding of the components, and a detailed design with the device of mechanical positioning of the components is carried on. To realize the automatic sorting of real-time online, the economic application of image acquisition equipment and image optical lighting system is selected, and the material sorting device is designed.Secondly, this article introduces the automatic control system and the working principle of the sorting system, focus on the processing difficulty in the defect in element surface image, and put forward some effective methods for image preprocessing, image segmentation, locating and character correction, character segmentation and recognition. By using morphological image processing technology, the process of image processing with a example is descripted.Finally, the system control software is programmed for Visual c + + 6.0 development environment with MVTec HALCON 8.0 professional image processing software, a design a friendly human-computer interaction interface of the system. Use of machine vision processing algorithm to implement the identification of a semiconductor element classification, has a good progressiveness. According to the test, the technical specifications of system is : sorted rate is 1-2 per second, recognition rate is more than 70%,misclassification rate is less than 2%, meeting the precision requirement of enterprise production, with a good practicality.