| As an important industrial electric component, contact component is broadly used in various fields. The performance of contact component is critical in deciding electric equipments'lifespan since it takes the key task of switching, for which higher standard is required for the measurement of contact component's quality. For the moment it is mainly checked by hand in domestic product line, which is quite ineffective. As a result, an automatic detecting system based on machine vision would be very practical. This paper presents a machine-vision based field detecting system focusing on contact component's various shapes and flaws.The main purpose of the system is to: 1.Detect and categorize the geometrical and texture flaws.2.Detect both sides of the contact component in one process. An integrated design was made upon analyzing the detecting requirements, as well as the hardware and the software part.The innovation points for the hardware part are: 1.A special transfer and overturn system was designed to turn over the components in the product line so that detection for both sides could be possible. 2. A special filtering system was designed to categorize the different flaws. What's more, further research on the parts operating in the system was done to better know their function and choosing criteria.According with the real-time and accuracy requirements of the system, this paper presents an adaptive threshold segmentation algorithm based on the improved local minimum point. By effective measurement of both the texture and geometrical features of the flaws, a criterion feature was generated. Finally, detection and categorization of the contact components'various flaws was achieved.The software part for the implementation of the above algorithms is included in this paper. Some tests made in the laboratory have seen fairly good results. At the end of this paper was some analysis for the factors relevant with the testing results, as well as directions for further improvement. |
PDF Full Text Request