Automatic Stake Test System Based On Machine Vision

In order to improve equity and veracity of automobile's stake test, this paper put forward a thought and means of automatic stake test, based on machine vision in which a number of digit image disposal technology and pattern identification theory are used.The paper recounts implementing thoughts of this system and some digit image disposal methods for this system. Point is how to raise accuracy and real time of this system. In pretreating image phase, in order to get a better threshold, I advance a improved algorithm based on luminance compensating, and get rather satisfying effect. On the aspect of identifying image, I use a structure of mosaic. The experiment proves this special structure can not be easily disturbed and can be quickly identified. This can satisfy the real time demand of the automobile's stake test system. In the phase of checking automobile's edge, I use a few of spots instead of all the edge spots which can reduce computer time consumedly. Finally, this article put forward the conceive of multicamera cooperation mechanism. In Windows environment the whole system runs in gear, which implements real time operation of the stake test.