Vision system for Command and Fault detection

This article concerns the development of algorithms treating the problems of luminosity in the industrial vision. Also this paper participates to have another look to the industrial equipments by exchanging the classic sensors by the vision system. The fault detection in the industrial environment has real problem which is the luminosity, and that depends on the location of the system. In fact, the vision systems are very sensitive to the luminosity variation and the dazzling event. To solve this problem, the vision systems are black and white (named also gray scale image) and have a complex system of lighting. The solution advanced in this paper is the treatment of coloured image (YCbCr) instead of gray scale image. By basing the treatment on the coloured image, the system will be less sensitive and very precise. The algorithms will eliminate the luminosity effect and continue to localize and identify faults. The profit of this method is to calculate in several plans (Cb and Cr) the information needed to command and to the fault detection algorithms