Detection Algorithm of Aluminum Surface Defects Using Machine Vision

With the growth of global economic and the widespread use of aluminum profile, the consumption of the global aluminum profile increases year by year. In this paper, we proposed a novel detection algorithm of aluminum surface defects using machine vision. Firstly, the aluminum images are acquired and analyzed sequentially, then a number of image processing strategies were used to detect various surface defects. The main contribution is a new area partition method, which can automatically assign texture and no-texture regions with texture information. The proposed method is proven able to detect defects on aluminum profile surfaces, such as cracks, pits, rust or scratches, rapidly and precisely. Robustness and effectiveness in the practical aluminum casting process are improved by using the proposed system.