A robust bread defect detection and counting system

In this paper, an automated visual inspection method is presented to detect defect and count breads for the use in bread production. Cost reduction in production and inspection processes is of big concern to bread manufacturers. Inspecting bread by human eyes must be a tedious job for human inspectors, which might result in high inspection error. Hence, automated inspection systems are greatly required. In this work, the bread image is binarized firstly. If several breads are adjacent to each other, those are separated using K-cosine corner detection algorithm. Next, defect detection is conducted by taking object´s area and Hu´s invariant moments. Experimental results indicate that the proposed method can be efficiently used in the bread inspection system.