COLOR-BASED DETECTION OF DEFECTS ON CHICKEN MEAT

Existing vision-based automatic inspection systems are mainly devoted to mechanic and electronic applications, their introduction into other fields being strongly limited by the need for operating in non-contro lled environments and by the lack of an accurate definition of the ins pection task. In this paper, an intelligent vision system aimed at the detection of defects on chicken meat before packing is presented. The detection of defects relies on the analysis of the chromatic content of chicken images. Possibly defective areas are first extracted by mea ns of morphological image reconstruction, and then classified accordin g to a predefined list of defects. Experimental results show the effec tiveness of the proposed approach, thus proving the feasibility of aut omatic inspection of alimentary products. (C) 1997 Elsevier Science B. V.