Evolutionary learning of a fuzzy edge detection algorithm based on multiple masks

In this paper we demonstrate how an evolutionary algorithm (EA) can be employed to learn a fuzzy knowledge base (FKB) utilised in an edge detection algorithm based on the fuzzy paradigm. The proposed method calculates a fuzzy measure 'edginess' at each pixel of the image using masks of different sizes. Then the edge strengths calculated using these masks are used to form a fuzzy knowledge base which in turn is used to decide whether a given pixel belongs to an edge or not. When calculating the above mentioned fuzzy measures, the algorithm takes into account both step like edges and 'line edges' in the image being processed. The final edge map of a given image is produced by generating output pixel values 'non-linearly' proportional to the above mentioned fuzzy measure 'edginess'. The results are presented for several real and synthetic images to show the effectiveness of the proposed technique.