Contour recognition based on associative memory

In contemporary digital world, image digitization benefits the pattern recognition based on the computational methods. By the nature of associative memory, which is a practical usage of discrete Hopfield neural network, it can be adopted for the pattern recognition application. This paper proposes the design and the prototype implementation of pattern recognition through associative memory. Our approach is semantics oriented through figure contour extracted by edge detection. Experimental results show that our approach can recover the polluted or fragmentary image on a high confidence level.