Pattern recognition based on heuristic neural network

A heuristic neural network (HNN) algorithm is proposed for pattern recognition. To achieve faster convergence and to promote the recognition accuracy, the proposed neural network is trained by adjusting the weighting values between layers, bias values of neurons and the learning rate according to the classification error. The whole architecture includes the image capturing by CCD camera, image pre-processing, neural network constructing and the developed heuristic algorithm. To verify the effectiveness and performance of the proposed method, a total of 36 patterns are collected as samples, 0 to 9 and A to Z respectively. From the experimental results, the proposed method has faster convergence and better recognition accuracy compared with the conventional back-propagation momentum neural network.