A novel edge-parameter analysis approach of the blur identification based on the single threshold PCNN

A novel edge-parameter analysis method of the blur' identification based on Pulse Coupled Neural Networks (PCNN) model is proposed for image de-blurring application, which is suitable for the identification of the horizontal linear motion blur. This new identification method uses the improved single-threshold PCNN model and the normalized local entropy transformation to form a novel edge factor and extract edge and texture information from the blurred image. From the edge-parameter curve caused by the edge factor and the uniform linear motion parameter, the displacement parameter can be recognized accurately. The experimental results obtained from the different images and the same image with the different resolution show that the new algorithm is very effective and the curve performs quite stably. The identification displacement can ranges from 4 to 30 pixels.