Aerial extended target detection based on unit-linking pulse coupled neural networks

A novel method of aerial extended target detection based on pulse coupled neural network (PCNN) is presented. Some key techniques are improved and network model is simplified, which are more suitable for image processing, such as linear modulation, linear attenuation of dynamic threshold and step selection, and iteration time control. Meanwhile, maximum line contour point number and color negative are adopted to determine the optimal iteration times and the optimal segmented results for targets with low contrast and continuously varying background. The experimental results show that the proposed method can maintain perfect and distinct targets contour and detect aerial extended targets against complex background.