Hopfield neural network image matching based on Hausdorff distance and chaos optimizing

Due to its capability of high-speed information processing and uncertainty information processing, Feature point based Hopfield Neural Network image matching method has attracted considerable attention in recent years. However, there often exists much difference between two images, especially under the influences of distortion factors, thus the result of image matching is affected greatly. In addition, Hopfield Neural Network is often trapped in local minima, which gives an optimization solution with an unacceptable high cost. To overcome the defects mentioned above, in this paper, Hausdorff distance is used to measure the degree of the similarity of two images. Chaos is used to optimize the search process of Hopfield Neural Network, and a new energy formulation for general invariant matching is derived. Experimental results demonstrate the efficiency and the effectiveness of the proposed method.