Surface targets recognition method based on LVQ neutral network

These A method to identify the different surface targets with combination features was proposed on the conditions of pretreatment that the video image sequence was preprocessed by removing noise and image stabilization. Firstly, Targets and background were separated by segmenting the clearer images. Secondly, the geometrical feature and the moment invariant feature in different targets were extracted. The LVQ (Learning Vector Quantization) neutral network was trained to identify surface targets by using combination features. Finally, the simulation study of identifying test targets was done. The results of simulation research show that the proposed method based on combination features of different surface targets can recognizes the three types of common surface targets effectively. And, the convergence speed of LVQ neural network is fast compared with the BP neural network and the recognition has a good effect.