Image Recognition Of Moment And Neural Network Based On Wavelet Transform

As we know, image recognition is an important branch of pattern recognition. Through few decades, it has been applied successfully in the military, space exploration, medical science and post, etc. So it has greatly importance and practical value.In this paper we look into the application of moment function and neural network in feature extraction and pattern classification of image recognition respectively.We use moment algorithm to extract the invariant features of image, and discuss zernike moments, wavelet moments, Hu moments and improve algorithm of Hu moments in detail. Because the other algorithms based on Hu moments can only be used in different circumstances, this is the disadvantage in application. Experiments show that the relative moments can substitute the other algorithms based on Hu moments under some situations. Because the wavelet moments that Shen proposed.cannot be applied in computer, we put forth a discrete method of wavelet moments. Experiments show that the discrete wavelet moments are superior to the traditional moments.Because of the disadvantage of K-means and BP algorithm, we use the wavelet neural network in system of image recognition. We depict the algorithm and design of wavelet neural network. Experiments show that the wavelet neural network is better than the traditional BP algorithm.