The Study, Based On The Combination Of Moments And Neural Network Image Recognition Technology

With the development of Artificial Neural Network(ANN) and imagine science, the technology of image recognition based on NN overcomes some shortcomings of conventional Pattern Recognition, so that the application of image recognition technology based on NN is becoming more widely and deeply than previous any period in many fields.The paper mainly includes three aspects.They are introduced as follows:Firstly, preprocess target images. The target images are sampled, grayed, filtered and segmented in this phase.Secondly, extract moment features from target images. The paper presents 5 Combined-moment invariants on the basis of Hu 7 moment invariants, then extracts the combined-moment invariants from binarization images by program.Lastly, implement recognize and classify among different target images. Establish the BP NN model, then combined-moment invariants which are inputted the ANN is used to recognize different target images. The recognition is completed by trained BP NN.This paper adopts VC, C++ language to write some algorithms which are used to carry out different operations.Experiment and simulation indicate that the system can achieve classification and recognition among different target images with high classification rate and short time.