Research On Classification Of Image Of The Web Based On Neural Network

With the development of Internet and digital technology, the information on the Internet has been increasing rapidly. The Internet has become a huge storehouse of the multi-media documents such as texts, pictures and audios. A lot of attention has been paid to the employment of web pictures, so it has also become an urgent problem. On one hand, it can enable internet user to locate the information needed. On the other hand, it can filter the picture noise. However, web pictures vary in terms of themes, patterns, functions and sources. It is hard to classify them with standard criteria for effective retrieval. Therefore, the classification of web pictures is restricted to certain areas, which in turn affects the effective application of web pictures. Aim of this research is to build one classification Model of high efficiency, which will help to classify web Images. This research has applied the algorithm of Artificial Neural Network (ANN), which is one classification method of Data Mining. At the same time, the research applies the flexible BP arithmetic to make it optimize. Due to the absence of some attributes of some images, which making the use of ANN more appropriate, because ANN has the advantage of Robustness and Fault Tolerance. The experiment has build one Neural Network by processing one practice set first, and then classifies the test set by applying the built Neural Network. Result of the experiments have proved that after constant and repeat of the training, making the Classification Model have strong ability to adapt, and could predict the classification of web images with high accuracy。The First Part mainly states the background, aim, and meaning of this research.The Second Part mainly states the methods which are applied in the field of Images' classification, making the reader have a clear understanding of it.The Third Part mainly states the overall design of the researchThe Fourth Part mainly states the Model of collecting data for the research.The Fifth Part mainly states the relative content of algorithm of Artificial Neural Network (ANN).The Sixth Part mainly states the process of building ANN model and Relative results of this research.The Seventh Part mainly comments on the research according to the results of the experiment, and the improvements which will be made in the future.The whole research is based on Windows XP, developed by Visual C++ and MS SQL Server Database, MATLAB7.0.