Font Size: a A A

The Research On The Artificial Neural Networks And Its Application On Image Recognition

Posted on:2004-06-28Degree:MasterType:Thesis
Country:ChinaCandidate:Z M YangFull Text:PDF
GTID:2168360095456833Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
This thesis was Carried out basing on the Seals Automatically Reorganization project in a Information Science Company of ChongQing municipalization. In the paper, based on the analyzing of the mathematical models and their algorithms of kinds of Artificial Neural Networks (ANN), the author analyzed the algorithm of the Back-Propagation (BP) networks model, at the same time, the author improved the algorithm practically and effectively, and applied it in the recognition technique of digital image successfully. As a result, a satisfying result was achievedIn the paper, the writer analyzed the basic theories and the basic models of the classic Artificial Neural Networks. During this work, the writer worked out the reason that cause the Error Shock of the classic ANN, and bring out his improved project. Furthermore, the writer improved the arithmetic, the structure of the classic BP, and the processing methods to the sample data. Based on this work, the writer programmed a procedure to carried out his idea, and proved it with some instance. Focusing on the master paper, the writer carried out the application research work, based on the achieves of the Artificial Neural Networks researchers, the author applied the BP theory in the pattern recognition creatively, and have tried a series of experiments, especially in the clarity of digital image, as a result, the outlets are satisfy. The digital image recognition mathematic model that based on the BP networks provided a reference to the process of digital image.This thesis is organized as follow: Chapter 1 gives a general introduce of the basic theories and basic mathematic models of the Artificial Neural Networks, it briefly introduce some main mathematic models and some Machine Learning algorithm, such as BP,Hopfield,SOFM,Boltzmann machine,CMAC and etc. In the chapter 2, the author go deep into the BP mathematic model and it's algorithm, and emphases on the faults of the BP algorithm, analyzed the reason that cause the BP networks' error shaking, at the same time the writer advanced a improving algorithm. The 3rd chapter give the implement of the improved BP algorithm with VC++, and parts of it's source codes, a brief introduce of the machine learning software that base on the BP networks and is coded by the author is give out too. The application and application effect of the Machine Learning software that based on the BP networks is introduced in the 4th chapter. The lastchapter is introduction of the achievements of the application of BP networks and the application of machine learning software that based on the BP networks in the recognition of digital image, including the introduction of mathematic model and the application effect of the recognition of digital image.
Keywords/Search Tags:Artificial Neural Network, Back-Propagation networks, Machine Learning, Artificial Intelligence, Digital image
PDF Full Text Request
Related items