Research, Remote Sensing Image Based On Bp Neural Network Pattern Recognition Method

"Remote sensing", as its name implies, is sensing remotely . A large number of human beings have found that every object on Earth are continuously absorbing and reflecting launch of information and energy through the practice, and a human form - electromagnetic waves has been recognized, also found that the electromagnetic properties of different object are different. According to the principles of remote sensing is to detect objects on the reflection of electromagnetic waves and then to extract information from these objects in order to complete the remote identification. And remote sensing images is to obtain the picture from distant objects. It so as to show the image of the authentic environment surface in a form of tiny, which help human senses beyond their limits , and to monitor the dynamic changes of the earth's environment in a timely and quickly perceived manner with different spatial scales, It has become an important means to obtain the information of Earth's resources and environment.In conclusion, this article studies on the current status quo at home and abroad, bases on the use of BP neural network algorithm, wavelet transform combined with professional software and remote sensing PCI, and engaged in remote sensing image classification research, primarily to complete the following tasks:1. First review a number of research work made by the domestic and foreign scholars in the field of remote sensing image classification, focuse on principles and methods of BP neural network for remote sensing image classification and the principle of wavelet transform.2. Correct the original remote sensing image data integrate pre-processing, reduce noise by wavelet transform data processing.3. BP neural network is widely used remote sensing image classification in recent years. In this paper, the use of remote sensing image processing software PCI, the use of LandSat7 ETM+ remote sensing image data to extract the characteristics of four types of features, and then combined with BP neural network classifier for remote sensing images of their pattern recognition, and noted that the wavelet transform in BP neural network to enhance remote sensing the role of image recognition results.