Residents To Non-supervised Classification Of Satellite Images Feature Extraction

With the rapidly development of space techniques and information techniques, the high resolution secondary planet can supply human with the digital images of higher accuracy and larger capability. Target extraction from images has become the important means of the space information updating, and has been used extensively in the national economic production and martial target detection.QuickBird, launched by the American Digital Globe Co., is the most high resolution commercial remote sensing satellite in the world, whose resolution can even be 0.61m. This paper mainly studied the large residential area extraction of a county in YunNan province caught by QuickBird, by using the software of ERDAS. First, we did the data processing, including the geometric correction, histogram breakpoints and so on. Then did the initial classification of the image using iterative self-organizing data analysis technique (ISODATA), because of without the acquaintance of this area. We concluded why some areas belonged to a wrong class. It's necessary to do the post-classification process after this, such as clustering statistics and recode, because of the uncertainty of image. At last, we concluded that this research is successful according to the comparison of the extracting result with the large-scale land use vector-graph of the same area.The cooperation of multi-subject and unremitting effort is the essential condition for automatic spatial image understanding and interpretation. With long term researching and studying, the era of intelligent acquisition and processing spatial information from space imagery will by all means come true!...