The Research Of Man-made Objects Extraction Methods Based On QuickBird Images

The rapid development of remote sensing technology, especially the emergence of high-resolution images provides a large number of datum for obtaining the space information of ground. However, the application of modern remote sensing technology, especially the high-resolution remote sensing technology lags far behind the level of remote sensing data obtaining, thus greatly hinders the popularity and applications of high-resolution remote sensing. The current development of remote sensing tells us that the application of remote sensing research has a long way to go, high-resolution ones need more efforts. Man-made objects are the mainly component in urban area, whose distribution and amounts have a close relationship with the population, economy, and culture of the city. Meanwhile, the concentration extents of man-made objects reflect the development level of such a city. This paper selects the latest QuickBird remote sensing image, which is located in Yong-ding District of Zhangjiajie. A new method of extracting man-made objects was proposed based on the QuickBird image by making full use of spectral information and morphological characteristics. The main contents include:(1) From the analysis of the main spectral features of various land style especially man-made objects in the study area, the types of land-use can have a simply merged processing by referencing the V-I-S model. According to the practical application, the land utilization patterns in the study area are divided into three major urban land classes finally, including manmade object, vegetation and water.(2) According to the spectral and morphology features of three land types (vegetation, water, manmade object), try to find an effective method for extraction of man-made objects. After tested repeatedly, selects image-related band to enhance the information of artificial objects and weaken the others, and then based on the result of Band operation finishes morphological filtering by using morphological characteristics. Ultimately, man-made object can be extracted from the city successfully.(3) Utilized the actually measured data that comes from the second national land survey of city to evaluate the new model which is proposed in the article. It proves that the new method of man-made objects extraction in high-resolution images is fast, accurate and highly-efficient. This method has important application value in intentionally obtaining the thematic information of man-made objects from QuickBird images and updating the GIS databases.