Acquiring the information for rural settlement timely and accurately has an important significance for construction and development of rural areas. The development of remote sensing technology provides advanced means of the acquirement of the information of settlement. The study of extracting rural settlement information from Quickbird images in Xindu district, Chengdu City, P.R.of China was discussed here. Firstly, The Quickbird images such as panchromatic image and multi-spectral images were processed by geometric correction, enhancement and fusion. Secondly, the homogeneous image objects were formed by using multi-scale segmentation technology based on knowledge. Thirdly, the features such spectral feature, spatial relationship feature, texture feature and geometric feature of the image objects were obtained for each image object by using feature calculation. Fourthly, the feature knowledge of rural settlement unit and its component were obtained by using knowledge discovering. Finally, the rural settlement unit and its component information were extracted by matching the features with the feature knowledge of rural settlement unit and its component based on reasoning. It was shown that the rural settlement unit and its component information can be effectively extracted from Quickbird images by using our proposed method in this paper.

Extraction of the housing generally based on the extraction pixels, unable to spatial information of images. This paper presents a object-oriented high resolution satellite images extraction housing me- thods. get image objects from image segmentation and then through analysis housing characteristics of the QuickBird images, using the spectral characteristics, geometry and spatial characteristics of image objects to establish the knowledge database. use the rules of the knowledge database to extract housing from images.

The correlation analysis of the hyper-spectrum images in different bands is very important to the applications of the hyper-spectrum images.The correlation of the hyper-spectrum images acquired by airplane in the north part of Tengchong County was analyzed in this paper.Firstly,the hyper-spectrum images were divided into low quality images and high quality images by visual interpretation.The high quality images were used for further study.Secondly,the mean and standard deviation were calculated for each band image.Thirdly,the correlations among images were analyzed and the images were divided into four groups.The correlation features among groups and in each group were discovered.Fourthly,the distribution features of the correlation,the relationship features of the correlations above 0.9 or below 0.1 and the interval spectral distance among images were discussed.Some conclusions are drawn as follows: 1) Among all bands,band 24(2.217~2.249 μm) is of highest information amount.2) In each band group,the correlation coefficient is above 0.93.3) The frequency distribution of the correlation coefficient grades is of the characteristics with higher value in the two ends and lower value in the middle.The correlation coefficient above 0.9 accounts for 38%.4) The correlation coefficient above 0.9 mainly appears among images with the interval spectrum distance of 0,which account for 20%.The results are useful to the design and selection of the hyper-spectrum bands.