Remote sensing (RS) is a rising technology. It is widely used at lots of fields in recent year, such as topography, environmental engineering, agriculture, forest research. Base on each field's feature, some special RS technology appear.Information classification is one of the key techniques in the Remote Sensing (RS) domain. With the rapid development of the RS, Information classification has been playing an increasingly important role. However, because of the infection of noise and limitation of classification algorithm, the classification result always has a lot of fragmentations. In this paper, we present a new clustering algorithm base on traditional K-mean clustering and SVC clustering. Through this algorithm, we can get a relative accurate result compared with real instance.Both the original RS photo and the classification result are the gird picture. There are many limitations when using grid picture. In this paper, we also present a algorithm to convert grid picture to standard Shape File format vector file.Finally, we introduce a scheme of vector system. |