摘要： Lately, with progresses in remote sensing information techniques and the growingly and unprecedented uses of its uses, remote sensing became a science that cannot be dispensed with in most ?elds and ground object extraction has turned out to be more exact. Remote sensing image of high spatial resolution gives more inconspicuous components for instance, shape, color, size. The use of the old pixel-based method of classi?cation images inevitably leads to a signi?cant sacri?cing of image classi?cation accuracy. The use of unconventional methods such as object based image analysis (OBIA) to obtain data from image of high spatial resolution becomes the focus of many researchers. The initial phase of the OBIA technique is segmentation, which is a procedure that partition an image into moderately homogeneous areas named segments. Because the conventional pixel-based method does not suit the classi?cation of remote sensing images with spatial resolution, it has been replaced by a new standard method OBIA. Choosing the parameters of segmentation is a fundamental stage in the image segmentation process, the main purpose of this research is to try to ?nd the best values or near the best values for the parameters of image segmentation. It is expected to obtain an image object that expresses the reality and therefore obtain the accuracy of the classi?cation of the satellite images if the selection of good and appropriate segmentation parameters well done. There are three parameters that have a signi?cant impact on the accuracy of the results of the segmentation must be determined their values with high precision, where they can be arranged from the lowest up, these are compactness, shape scale. Dependence on use of visual analysis alone in determining the values of these parameters is a waste of time. Consequently, in this paper, a set of segmentations was carried out utilizing the Worldview-3 image with different values for the segmentation parameters to de?ne ideal or close ideal segmentation parameters used to extracting building from remote sensing images.