The Object Detection And Extraction Of Surface Features In Real Scene Image

With the development of remote sensing technology, the main surface features object detection and extraction from real image has an important academic value and application prospects in the national economy and national defense. This paper studies the detection and extraction technology of surface features object based the real image of Google Earth, the main research target is the extraction of buildings, roads and water surface. For the detection and extraction of main building, first, use green vegetation index to preprocess origin image, to remove the interference of vegetation in the image, second, use region growing watershed algorithm, combined with prior knowledge of the building, and the mathematical morphology to extract the potential region of building. At the same time, extract the shadow area of the buildings, finally, accord to object-relational features of the shadow of the adjacent buildings and their characteristics, excluding roads, squares and other non-building area, the rest are building area. For the detection and extraction of main roads, first, use vegetation index to preprocess image, second, use two thresholds OSTU algorithm and local gray level consistency algorithm to segment image, and then, combine segmentation images, use the shape index to validate, finally, use morphological processing to remove noise and get the road boundary. For the detection and extraction of main water area, first, use threshold neighborhood algorithm to obtain the initial outline of the water area, second, combine with the shape index to reduce the noise of initial extraction, finally, use mathematical morphology to eliminate the impact of other objects on the water surface and get boundary of water area. Finally, choose different types of structure of real scene image and use the method proposed in this paper to extract the buildings, roads and water areas. The Experimental results show that the method proposed in this paper is effective to detect and extract the main surface features.