Building extraction in towns and villages based on Digital Aerial Image by texture enhancing

With the continuous improvement of remote sensing image, especially the appearance of Digital Aerial Image, make the texture features of image more significantly and the deeply mining possibly and the object-oriented information extraction method has been applied on digital aerial image since the object-oriented information extraction method has been proposed, it has been applied in many fields and achieved better classification results than the traditional methods. In the process of object-oriented classification, the image quality of segmentation is the guarantee of the image classification accuracy. The current mainstream thinking of segmentation mainly considered about the four features of image, such as color, shape, smooth and compactness. The method based on texture enhancing of digital aerial image is brought on. In this article, firstly the original Digital Aerial Image is preprocessed by edge detection, principal component analysis and the texture filter of second-order probability statistics; Secondly the gray image of Contrast Texture was got through the sharpening window of 7×7; Then taking the gray image as independent band, false color composite were processed with the band combination(Contrast, the original Digital Aerial Image G, the original Digital Aerial Image B). The abundant of texture features in digital aerial image are translated into sensitive multi-scale segmentation spectral feature by using image enhancement technology, as it can promote segmentation effects. Building detail texture features have been involved in the process of multi-scale segmentation, and building segmentation will be more fully and sensitively. Finally, based on the false color image, multiple segmentations and building extraction in towns and villages were processed. Taking the Yanqing County Beijing Digital Aerial Image as an Example, building extraction was processed by method that mentioned above. Compared to the object-oriented classification method, it n- - ot only highlighted the edge of the buildings, but also reduced the redundant segmented objects. Besides it gets an effective solution to the shadow of the building and its confusing area, optimized the feature space, and improved the accuracy of classification.¹