Side View Image Segmentation Algorithm of Buildings Considering Edge Features and Image Peaks
In recent years,building image segmentation technology has been rapidly improved,but most of them are segmentation algorithms for remote sensing orthophoto images.With the development of street view and augmen-ted reality technology,the demand for building side view image segmentation is increasing.To solve this problem,a combination algorithm combining edge features and image peaks is proposed.Firstly,the preliminary image seg-mentation is performed by using Canny edge detection algorithm.Then the improved skyline detection algorithm is used to form the boundary between building and sky,and the adjacent buildings are separated by means of image peaks.Secondly,the green channel of the original image is used to remove the interference of trees,and realize building edge extraction and shape extraction.Finally,the images in MSRA10K dataset are used to verify the pro-posed method.Experiments show that the algorithm does not depend on human-computer interaction and has better extraction effect in the case of small field of vision and good background compared with GrabCut algorithm.
万方数据
维普
