基于灰度与纹理信息融合的影像分割方法

文章提出了一种融合遥感影像像元各波段灰度信息与影像纹理信息的图像分割方法, 设计并实现了一种基于四分树的区域分开-合并算法。该方法有效地利用了遥感影像像元各波段灰度信息以及影像纹理信息, 用SPOT 5 卫星数据的实验结果表明, 该方法对影像分类以及制图工作有一定的促进作用。

Image Segmentation Method Based on the Fusion of Gray Information and Texture Information

The image segmentation is not only a key step from the image treatment to the image analysis,but also one of the oldest and the most difficult problems in the image treatment.The significance of image segmentation in the image treatment is analyzed in this paper,followed by the characteristics and existing problems of the current image segmentation methods.Based on these analyses and aiming at the remote sensing image,an image segmentation method based on the information fusion of gray information of remote sensing image pixels of every band and texture information of remote sensing image is put forwards from the angle of information fusion,and thus a quadtree-based region split-and-merge algorithm was designed and realized.The SPOT5 satellite image obtained on 31 March 2004 in the Zhixin County of Jiangsu province,China,was treated by the above image segmentation method and the proposed quadtree-based region split-and-merge algorithm.It is found that this image segmentation method could utilize the gray information of remote sensing image pixels of every band and texture information of remote sensing image effectively,reduce the partial working capacity in practical drawing work,and provide some serves for the classification of remote sensing image.Based on these treating results,it is inferred that the proper combination of many kinds of information may provide a new idea for solving some problems,but the methods how to make many kinds of information fused more effectively,need to be deeply studied in the practical research work.