Research Of Agglomerative Hierarchical Clustering Method Based High-Resolution Remote Sensing Image Segmentation Method

Remote sensing image segmentation is the basis of pattern recognition and the first step for extracting the objects automatically from image using computer. This is significant for the application and analysis of remote sensing images. Clustering analysis as a non-supervised learning method is widely used in the remote sensing image segmentation. It has made good results in the segmentation of low-resolution remote sensing image. But as the improvement of image resolution, the existing common clustering segmentation algorithm has low segmentation accuracy and too many broken spots and other deficiencies. They can not satisfy the requirements of the high-resolution remote sensing imaging application. For these issues, this paper studies the existing common clustering segmentation algorithm and analyzes their deficiencies in the segmentation of high-resolution remote sensing image through experiment. Considering their deficiencies, this paper studies the Agglomerative Hierarchical Clustering Method based High-Resolution Remote Sensing Image Segmentation Method. The main research works of this paper are as follows:Study the basic principles of data mining, functions and processes and summed up the application of dada mining in various fields. Have an in-depth study of Clustering Analysis and analyze the advantages and disadvantages of various kinds of Clustering Analysis algorithms.Study the related knowledge of remote sensing and remote sensing image segmentation. Analyze the advantages and disadvantages of various kinds of remote sensing image segmentation algorithms. Introduce the clustering segmentation algorithm of remote sensing image: K-Means and ISODATA and uses high-resolution image to test and analyze them.For the deficiencies of K-Means and ISODATA in remote sensing image segmentation, study the agglomerative hierarchical clustering method based High-Resolution Remote Sensing Image Segmentation Method. This algorithm takes the spectral and shape characteristics into account and has image meshing, broken grouper and noise reduction and other functions. This algorithm can effectively improve the availability and accuracy of image segmentation results. To test the effectiveness of the approach, this paper takes parameters experiment, the efficiency of image meshing experiment and the comparison experiment. The experiment results show that the algorithm can solve the deficiencies of current clustering segmentation algorithms and get a good segmentation results.