Research And Application Based On K-means Algorithm And Hierarchical Clustering Algorithm

With the rapid development of various industries in the world, people are faced with more and more data, the era of the big data is coming. People began to explore the impact of big data that brings to our life and social activities. In the study, the processing method of the big data has become one of the central issues of concern.In the processing of big data, cluster analysis is an important means. It has a very wide range of applications, like in the field of medical diagnostics, image processing, information retrieval, statistics, biology and some other disciplines. Because of clustering algorithm enables better segmentation effects in the process of image segmentation, it attracted wide attention of experts and scholars. Image segmentation is not only an important step from image recognition processing to image analysis, but also a key link which using computer vision from a lower to a higher one.This paper on the basis of the principles and applications of clustering algorithm which based on the background of big data, drawing lessons from previous research experience, research and implement the generation, principle and realization of algorithm. Meanwhile, clustering algorithms were compared, verify the feasibility of the algorithm and the effect of the clustering, specific describe the principal, the implementation, and the pros and cons between the k-means algorithm and the hierarchical clustering algorithm, use the Java software to simulate the segmentation algorithm and have got some research experience. On this basis, I analyzed the existing shortcomings of clustering algorithm, thus, put forward to the improved proposals, and then use the Java language to achieve it. In the meanwhile, I also analyze the advantages and disadvantages of the basic clustering algorithm. For the optimization of clustering, I start from its shortcomings, gradually eliminate the problems arising from it, so that make the improved algorithm can produce a better result than the basic clustering algorithm does. At last, record the improvement processing of clustering algorithm, test the improves algorithm, make it gives the better clustering results in a certain environment and a certain range of data, provide a reference of the application of clustering algorithm in image processing, and it provides important inspiration in studying and solving the other similar problems like the recognition of the complex patterns.