Research And Application Of Density Clustering Based On Image Segmentation

Clustering is an important method for data processing in the field of machine learning.It is widely used in many academic fields.Such as the group division of target users,the value combination of different products,and the detection of outliers.In these scenarios,the density of data is usually uneven.Clustering on such data sets requires that the clustering algorithm is suitable for non-uniform density data sets.However,the clustering effect of traditional clustering algorithms on non-uniform density data sets is not ideal,and the traditional clustering algorithm relies heavily on parameters.To solve the above problems,a density clustering algorithm DCABISM based on image segmentation model is proposed.The algorithm does not rely on any parameters,and can better display the visual contour of each cluster when clustering non-uniform density data sets.The main work of this thesis is as follows:(1)A density clustering algorithm DCABISM based on image segmentation model is proposed.(2)A training set generation program is designed to generate 2-D point data with different distribution shapes and densities.(3)In order to meet the input needs of image segmentation model,a gray image generation program that can convert 2-D points into gray image is designed.(4)In order to obtain the final clustering results,a 2-D point generation program is designed to convert gray scale image to 2-D point.(5)In order to verify the clustering effect of DCABISM algorithm,a comparative experiment with other density clustering algorithms on the density uniform data set D31,R15 and the density non-uniform data set Pathbased,Jain is conducted by dissertation.(6)A algorithm of DCABISM is applied to the identification of wine.