Research On Image Segmentation Methods Based On Density Peak Clustering

Image segmentation is an important preprocessing technology,which has been widely used in various fields of computer vision.With the development of artificial intelligence,the demand for research methods of image is increasing,and various image segmentation methods are emerging one after another.Among them,clustering-based image segmentation methods have attracted great attention from researchers because of their simple ideas.Therefore,based on the previous studies,the thesis combines the advantages of density peak clustering to propose some new methods of image segmentation.The main contents of the thesis include:1.An image segmentation method based on density peak clustering is proposed in this thesis.Firstly,this thesis proposes an improved density peak clustering method based on the traditional truncated distance of density peak clustering because of empirical value and the disadvantages of its clustering center relying on manual selection: the adaptive truncation distance is obtained by introducing information entropy,and the clustering center is automatically selected according to the slope.Then,the improved peak density clustering algorithm is applied to image segmentation.Firstly,the pixel of the image is inputted as the data of the peak density clustering.After the peak density clustering is processed,the subsets generated by the clusters are marked with different colors,and then the corresponding subsets of the targets are searched for and displayed to achieve image segmentation.Finally,the experiment verifies the feasibility of the method.2.Aiming at the problem that the time of segmentation method based on density peak clustering is too large,the image segmentation algorithm based on grid and density peak clustering is proposed.The grid can convert large amounts of data to small amounts of data to some extent.In this method,the pixels of the image are meshed regularly,the method calculates the color value of the grid that is used as the basic element of the density peak clustering to realize image segmentation.But this method is easy to cause the boundary of the segmented image to be blurred,therefore on the basis of the proposed image segmentation method based on grid and density peak clustering,an image segmentation method based on simple linear iterative clustering and density peak clustering is proposed in this thesis.Simple linear iterative clusteringpreserves image edge information while reducing the data during image segmentation.In this method,a suitable number of seeds are selected,and a simple linear iterative clustering is carried out to get the corresponding superpixels and the average color value of all the pixels in the superpixel is taken as the color value of the superpixel.Then,the superpixel is used as the basic element of the density peak clustering to realize image segmentation.Finally,a series of experiments verify the effectiveness of the methods...