Image division method based on inter-class maximized PCM (Pulse Code Modulation) clustering technology

Abstract

The invention discloses an image division method based on an inter-class maximized PCM (Pulse Code Modulation) clustering algorithm. The method comprises the following steps: carrying out classified labeling on pixel points of an input image according to a gray value; obtaining a clustering label when a clustering analysis method is used for dividing a target image; and carrying out performance evaluation on a label obtained by image division and an original label according to an evaluation index by a clustering method. The novel inter-class maximized PCM clustering algorithm considers the inter-class penalty, and parameters are adjusted and adjusted to enlarge the distance between class centers, so that the optimal classification of the pixel points in the image is realized.

Description

【Technical field】 [0001] The invention relates to the technical field of data mining and pattern recognition, in particular to an image segmentation method based on cluster analysis. 【Background technique】 [0002] Image segmentation is the technology and process of dividing the image into regions with different characteristics and extracting effective targets. It is an important link from image processing to image analysis, and the target of segmentation refers to dividing the objects in the image into Divided into different regions or categories, in essence, image segmentation is the process of clustering pixels. Image segmentation algorithms play a vital role in applications in many fields, such as medical biology, military, remote sensing, meteorology and so on. The existing segmentation algorithms can be roughly divided into the following categories [2] : Threshold-based segmentation algorithms, detection-based, region-based, cluster-based, and methods based on some s...

AI Technical Summary

Problems solved by technology

Although PCM is more superior than FCM in dealing with noise, when dealing with data with high overlap between classes, PCM does not fully consider the distance between cluster centers, which makes the PCM clustering algorithm difficult to process cluster centers. When the phenomenon of overlap occurs, the satisfactory effect is still not achieved

Someone proposed the kernel-based possibility C-means algorithm (KPCM), which is based on the idea of ​​the kernel algorithm. The points in the original space are mapped to the feature space through the kernel function, and the algorithm design, Analysis and calculation, so as to obtain the clustering division of the original space. This type of algorithm has the ability to discover non-convex clustering structures, and cluster from low-dimensional mapping to high-dimensional, which reduces the time complexity. However, from the analysis of a large number of experimental results It shows that there are still some problems in the original algorithm, which are sensitive to the initialization of parameters, and it is easy to form overlapping clusters

In addition, some people have proposed an algorithm to strengthen subspace clustering. Compared with other algorithms, this algorithm takes into account the distance between class centers, but the disadvantage is that it considers the distance between each class center and the center points of these class centers. The maximization of the distance between classes does not take into account the maximization of the distance between the class centers

Images