Color image segmentation method based on Gaussian mixture model and support vector machine

Abstract

The invention discloses a method for segmenting color images by means of combining a Gaussian mixture model and a support vector machine. The method mainly includes extracting characteristics of an image; building the Gaussian mixture model; and classifying by the aid of the support vector machine. A particular process mainly includes firstly, extracting color characteristics and texture characteristics of the image; then building the Gaussian mixture model and obtaining new characteristics of the extracted original characteristics by the aid of the Gaussian mixture model; secondly, obtaining an initial segmentation result by the new characteristics; and finally selecting training samples according to the initial segmentation result, classifying the training samples by the aid of the support vector machine and obtaining a final segmentation result. When the characteristics are described, the image is initially segmented according to the new characteristics obtained via the Gaussian mixture model (GMM) without being based on the original characteristics, and the final segmentation result is obtained by the aid of the support vector machine (SVM). Time-space domain information of the image is sufficiently utilized, shortcomings of the Gaussian mixture model (GMM) which builds a module with a complicated background only by the aid of time domain information are overcome, and segmentation accuracy is effectively improved.

Description

technical field [0001] The invention belongs to the technical field of image segmentation for multimedia information processing, and in particular relates to a color image segmentation method combining a Gaussian mixture model and a support vector machine. Background technique [0002] There are many methods of image segmentation at present, and the early image research mainly divided the segmentation methods into two categories, one is the segmentation method based on the boundary, and the other is the segmentation method based on the region. The region-based segmentation method relies on the spatial local characteristics of the image, such as the uniformity of grayscale, texture and other pixel statistical properties. Typical methods based on region segmentation include region growing, region splitting and the combination of region growing and splitting. Because these methods rely on the gray value of the image, their main advantage is that they are not sensitive to noise...

AI Technical Summary

Problems solved by technology

These methods do not depend on the results of processed pixels and are suitable for parallelization, but the disadvantage is that they are sensitive to noise, and when the edge pixel values ​​do not change significantly, it is easy to produce false boundaries or discontinuous boundaries.

Images