Image segmentation method based on local region homogeneity manifold constrained MRF model

Abstract

The invention discloses an image segmentation method based on a local region homogeneity manifold constrained MRF model. On the basis of a PairwiseMRF model, an image segmentation model based on a local region MRF model is constructed on an extended neighborhood of MRF nodes, and the prior distribution of a local region effectively avoids the interference of noise or texture mutations; and meanwhile, on the basis of a manifold learning theory, a manifold regularization term under a probabilistic framework is established, a local region probability distribution is used to effectively describe the local spatial geometric structure prior with complex natural images, and the local spatial geometric structure described by this manifold learning is introduced into the local region MRF segmentation model. Experiments have proved that compared with the prior art, the method of the invention not only avoids the local region prior oversmoothing penalty, but also effectively maintains the local geometric structure information of the image segmentation.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to an image segmentation method. Background technique [0002] Image segmentation methods based on the Markov Random Field (MRF) model are widely used. This type of segmentation method is based on the local spatial correlation of image information, and uses a two-dimensional random field model to describe the feature information of the image. The image segmentation model of MRF enhances the correlation of structural information, which greatly improves the speed and accuracy of natural image segmentation, but for natural images with rich statistical features, the currently commonly used MRF based on point-to-interaction structure The (Pairwise MRF) model cannot fully describe the complex statistical information and prior knowledge of natural images. Therefore, in the image segmentation problem, this image processing algorithm based on point-to-MRF often mis-segments in the i...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide an image segmentation method based on the local region consistency manifold constraint MRF model to solve the problem that the conventional regional MRF image segmentation model based on Euclidean space distance measurement technology cannot describe the global consistency characteristics of complex high-dimensional data Problem; the present invention can effectively improve the effect of image segmentation

Images