Remote Sensing Image Segmentation Method Based on Markov Random Field and Hybrid Kernel Function

Abstract

A remote sensing image segmentation method based on a Markov random field and a mixed kernel function belongs to the field of remote sensing image segmentation and identification technology. The remote sensing image segmentation method is characterized in that the method mainly settles a problem of incapability of simultaneously satisfying requirements of area consistency and detail information ofa segmentation result in prior art. The method comprises the realizing steps of 1, inputting a remote sensing SAR image with spatial resolution of 0.3 meter; 2, according to an area diaphragm of theremote sensing SAR image, dividing the remote sensing SAR image to a structural area subspace and a homogenous area subspace; 2, dividing the homogenous area subspace by means of the Markov random field based on a Gaussian radial kernel function; 3, dividing the structural area subspace by means of the Markov random field based on a threshold Ridgelet kernel function; and 4, combining the dividingresult of the homogenous area subspace and the structural area subspace, and obtaining a segmentation result of the remote sensing SAR image. The remote sensing image segmentation method realizes a good segmentation effect of the remote sensing SAR image and can be used for remote sensing image segmentation.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a remote sensing image segmentation method, which can be used for image classification, recognition and detection. Background technique [0002] At present, some research results have been obtained in remote sensing image segmentation technology, and various segmentation methods have been proposed. Classical remote sensing image processing methods include threshold-based methods, clustering methods, and so on. Based on the grayscale information of pixels, this type of method designs a variety of features suitable for remote sensing images, such as gray-level co-occurrence matrix features, Gabor features, SIFT features, half-difference features, and so on. Extract features for each pixel, and use clustering methods to obtain image segmentation results for the extracted features, such as K-means clustering, hierarchical clustering, AP clustering, fuzzy C-means and other c...

AI Technical Summary

Problems solved by technology

Because the different spatial structure characteristics of remote sensing images are not considered, the regional consistency and detail information retention in the segmentation results cannot be satisfied at the same time, which affects the subsequent classification, recognition and detection of remote sensing images.

Images