Segmentation Method of High Resolution Remote Sensing Image Based on Fuzzy Gaussian Membership Function

Abstract

The present invention proposes a high-resolution remote sensing image segmentation method based on a fuzzy Gaussian membership function, supervises sampling and extracts training samples, and calculates the frequency value at which the gray value of each pixel in the training samples appears in the corresponding feature category; Establish Gaussian membership function models for different types of ground objects; fuzzify the Gaussian membership function model parameters, and establish fuzzy membership functions; establish a linear neural network model as the objective function of high-resolution remote sensing images, and integrate the spatial relationship to obtain the target of high-resolution remote sensing images Function matrix; divide the objective function matrix of high-resolution remote sensing images according to the principle of maximum membership degree; change the adjustment factor according to the set step size, and take the optimal segmentation as the final result. Based on the boundary information of the fuzzy membership function and the original membership function, the present invention establishes the objective function and integrates the spatial relationship, realizes the accurate fitting of the complex distribution characteristics of the high-resolution remote sensing image, effectively overcomes the noise, and improves the Segmentation accuracy.

Description

technical field [0001] The invention belongs to the field of image processing, in particular to a high-resolution remote sensing image segmentation method based on a fuzzy Gaussian membership function. Background technique [0002] Image segmentation is the most basic and critical task in remote sensing image processing, and has always been a hot and difficult issue in image processing. Because high-resolution remote sensing image data can present surface coverage information more clearly and meticulously, it has great potential and advantages in the precise segmentation of ground objects; at the same time, high-resolution features also bring new segmentation problems:( 1) The overlapping range of the distribution curves of different categories of ground objects increases the uncertainty of the pixel category; (2) The local feature data of the same category area is significantly different, which leads to the uncertainty of modeling. [0003] The above two uncertain features...

AI Technical Summary

Problems solved by technology

Because high-resolution remote sensing image data can present surface coverage information more clearly and meticulously, it has great potential and advantages in the precise segmentation of ground objects; at the same time, high-resolution features also bring new segmentation problems:( 1) The overlapping range of the distribution curves of different types of ground objects is large, which increases the uncertainty of the pixel category; (2) The local feature data of the same type of area is significantly different, which leads to the uncertainty of modeling

However, the above process of modeling the objective function does not take into account the influence of the spatial correlation between pixels on the segmentation results, so it is sensitive to noise

Images