Remote sensing image classification marking method based on composite graph conditional random field

Abstract

The invention discloses a remote sensing image classification marking method based on a composite graph conditional random field in allusion to a classification marking problem of remote sensing images. The method comprises the steps of acquiring a training sample and a test sample through manual sample collection, wherein the training sample contains a category marking truth value; building a composite graph; defining a correlation potential function and an interaction potential function of the conditional random field; optimizing a conditional random field model through a quasi-Newton method; and carrying out reference on the test sample through a loop belief propagation algorithm so as to acquire a classification marking result. In the invention, the composite graph is built through combining a sparse graph which is capable of expressing global interaction information and a spatial graph which is capable of expressing local spatial interaction information, the data expression capacity of a graph structure is enhanced, thus the classification marking performance of the conditional random field is enhanced, and high classification marking precision is provided.

Description

technical field [0001] The invention belongs to the technical field of image processing and pattern recognition, and in particular relates to a method for classifying and marking remote sensing images based on conditional random fields of compound graphs. Background technique [0002] Remote sensing technology is the most intuitive, richest and most effective technical means to detect the comprehensive information of land cover. With the rapid development of sensor technology, remote sensing images are gradually showing the characteristics of multi-spectrum, high resolution and large amount of data. The effective acquisition of a large number of comprehensive and information-rich remote sensing images provides complete information resources for the development of related scientific research, but also puts forward higher requirements for remote sensing image processing technology. [0003] In the field of remote sensing, the classification and labeling of remote sensing imag...

AI Technical Summary

Problems solved by technology

However, the spatial graph can only fuse the local interaction information of the spatial neighborhood, and lacks the ability to fuse the global interaction information, which affects the classification performance of the conditional random field to a certain extent.

Images