Texture Analysis Of Remote Sensing Image And Its Appliction In Seismic Disaster Information Extraction

Texture features are an important feature of remote sensing images,and texture analysis plays an increasingly important role in remote sensing image recognition.A lot of research shows that texture analysis is an important means to improve the accuracy of image classification,and the key to image classification is whether the image texture features as well as select the appropriate feature dimension that has been popular for research.This paper conduct a research for feature extraction of image texture based on the texture feature of typical feature in the image,and the image classification is extracted according to the extracted texture eigenvalues and applied to the earthquake disaster information extraction,which provides information reference for the post-earthquake rescue.The details are as follows:Firstly,on the basis of reading a large number of previous research results,we analyze the texture features of typical objects such as buildings,mountains,water and so on.These features can well distinguish different objects.In this paper,the commonly used methods of texture extraction,such as statistical analysis method,geometric feature method,signal processing method and model method,are summarized.The gray level co-occurrence matrix(GLCM),the travel length matrix and the gray-scale difference matrix method in the statistical analysis method are defined as the feature extraction method by analyzing the research status and characteristics of these texture analysis methods.The feature extraction algorithm is implemented to calculate the eigenvalues of the image.Besides,the feature dimension of these methods are selected and the weight classification algorithm is used to compare the image classification accuracy with different dimension features,so the image classification experiment is carried out by using 15-dimension feature to achieve great speed and good classification accuracy.The remote sensing images before and after the Yushu earthquake are selected as the experimental samples,and three samples representing earthquake disaster information are experimented,which are good buildings,serious collapsed buildings and slight collapsed buildings.The classification method is selected as the decision tree algorithm and the support vector machine(SVM).The classification contains a single classification and mixed classification: the buildings are the samples of single classification and the buildings,mountains,water,roads are the samples of mixed classification.The experimental results show that the method used in this paper can achieve a classification accuracy over 87% and can identify buildings from a large number of samples.Based on these experimental foundations,disaster information extraction is carried out,and the objects are classified directly on the image area after the earthquake and the building samples representing earthquake disaster information are identified.The experiment results show that the proposed method can well extract the earthquake information and the feature dimension has the advantages of high classification accuracy and fast speed.