结合二维EEMD和小波分解的遥感图像纹理方向检测  

Detection Method for Remote Sensing Image Texture Direction Based on Bi-dimensional EEMD and Wavelet Transform

作　　者：刘晓杰[1] 范虹[1] 党小虎[2] 

机构地区：[1]陕西师范大学计算机科学学院,陕西西安710062 [2]西安科技大学地理与环境学院,陕西西安710054

出　　处：《地理与地理信息科学》2016年第1期66-70,共5页Geography and Geo-Information Science

基　　金：国家自然科学基金项目(41271518);陕西省自然科学基金项目(2014JM2-6115);中央高校基本科研业务费项目(GK200902018)

摘　　要：遥感图像纹理特征的提取和分析,可以弥补光谱特征在提取遥感图像信息特征方面的不足,提高遥感图像分类、识别的精度。为了得到遥感图像纹理特征中较重要的方向性特征,提出一种检测遥感图像纹理方向特征的新方法:先用改进后的二维集合经验模态分解算法(BEEMD)对原始图像进行处理;再对分解结果进行小波分解与Radon变换来检测图像纹理的方向性特征。实验结果表明,所提算法既能克服小波分解高频信息泄露的缺陷,增强小波分解的适用性;又能获得遥感图像纹理在低频、垂直和水平等部分较为准确、精细的方向特征检测结果,为遥感图像的识别与区分提供更有效的依据。The extraction and analysis of remote sensing image texture feature can make up for the inadequacy of spectral feature extraction,can improve the accuracy of remote sensing image classification and recognition.In order to get the direction of the more important features of remote sensing image texture feature,this paper presents a new method of processing the texture direction features of remote sensing image detection technology based on digital image.First of all,the original image is decomposed to a number of BIMF function via two-dimensional improved ensemble empirical mode decomposition algorithm（BEEMD）;then,the BIMF1 is decomposed by wavelet;finally,the results of wavelet decomposition are performed on Radon transform to obtain the remote sensing image texture directional characteristics.The experimental results show that the combination of BEEMD and wavelet transform algorithm,which can overcome the wavelet decomposed high frequency information leakage defect,enhance the usability of the wavelet,and get more accurately detect the direction of the remote sensing image texture feature in the low frequency,vertical and horizontal parts,which can provide a more effective basis for the subsequent recognition of remote sensing image and distinguish work.

关 键 词：BEEMD 小波分解 RADON变换 遥感 图像纹理方向检测 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]