辅以纹理特征的遥感影像神经网络分类  

The Neural Network Classification of Remote Sensing Image Supplemented by Texture Characteristic

作　　者：裴亮[1] 谭阳[1] 

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000

出　　处：《测绘与空间地理信息》2008年第4期66-67,73,共3页Geomatics & Spatial Information Technology

摘　　要：将由变差函数提取出的纹理信息与光谱信息相结合,运用神经网络方法进行分类和地物识别,并与最大似然法的分类结果进行对比分析,结果表明该方法明显提高了遥感影像的分类精度。The study adopts the BP neural network method to classify and recognize the features combining the textural information extracted by variogram and spectral information. Then it compares and analyzes the classification results by maximum likehood method. The analysis result shows that this method can improve the classification precision.

关 键 词：遥感影像 纹理 变差函数 BP神经网络 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]