【摘要】 滩涂在光学遥感影像中具有光谱信息综合的特点,单纯基于光谱特征的滩涂信息提取方法精度不高。本文在海岸带地物分类过程中,引入了纹理特征分析,发展了一种基于光谱和纹理特征融合的滩涂信息提取方法。该方法首先利用神经网络分别实现基于像元光谱特征和基于图像纹理特征的海岸带地物分类;在两种分类结果的基础上,建立融合模型,完成决策级的信息融合,从而得到最终的滩涂信息提取结果。试验结果表明,在海岸带地物分类过程中引入纹理特征分析是有效的,决策级的光谱与纹理信息融合可以使滩涂信息提取的精度达到92.3%。另外,神经网络方法用于海岸带地物分类亦是可行的。更多还原

【Abstract】 Usually there is comprehensive spectral information in optical image in the tidal flat,but the accuracy of the extraction information in the tidal flat based on the material spectrum features is lower.In this paper,the texture features analysis is introduced to the material classification in the coastal zone,and a method was developed based on the fusion of spectral and textural information.By means of the neural networks the material classification in the coastal zone based on the pixel spectrum features and image texture features are carried out.A decision fusion mode is constructed by above the two methods.The accuracy of the extraction information in the tidal flat was improved.The results show that the introduction of the texture features analysis is effective to the material classification of the coastal zone.The decision fusion of spectral and textural information can improve the accuracy of the extraction information in the tidal flat.The method of the neural networks is feasible in the material classification in the coastal zone.更多还原