一种小波域K-Means遥感图像分类标注算法

由于合成孔径雷达图像（遥感）的相干斑噪声数据丰富，导致传统的遥感图像分割方法分割效果不佳，采用学习理论和神经网络改善图像处理性能。根据图像统计特征，采取神经网络语义提出一种高效的图像纹理特征分割方法。首先，利用K-means聚类提取遥感图像的纹理特征，然后根据遥感图像在小波域中的分布特征对其进行滤波，最后利用语义对滤波后的遥感图像纹理特征和灰度组成的矢量进行分割归类，在遥感图像分割中快速标注分类以便于视觉分析。利用区域一致性分割分类，由聚类样本特征匹配进行图像分类标注，对变化检测进行统计分析，过分割或欠分割误差聚类样本不做标注，选取最佳样本聚类k值标注分类结果。

A Remote Sensing Image Semantic Classification Label of K-means Clustering on Wavelet Transform

Because of the large amount of noise data of polarized remote sensing image， the traditional remote sensing image segmentation method has a poor effect on obtaining better segmentation images. The remote sensing image contains rich texture information to facilitate the classification of remote sensing images. Therefore， the learning method uses the advantages of theory and neural network to improve the performance of image processing. According to the statistical characteristics of the image， this paper proposes an efficient image literary feature and semantic analysis based on neural network semantics. Firstly， the texture features of remote sensing images are extracted by K-means clustering， and then filtered according to the distribution characteristics of remote sensing images in the wavelet domain. Finally， the texture features of the filtered Remote Sensing image and the vector of the gray component are segmented and classified by semantics. The algorithm quickly labeled the classification in remote sensing image segmentation for visualization-analysis. Using regional consistency segmentation classification， finally image classification labeling based on cluster sample feature matching， statistical analysis before and after change detection， over-segmentation or under-segmentation error clustering samples are not labeled， and the best sample clustering value k is selected to label classification results.