基于关联规则的含噪高光谱图像分类系统

Design of noisy hyperspectral image classification system based on association rules

为消除传统高光谱图像分类系统使用特征单一、分类准确率低的不足,提出了基于关联规则的含噪高光谱图像分类系统设。将关联规则挖掘算法应用于遥感领域,并依据关联规则设计了高光谱图像分类系统的特征模块,模块单元由滤波器、灰度共生矩阵模块、ASM模块、对比度处理模块、均匀度处理模块等部分构成;原始含噪高光谱图像输入分类系统后,先对其进行滤波降噪处理,基于灰度共生矩阵提取除噪后高光谱图像的细节特征,再依据图像之间的内在关联规则,实现对高光谱图像的精确分类。结果表明提出的系统设计,在含噪高光谱图像分类准确率改善方面效果显著。

In order to eliminate the shortage of traditional hyperspectral image classification system with single feature and low classification accuracy,a new classification system based on association rules for noisy hyperspectral images is designed and studied. The association rule mining algorithm is applied to remote sensing field,and the feature module of hyperspectral image classification system is designed according to the association rules. The module unit is composed of filter,gray level co-occurrence matrix module,ASM module,contrast processing module,uniformity processing module and more. The original noise high spectral image input classification system is first introduced. After that,the filtering and noise reduction processing is carried out. Based on the gray level co-occurrence matrix,the detail features of the hyperspectral image are extracted and the accurate classification of hyperspectral images is realized according to the inherent association rules between the images. The results show that the proposed system design is effective in improving the classification accuracy of noisy hyperspectral images.