论文部分内容阅读
提出了两种基于主成分分析与局部二值模式的高光谱图像分类算法。利用主成分分析去除高光谱图像的谱间冗余信息,对降维后的图像利用局部二值模式进行空间纹理特征分析,采用稀疏表示分类和支持向量机分别对提取的特征进行分类。其通过将主成分分析与局部二值模式相结合对高光谱图像进行特征提取,保证了高光谱图像的谱间冗余的有效去除,同时保护了高光谱图像的空间局部邻域信息,因此,此类算法不但能充分挖掘高光谱图像的谱间-空间特征,在较大程度上提高分类精度和Kappa系数,而且在高斯噪声环境中和小样本情况下也具有良好的分类性能。
Two hyperspectral image classification algorithms based on principal component analysis and local binary mode are proposed. The principal component analysis (PCA) is used to remove the spectral redundancy information of hyperspectral image, the local binary pattern is used to analyze the feature of the space-reduced image, and the sparse representation and SVM are used to classify the extracted features separately. By combining the principal component analysis with local binary pattern, the feature extraction of hyperspectral image is carried out to ensure the effective removal of hyperspectral image spectrum redundancy and to protect the spatial local neighborhood information of hyperspectral images. Therefore, Such algorithms can not only fully exploit the spectral-spatial features of hyperspectral images, but also improve classification accuracy and Kappa coefficient to a great extent, and also have good classification performance in Gaussian noise environments and small samples.