结合Gabor滤波和同质性判定的高光谱图像分类

针对传统高光谱图像分类算法多利用目标类别光谱信息而忽略空间信息的问题，提出了一种综合利用空间信息与光谱信息的分类算法．首先，利用主成分分析（PCA）和无参数加权特征提取（NWFE）分别对高光谱数据进行特征提取；然后，在PCA第一主成分的基础上进行二维Gabor滤波得到像元纹理特征，结合纹理信息与光谱信息利用支持向量机对图像分类；最后利用多尺度区域同质性判定进一步改进图像分类精度．实验表明，该算法能够消除“噪声”像元，有效地提高图像分类精度．

Hyperspectral imagery classification combined with Gabor filtering and homogeneity discrimination

Most of traditional hyperspectral image information, while ignore spatial information. To classification algorithms make use of target classification spectral solve this problem, a classification algorithm which comprehen- sively makes use of spatial information and spectral information is proposed. Firstly, two different feature extraction methods ： principal component analysis （PCA） and nonparametric weighted feature extraction （NWFE） are applied to the feature extraction of hyperspectral data, respectively. Then based on PCA, the two-dimensional Gabor filte- ring is applied to get texture feature of the pixel, then by integrating texture and spectral information, using the support vector machine to classify the image. At last, the image classification accuracy is further improved by using multi-scale regional homogeneity discrimination. Experimental results show that the proposed classification algo-rithm can eliminate noise pixels, achieving better accuracies than the conventional spectral classification.