论文部分内容阅读
为了进一步提高稀疏表示分类能力,提出了基于联合稀疏表示算法与形态学特征的高光谱图像(HSI)分类算法。该算法对高光谱图像提取主成分特征图,并利用结构元素对主成分特征图进行多维的空间结构特征提取,结合提取的形态学特征与原始光谱特征,利用联合稀疏表示算法将同一空间区域中的像元联合进行稀疏系数矩阵的求解,最终通过最小残差判断准则确定像元类别。在AVIRIS与ROSIS HSI上的实验结果表明,该算法在分类效果和分类总精度上都有显著提高。
In order to further improve the classification ability of sparse representation, a HSI classification algorithm based on joint sparse representation algorithm and morphological features is proposed. The algorithm extracts the principal component features of the hyperspectral image, extracts the features of the principal component features from the features of the principal components by using the structural elements, and combines the extracted morphological features with the original spectral features. By using the joint sparse representation algorithm, Of the pixels jointly sparse coefficient matrix solution, and ultimately determine the minimum residual pixel criterion categories. The experimental results on AVIRIS and ROSIS HSI show that the proposed algorithm has significantly improved the classification accuracy and overall classification accuracy.