高光谱图像分类是高光谱图像处理领域的研究热点，它旨在通过标记样本上学习

Hyperspectral image classification is a hot topic in the field of hyperspectral image processing, it is designed by the model labeled samples obtained by learning to classify unlabeled surface features. Since the acquisition difficult and costly manual tagging, the number of labeled samples typically very limited. Labeled sample under limited circumstances, the performance of the algorithm hyperspectral image classification model is usually poor. How limited in sample mark or no mark of the sample, the realization terrain classification is a challenging task. This paper discusses the adaptation domain (Domain Adaptation, DA) techniques, using known markers or marker region hyperspectral image data to classify unlabeled, labeled so as to reduce the dependence on sample has very important practical significance. <br>This paper introduces the related alignment (Correlation alignment, CORAL) domain adaptive algorithm, discusses the covariance its presence in the region to adapt hyperspectral image classification and estimation; for the small sample covariance estimation instability problem, a species associated sparse matrix transformation (sparse matrix transform, SMT) alignment algorithm (CORAL-SMT) based on the estimated covariance matrix of the source domain and the target domain to obtain stable and accurate estimate of the covariance. Specifically, in CORAL-SMT, the covariance matrix of the source and target domains are decomposed to have a constrained feature, this feature can be represented as a series decomposed product form Givens (the Givens) rotation. In the maximum likelihood estimation framework, greedy strategy can effectively minimize the estimated covariance matrix, and can secure the estimated covariance matrix is <br>positive definite. <br>In the final experimental part herein Pavia of the hyperspectral image data of the proposed algorithm in high quiz port Yellow No. 5 made hyperspectral image and The City, and the use of the overall classification accuracy (Overall Accuracy, OA) and κ coefficient as the evaluation criteria to measure the performance of the algorithm. Experimental results show that the proposed algorithm has good performance on a test image.

Hyperspectral image classification is a research hotspot in the field of hyperspectral image processing, which aims to classify unmarked objects by the model learned on the marked sample. The number of marked samples is generally very limited due to difficulty in collecting and the high cost of manual marking. The performance of the hyperspectral image classification algorithm model is usually poor when the labeling sample is limited. How to achieve the classification of matter in the case of limited or even unlabeled samples is a challenging subject. This article explores the practical significance of Domain Adaptation (Domain Adaptation, DA) technology to classify unmarked hyperspectral data using known marker areas or marker images, thereby reducing reliance on marker samples.<br> This paper first introduces the correlation alignment (CORRELATION, CORAL) domain adaptation algorithm, discusses the problem of covariance estimation in hyperspectral image domain adaptation and classification, and proposes a sparse matrix transformation based on the instability of covariance estimation in the case of small samples. The Association Alignment Algorithm (CORAL-SMT) of Sparse Matrix Transform, SMT, estimates the covariance matrix of the source and target domains for accurate and stable covariance estimates. Specifically, in CORAL-SMT, the covariance matrix of the source and target domains is constrained to have a feature decomposition that can be expressed as a product form of a series of Givens rotations. In the framework of a very similar estimate, the covariance matrix can be effectively estimated using the greedy minimization strategy, and the estimated covariance matrix can be guaranteed to be<br>Positive.<br> In the final experimental part, the proposed algorithm is tested on the high-spectrum image 5 of the Yellow River estuary and the City of Pavia hyperspectral image data, and the overall classification accuracy (Overall Accuracy, OA) and the coefficient are used as evaluation criteria to measure the performance of the algorithm. The experimental results show that the algorithm proposed in this paper has good performance in the test image.

Hyperspectral image classification is a research hotspot in the field of hyperspectral image processing. It aims to achieve the classification of unlabeled objects through the model learned from the labeled samples. Due to the difficulty of collection and the high cost of manual labeling, the number of labeled samples is generally very limited. In the case of limited labeled samples, the performance of hyperspectral image classification algorithm model is usually poor. It is a challenging task to realize the classification of ground objects in the case of limited or even no labeled samples. In this paper, domain adaptation (DA) technology is discussed. It is of great practical significance to classify unlabeled hyperspectral data by using known labeled regions or labeled images, so as to reduce the dependence on labeled samples.<br>In this paper, we first introduce the correlation alignment (Coral) domain adaptation algorithm, and discuss its covariance estimation problems in hyperspectral image domain adaptation and classification. Aiming at the instability of covariance estimation in the case of small samples, we propose a sparse matrix transform, The correlation alignment algorithm (coral-smt) of SMT estimates the covariance matrix of source domain and target domain, and obtains the accurate and stable covariance estimates. Specifically, in coral-smt, the covariance matrix of source domain and target domain is constrained to have a feature decomposition, which can be expressed as a product of a series of givens rotations. Under the framework of maximum likelihood estimation, the covariance matrix can be effectively estimated by greedy minimization strategy, and the estimated covariance matrix can be guaranteed to be<br>Positive definite.<br>In the last part of the experiment, this paper tests the proposed algorithm on the data of gaofen-5 hyperspectral image and the city of Pavia hyperspectral image, and uses the overall accuracy (OA) and kappa coefficient as the evaluation criteria to measure the performance of the algorithm. The experimental results show that the algorithm proposed in this paper has good performance in the test image.