Matrix Cofactorization for Joint Unmixing and Classification of Hyperspectral Images
read more
Citations
References
Assessing the accuracy of remotely sensed data : principles and practices
Random forest in remote sensing: A review of applications and future directions
Vertex component analysis: a fast algorithm to unmix hyperspectral data
Hyperspectral Unmixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches
Related Papers (5)
Hyperspectral data unmixing using constrained semi-NMF and PCA transform
Frequently Asked Questions (13)
Q2. What are the future works in this paper?
M. Belgiu and L. Drăguţ, “ Random forest in remote sensing: A review of applications and future directions, ” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 114, pp. 24–31, 2016. [ 3 ]
Q3. What is the classification rule for a linear classifier?
Considering a linear classifier parametrized by the matrix Q ∈ RC×K , a vector-wise nonlinear mapping φ(·), such as a sigmoid or a softmax operator, is then applied to the output of the classifier.
Q4. What are the two conventional metrics used to evaluate the classification accuracy?
To evaluate the classification accuracy, two conventional metrics are used, namely Cohen’s kappa coefficient and the averaged F1-score over all classes [18].
Q5. What is the attribution term for the clustering term?
More precisely, the coupling term is expressed as a clustering term over the abundance vectors where the attribution vectors to the clusters are also the feature vectors of the classification as detailed in Section II-C.
Q6. What is the penalization of the kmeans clustering problem?
Two constraints are considered in this kmeans clustering problem: i) a positivity constraint on B since centroids are expected to be interpretable as mean abundance vectors and ii) the vectors zp (p ∈ P) are assumed to be defined on the K-dimensional probability simplex SK .
Q7. What is the definition of the unmixing method?
It should be noted that all unmixing methods use directly the correct endmember matrix M which has been used to generate the data.
Q8. What is the purpose of this paper?
This paper introduces a unified framework to perform jointly spectral unmixing and classification by the mean of a cofactorization problem.
Q9. What is the index subset of unlabeled pixel?
The index subset of labeled pixel is denoted hereafter L while the index subset of unlabeled pixel is U ( L ∩ U = ∅ and L ∪ U = P).
Q10. What is the RMSE for the unmixing?
To evaluate the unmixing results quantitatively, the reconstruction error (RE) and root global mean squared error (RMSE) are considered, i.e.,RE =√1PL∥ ∥ ∥ Y −MÂ ∥ ∥ ∥ 2F ,RMSE =√1PR∥ ∥ ∥ Atrue −
Q11. What is the definition of a global cofactorization problem?
To define a global cofactorization problem, a relation is drawn between the activation matrices of the two factorization problems, namely the abundance matrix and the feature matrix.
Q12. What is the cost function for a classifier?
The weighing coefficients dp adjust the cost function with respect to the sizes of the training and test sets, in particular in the case of unbalanced classes.
Q13. What are the results of the proposed cofactorization framework?
Results reported in Table The authorshow that the proposed cofactorization framework outperforms both RF and D-KSVD in term of classification.