Minimum-Volume Non-Negative Block-Term Decomposition: Blind Data Fusion and Unmixing with Estimation of the Number of Endmembers

This paper introduces a family of efficient algorithms for joint super-resolution and unmixing in remote sensing. They utilize coupled tensor decompositions with minimum-volume regularizations and incorporate the $\beta$ -divergence to handle diverse noise statistics, while estimating degradation operators mapping the unknown super-resolution image to the degraded observations. They aid in identifying factors in the tensor decomposition and managing rank deficiencies effectively. These algorithms show competitive performance compared to existing methods, even in challenging scenarios: unknown degra-dation operators and estimating the number of endmembers.