|
Linear Spectral Unmixing (LSU) is a widely used technique in the field of Remote Sensing (RS) for the estimation of fractional abundances of endmembers and their spectral signatures. Large data size, poor spatial resolution, non-availability of pure endmember signatures in data set, mixing of materials at various scales and variability in spectral signature make LSU a challenging and inverse-ill posed task. Broadly there are three basic approaches to manage the LSU problem: geometrical, statistical and sparse regression. First and second approaches are types of Blind Source Separation (BSS). The third approach assumes the availability of some standard publicly available spectral libraries, which contains signatures of many materials measured on the Earth's surface using advanced spectro radiometer. In sparse regression approach, the problem of LSU is simplified to finding the optimal subset of spectral signatures from the library known in advance. In this paper, the concept of collaborative sparse regression is incorporated to improve the performance of existing SUnSAL-TV algorithm. SUnSAL-TV is a recently proposed Total Variation (TV) spatial regularisation-based approach. Our simulation results conducted for standard and publicly available synthetic fractal data set show 10% to 15% performance improvement in Signal to Reconstruction Error (SRE) for different data cubes. Simulation is also performed for a subset of real cuprite data cubes and compared with the outcome of recent algorithms.
|
