Abstract
This paper proposes a whale optimized subspace matching pursuit algorithm to improve the precision of sparse unmixing of hyperspectral data. The proposed algorithm modifies the traditional whale group optimization algorithm by introducing the nonlinear population control parameters and evolution strategy and thus improves the convergence speed and convergence precision of the whale cluster optimization algorithm. Based on the improved subspace matching pursuit algorithm, the new algorithm uses the whale optimization algorithm to solve the abundance coefficients of the known end members, where the objective function is modeled as a constrained sparse regression. In order to improve the precision of the subspace matching pursuit algorithm, the minimum reconstruction error is taken as the criterion and the redundant end members with smaller coefficients are removed to the greatest extent. The simulation and real data experiments show that the proposed algorithm can effectively remove the redundant endmembers and has a higher precision of unmixing.
Introduction
Hyperspectral imaging captures detailed spectral information across hundreds of narrow bands, providing a unique "fingerprint" for materials. However, due to the limited spatial resolution of sensors, a single pixel often covers multiple materials (e.g., soil, water, and vegetation), resulting in a mixed pixel.
Hyperspectral Unmixing (HU) is the process of decomposing these mixed pixels into a collection of pure constituent spectra (endmembers) and their corresponding proportions (abundances). This is fundamentally a sparse regression problem: while a spectral library may contain hundreds of potential minerals, any single pixel typically contains only a few. Existing greedy algorithms like Subspace Matching Pursuit (SMP) often struggle with local optima or fail to prune redundant endmembers, leading to inaccurate results where materials are incorrectly identified as being present in trace amounts.
Methodology: The IWOSMP Framework
The authors propose a multi-stage approach called Improved Whale Optimized Subspace Matching Pursuit (IWOSMP) to address these challenges.
1. Improved Whale Optimization (SEWOA)
The standard Whale Optimization Algorithm (WOA) mimics the bubble-net hunting behavior of humpback whales. The authors improved it into the Strategy-based Evolutionary Whale Optimization Algorithm (SEWOA) by introducing two key modifications:
- Nonlinear Population Control: Replacing the standard linear convergence factor with a cosine-based nonlinear parameter to better balance exploration (global search) and exploitation (local refinement).
- Evolution Strategy: Introducing a mutation-like strategy to help the algorithm escape local optima.
2. Subspace Matching Pursuit Integration
The algorithm uses the SMP framework to select a candidate set of endmembers from a large spectral library. SEWOA is then employed to solve the constrained sparse regression problem for abundance estimation, ensuring that abundances are non-negative (ANC) and sum to one (ASC).
3. Redundancy Pruning
To further improve precision, the algorithm uses a minimum reconstruction error criterion to aggressively remove redundant endmembers with very small coefficients. This step ensures that only the most significant materials are retained in the final unmixing result.
Experimental Results
The researchers evaluated IWOSMP using both simulated and real-world datasets:
- Simulated Data: Using the USGS spectral library, the algorithm was tested against varying Signal-to-Noise Ratios (SNR) and different numbers of endmembers. IWOSMP consistently achieved lower Root Mean Square Error (RMSE) than standard SMP and popular methods like SUnSAL-TV.
- Real-World Data: On the classic AVIRIS Cuprite dataset, IWOSMP produced abundance maps that were visually sharper and exhibited higher sparsity (fewer false positives) than comparison algorithms. Its reconstruction error was approximately 15-30% lower than that of the standard SMP.
Conclusion
By hybridizing a greedy pursuit framework with a nonlinearly optimized meta-heuristic, the IWOSMP algorithm significantly improves the precision of hyperspectral unmixing. The introduction of SEWOA provides a robust optimization engine that excels in finding global optima in sparse regression tasks. Future research may explore integrating spatial information to further enhance performance in noisy environments.
Summary updated