Robust Subspace Clustering with Block Diagonal Representation for Noisy Image Datasets

As a relatively advanced method, the subspace clustering algorithm by block diagonal representation (BDR) will be competent in performing on dataset if is assumed to noise-free and drawn from union of independent linear subspaces. Unfortunately, this assumption far reality, since real data are usually corrupted various noises subspaces overlap with each other, performance algorithms, including BDR, degrades complex data. To solve problem, we design new objective function based which l2,1 norm reconstruction error introduced model improve robustness algorithm. After optimizing function, present corresponding pursue self-expressive coefficient matrix structure for noisy dataset. An affinity constructed matrix, then fed spectral obtain final results. Experiments several artificial image datasets show that proposed has better than compared algorithms.