Noise Characterization

Dictionary Learning with Structured Noise

We propose a novel dictionary learning with structured noise (DLSN) method for handling noisy data. We decompose the original data into three parts: clean data, structured noise, and Gaussian noise, and then characterize them separately.