Compressive Sensing and Recovery for Binary Images

The compressive sensing framework enables digital signal acquisition systems to take advantage of signal structures. However, non-sparse binary images don¡¯t share in the benefits to be applied by compressive sensing. Proposed is a new compressive sensing and recovery method for binary images. In the sensing step, the sensing matrix is specially designed instead of randomness. In the recovery step, L-infinity norm minimization scheme is used for binary dense representation to be recovered. The new approach is shown to be effective and useful with real binary images such as text and fingerprints.