Restricted Structural Random Matrix for compressive sensing

Compressive sensing (CS) is well-known for its unique functionalities of sensing, compressing, and security (i.e. equal importance CS measurements). However, there a tradeoff. Improving compressing efficiency with prior signal information tends to favour particular measurements, thus decreasing security. This work aimed improve the without compromising novel sampling matrix, named Restricted Structural Random Matrix (RSRM). RSRM unified advantages frame-based block-based together global smoothness low-resolution signals are highly correlated). acquired compressive measurements random projection multiple randomly sub-sampled signals, which was restricted (equal in energy), thereby observations equally important. proven satisfy Isometry Property showed comparable reconstruction performance recent state-of-the-art deep learning-based methods. • Propose matrix that improves scarifying democracy. Combine partial sampling, multi-images super-resolution, coded imaging, sensing. Proposed satisfies competitive performance.