A novel method for designing compressed sensing matrix

Compressed Sensing (CS) is a method to restore sparse signals from a small number of observed components using a CS-matrix, and is usually applied to data compression. In previous studies, Random-Matrix, whose elements are randomly generated variables, is considered to be the most suitable form for CS-Matrix. However, it cannot ensure the availability of CS due to its randomness. Moreover, it has a problem with the reproducibility. To solve this problem, in this paper, we propose a novel generation method of CS-Matrix with randomness removed. Through numerous experiments with numerical value and images, we confirmed the reconstruction precision using proposal CS-matrix were higher than Random-Matrix.