CS mainstream research direction
(1) spare reconstruction:
usage: reconstruct spasre signal S from a small mumber of measurements .
problem:Given measurement B and a measurement matrix A find the signal matrix S such that B=AS where S is a sparse matrix.
Main algorithm: Convex Optimization,Iterative Greedy Algorithm,Focuss Algorithm
1) Convex Optimization:BP(Basis Pursuit), Gradient Projection Method….
2) Iterative Greedy Algorithm: MP(Matching Pursuit), OMP(Orthogonal Matching Pursuit), cosaMP(Compressivesampling Matching Pursuit)….
3) FOCUSS Algorithm(Focal underdetermined system solver).....
(2) dictionary learning:
usage:design/learn adictionary adaptively to better fit the model and achieve signal spare representation.
problem:Given a signal matrix B find a sparse matrix S and a dictionary D such that B=DS.
(3) blind compressed sensing
usage:1) reconstruct signal from a small mumber of measurments without any prior for X(blind cs),this problem is impossible to solve.
2) reconstruct signal from a small mumber of measurments where signal are sparse under somebasis with additional constraints (half-blind cs).this problem is possible to solve.
Problem:Given measurements B and a measurement matrix A find the signal matrix X such that B=AX where X=PS for some basis P and k-spare matrix S.