An adaptive compressive sensing with side information

Compressive sensing measurements generally do not provide as much understandable information as traditional compressed domains. On the other hand, with the rise of compressive sensing image sensors, it becomes necessary to define relevant sensing schemes. Applying the same linear projection to reduced signal supports optimizes the reconstruction time and makes the sensing strategy more suitable. It often refers to block-based compressed sensing. This work proposes to add new statistical measurements to efficiently adapt the sensing strategy. Since these new measurements become practical on a real sensor it also improves the use of image compressive sensing by providing some useful features for other applications. In addition, those measurements significantly ameliorate the reconstruction quality.