Image reconstruction of compressive sensing using digital signal processing (DSP)

Compressive sensing has opened up a new path to reconstruct images from a number of samples which is far smaller than the desired resolution of the images. A novel single pixel camera system has been developed based on this theory, and it can reconstruct images with a single optical sensing element. However, the development reminds challenge because the image recovery process is complex. For example, large amount of data is computed during the image recovery process and this make real-time processing of image become difficult. In order to improve the recovery speed, Orthogonal Matching Pursuit (OMP) algorithm for image recovery of compressive sensing are designed. However, the general OMP algorithm does not fit for digital signal processing (DSP) platform because it requires large memory allocation. In this paper, the modified OMP algorithm is introduced to improve calculation time with less memory requirement. The implementation of the algorithm on DSP device is presented and verified. Besides, the integration of the designed embedded hardware structure also increases the portability of the camera system.