This talk will introduce some of our recent research progress in sparse representation based image reconstruction and face recognition. For image reconstruction, we will first introduce the adaptive sparse domain selection and regularization scheme, and then present the centralized sparse representation (CSR) model we recently developed. The CSR is very simple but very powerful, and it shows state-of-the-art results in image denoising, deblurring and super-resolution.

    Apart from image reconstruction, sparse representation has also been attracting lots of attention in face recognition, and it has shown many interesting results.  However, is it really the sparsity that helps face recognition? In this talk, we will demonstrate that it is not the sparse representation, but the collaborative representation, that truly helps face recognition.