In this paper, we propose a suspected criminal face identification on the watch list for video surveillance system via SVDD and SRC. Especially, the proposed criminal identification module is designed in a hierarchical manner via a mixture of support vector data description (SVDD) and sparse representation classifier (SRC). It has the following characteristics (system design requirements): 1) Since SVDD quickly recognizes only criminal suspects, it does not perform the unnecessary operation for the ordinary person; 2) Even in an inhibitory environment against face identification, it ensures a reliable and accurate identification performance via SRC that has been already proven as an excellent robust methodology for a face recognition; 3) Taking majority voting strategy with the repeated use of the same biological characteristics ensures the reliability of the system performance; 4) With the intrinsic incremental learning capability of SRC, this system can actively adapts itself according to the change of a watch list database. We carried out a feasibility experiment on the proposed prototype system with our KUFD (Korea University Face Database) collected from real experiments on Campus.