It is always easy to recognize a person from the image of his face which is taken in constrained situations and with the cooperation of the subjects. For example from the photos of a person taken for his driving license or passport will immediately make him recognizable. But the problem comes when the subject is uncooperative and the scene is unconstrained as in the case of a criminal or a thief for whom there is a look out by the investigation agencies. Recognizing a face from video surveillance throws number of challenges like variation in illumination, resolution problems, back ground muddle, facial pose and expression on it. In order to identify a person the investigating agency has to utilize all known sources of the person as the quality of the images and videos will be poor. Sometimes the onlookers of the crime may capture the pictures or videos of the person responsible for the act on their mobile phones which will be a useful source for recognition. From the verbal description of the person of interest specific information can be gathered like the gender of the person, his probable age, how he looks like, color of the hair, body build, height and his race etc. Face sketches can be drawn by the experts from such description which will be of great helpIn traditional face matching techniques a single media is used. Generally it uses a still face image, a video track or a face sketch as input. In this paper utilizing entire gamut of data from multiple media is discussed for recognizing a person of interest. Due to this type of approach of utilizing different fusion schemes, 3D face models, incorporation of quality measures for fusion, and video selection the accuracy of identifying the person of interest will improve many folds.