Who dunnit? An appraisal of two people matching techniques

The ability to compare a vague description of a person with a known dataset of people is an invaluable tool for various security related applications. This notion is built upon to allow the fusion of various descriptions of the same person to create a suspect description, which is then compared against the dataset of known people. Two different methods have been built that are based upon the various extensions of the transferable belief model to perform the fusion. Various methods are performed to overcome some of the issues of fusing within the transferable belief model in real world applications. Our results compare and contrast between the two methods show the effectiveness of each method of being able to extract the ground truth person, as described by various (real) vague descriptions from a dataset of people.