Forensic Odontology is the discipline of identifying, based on the recognition of certain features present in each person's dental structure. This area of forensic identification plays a major role in a natural or artificial disaster which has resulted in multiple deaths that are not possible to be identified by conventional methods like fingerprints. Forensic Odontology shall identify individuals based on dental records, mainly through their dental radiographs. In this paper we develop a biometric system for semi-automatic processing and matching of dental images, with the objective of identifying people. As a dental record, usually a post mortem dental radiography (PM), it will search the database, dental x-rays of the person before the accident (AM), consistent with dental X-rays of the autopsy (PM). To achieve the identification of the person, use a semi-automatic method to extract the contour of the teeth of the radiographs AM and P.M. to able to compare them, while also compare the separation of the contours of the teeth, x-rays taken earlier AM and PM. This paper presents a practical application of GRASP metaheuristic with double relaxation technique with double relaxation to solve the pattern comparison problem between the graphics approaches of dental records AM and PM.