Research on reassembling fragments of documentary evidence

A number of documentary evidence pieces are often left in a crime scene, and the key to effectively identify suspects is to collect and restore fragment evidence in time. Obviously, some fragments show strong similarities in paper texture, geometric landmarks, and document content. After much thought and consideration, a solution to reassemble image fragments was proposed and implemented based on shape matching. Firstly, edges are recognized by fully exploiting the potential of Canny algorithm, and the most outer contour is intelligently inferred. Secondly, the contour was transformed into a series of vertices which can represent a polygon. A similarity is then computed based on the geometric landmarks between polygons, and a best matching fragment is elected. Lastly, the original documentary evidence is reassembled step by step in order of priority. The software developed here can increase efficiency in forensic investigations for police.