Data acquisition in forensics science must be performed in a fast and an efficient way, so that the data acquired is maximized at the same time that disturbance and time on the scene are minimized. For this reason, the use of indoor mapping systems appears as a key solution, in contrast with static systems, either laser or photogrammetry based, in which representing big and complex scenes requires acquisition from a high number of positions, and long-time dedication for data processing. This paper presents a methodology for the segmentation of point clouds acquired with a mobile indoor mapping system, and their conversion to 3-D models in CAD format, based on parameterized geometric elements from the scene. This way, all the information required in forensic sciences is stored in an adequate digital format, enabling its availability in the future, and minimizing time dedication in both data acquisition and processing steps.