Managing Aircraft Simulation Requirements with Content-Based Image Retrieval

Requirements analysis for aircraft simulators is often driven by photographs and videos of the actual aircraft. An engineer may gather and organize hundreds or even thousands of source photos of various instruments and devices unique to the aircraft. Managing all of this source information and referencing it to generate software requirements can be challenging and time-consuming. This paper explores Content Based Image Retrieval (CBIR) techniques to automatically process and search those images to generate basic requirements and to facilitate reuse. An unsupervised clustering algorithm groups source images based on minimal user input. Images processed in this way can also be queried by image similarity, thereby allowing project managers to find common source material among projects. The effectiveness of these techniques is demonstrated on an example cockpit.