核心期刊网首页> 美国政府科技报告> AD

Registration of a Dynamic Multimodal Target Image Test Set for the Evaluation of Image Fusion Techniques


This report results from a contract tasking TNO Defence, Security and Safety as follows: The grantee will acquire and register a multimodal dynamic target image test set that can be used to assess the operational effectiveness of static and dynamic image fusion techniques for a range of relevant military tasks. TNO originally developed the TOD (Triangle Orientation Detection) method to predict human target detection and identification performance from static imagery. The TOD method produces a TOD-curve that resembles an MTF or MRTD (commonly used to evaluate thermal sensors). However, the TOD-method is more directly related to the target recognition task, and is better suited at capturing nonlinear effects of the kind that typically occur in image fusion methods, such as sampling and (local) contrast enhancement. In the proposed project grantee collected dynamic nighttime and daytime imagery using sensors with various spectral sensitivities. The sensors were thermal, SWIR, NIR, and visible wavelength cameras that will be mounted on a common base plate. The sensor suite as a whole will be placed in a vehicle, allowing one to register approach sequences to the targets. The targets were military relevant targets such as personnel or, if possible, military vehicles. Included with each target sequence will also be triangles with a range of different sizes, thermal and visual contrasts. This will allow TOD-measurements (characterizing object recognition performance) on the resulting set of multimodal images that can be compared with human task performance using the same imagery. At total of 16 runs were successfully registered, partly in dry and partly in rainy conditions. The imagery collected during this study, together with the GPS data and all other documentary material, was made available to the Air Force Research Laboratory organization at the end of the field trials. Researchers from both the Air Force Research Laboratory and TNO Human Factors will write a joint pa.......

【作者名称】: Toet, A.
【作者单位】: TNO Defence Security and Safety Soesterberg (Netherlands).
【关 键 词】: Multimode , Image registration , Target recognition , Target detection , Netherlands , Humans
【期刊论文数据库】: [DBS_Articles_01]
【期刊论文编号】: 104,448,742
【摘要长度】: 2,001
【上篇论文】: 美国政府科技报告 - Multi-Robot FastSLAM for Large Domains
【下篇论文】: 美国政府科技报告 - USAFA Cadet Homepage: Institute for Information Technology Applications (IITA)

【论文下载】: 免费获取 该期刊&论文全文内容