A longwave infrared (LWIR) handheld surveillance camera has been modified through the addition of a second sensor which provides both visible (RGB) and near-infrared (NIR) image streams. The challenges and constraints imposed on the development process of this Handheld Fusion Camera (HHFC) are described, and the approach to the dual and tri-band image fusion processing schemes is presented. The physical characteristics of the existing camera acted as a major constraint on the HHFC design with the Size, Weight, and Power (SWaP) requirements restricting the choice of both the additional sensor as well as the processor engine available within the camera. The primary use of the HHFC is in groundbased security and surveillance operations which is challenging in terms of variability in the scene content. Establishing an effective processing architecture is critical to both image interpretability by the user, and operational effectiveness. The HHFC allows the user to view different image streams including enhanced single-band image data as well as both dual and tri-band fused imagery. Such flexibility allows the user to select the best imagery for their immediate requirements. Power consumption and latency figures have been minimised by the use of relatively simple arithmetical fusion algorithms combined with an Adaptive Weight Map (AWM) for regional-based optimisation. In practice, the potential performance gain achieved is necessarily limited by the required performance robustness, and this trade-off was critical to the HHFC design and the final image processing solution.
|