Polarimetric Imaging: System Architectures and Trade-Offs

Polarimetric sensing provides additional scene information which can be used to enhance the target detection and recognition performance of an imaging system. Such information is particularly valuable in the discrimination of weak target signatures from their surroundings and, as such, is attracting a growing interest for future military and surveillance applications. However, the extraction of polarisation information adds complexity in terms of the physical sensor design and the required data processing. Imaging polarimeters typically require four images to be captured of the same scene in different polarisation orientations through either time or spatial division techniques. These sensor architectures introduce system performance constraints in terms of temporal and spatial resolution as well as the attendant degradations associated with the use of additional optical components. Issues associated with the physical envelope, operational robustness, and cost must also be considered. In terms of the processing of the polarimetric data, accurate registration and calibration is required to extract small polarisation signatures which are typically found in natural scenes. The polarimetric image data must then be processed using an image fusion or data fusion method which introduces further demands on the software design and system processor. For some applications, these limitations are acceptable relative to the polarimetric gain, whilst in others a conventional imaging sensor may offer a better overall solution. Consequently, it is important that a trade-off analysis is undertaken which evaluates the realistic performance gains with respect to the implementation and cost issues.