Effect of imaging geometry and noise model on polarimetric contrast optimization

We study the effect of the noise model and imaging geometry on contrast optimization with an active polarimetric imager. Considering scenes with various polarization contrast sources and fluctuation statistics of polarimetric properties, we show that to maximize the range of scene orientation angles for which contrast is sufficient, one has to take into account not only the polarization difference between the target and background but also the variance of fluctuations in the image. In particular, for scenes exhibiting depolarization contrast, the best strategy is often to set the illumination and analysis polarization states orthogonal to each other in order to reduce the variance of the image. These results are important for designing and optimizing remote sensing or robotics-oriented polarization imagers.