Search for: All records

When imaging underwater scenes from above the water surface, the reflection from the air–water interface creates an obscuring background that varies with illumination and viewing angles. It is well known that the reflected light is horizontally polarized, and using a vertically transmitting polarizing filter is a common technique to improve the contrast of underwater scenes. However, to our knowledge, no quantitative measurements of polarization-enabled contrast enhancement have been reported in the literature. In this work, panchromatic and RGB division-of-focal-plane polarization cameras were used to record images of black and white tiles submerged in water for determining contrast as a function of viewing angle, both without a polarizer and with a vertical polarizer. Experiments were conducted in two outdoor locations and in a black tub indoors with controlled color and brightness of the reflected background. The maximum contrast through a vertical polarizer occurred near the Brewster angle, but the amount of contrast enhancement (the ratio of contrast through a polarizer to contrast without a polarizer) was found to increase until much larger angles. Also, the observed changes in contrast resulting from changing properties of the reflected background were consistent with the Fresnel reflection coefficients. 

Spectral characterizations are performed on imagers to obtain a relative spectral response (RSR) curve. This process often utilizes a grating monochromator with an output that changes polarization as a function of wavelength (our monochromator’s degree of linear polarization was found to vary from less than 10% to more than 70%). When characterizing a polarization-sensitive imager, this introduces polarization artifacts into the RSR curve. We present a simple method to avoid these polarization artifacts for division-of-focal-plane polarization imagers by directly illuminating the camera with the monochromator output and calculating the S₀Stokes parameter at each super pixel, then we show consistent results from this method for two division-of-focal-plane polarization imagers. We also show that ignoring the monochromator polarization results in order-of-magnitude RSR errors. The recommended method uses an iris to limit the spatial extent of the monochromator output, which was found experimentally to increase the minimum signal-to-noise ratio by more than a factor of 2.