In light transmission processing, light is effected by the fog/haze/smoke/dust environment usually. Thus reduce the capability of traditional photo-electric instrument in detection, transmission, imaging, because of the attenuation, scattering, absorbing of the atmosphere. Polarization transmission characteristics exist the potential possibility in observing object through fog and smoke, increasing observation distance, improve the environment adaptability of the photo-electric instrument. The influence of polarization transmission characteristics are analyzed by environment testing experiment in this paper, including light intensity, degree of polarization, angle of polarization etc. We research smoke and fog environment simulation technology, builds the process-controlled and parameter-measured indoor equipment to simulate smoke/fog environment. The experiments of the polarization characteristic are tested in two kinds of transmission mediums separately, which are water fog and oil smoke. The attenuation ratio is measured in different kinds of smoke concentrations. Polarization imaging experiment are tested separately in different kinds of illuminance lights, which are daylight lamp, LED lights and projector lights. The images are captured by imaging CCD with linearly polarized filters on degree 0, 60, 120. The polarization images on three directions are processed by image infuse algorithm. The experiment results are provided as the theory foundation and data reference in field of polarization transmission.
An improved image fusion algorithm based on the NSCT is proposed in this paper. After decomposition NSCT method of multi-scale and multiple directions, polarization image was decomposed into two parts: low frequency sub-band and high frequency band-pass images. The fusion strategy of combining local regional energy and gradient structure similarity were used in low-frequency coefficients. While in the high-frequency band-pass coefficients part, the fusion strategy of the location spatial frequency as the correlation coefficient was used. The intensity image and polarization degree image are fused for improving the sharpness and contrast of the image. The experiments show that the algorithm is effective to improve the imaging quality in the turbid medium.