Light source condition is an important factor affecting image quality in active illumination polarization imaging system. It is of practical significance to study the variation trend of polarization degree contrast between target and water surface background under different light source conditions for water surface target recognition and classification using active illumination polarization imaging technology. In this paper, polarization imaging experiments of iron, aluminum, wood and foam are carried out under water background, and the variation trend of polarization contrast between target and water background is studied under different incident zenith Angle, different polarization states and different light source of visible light band. The results show that the polarization contrast between targets of different materials and water surface background has different sensitive range to the change of incident zenith Angle, and the appropriate incident zenith Angle can be selected according to the actual demand in water surface target recognition and classification. The smooth surface material is more sensitive to the change of incident light band than the rough surface material, and the depolarization ability of the rough surface target to the single-band polarized light is weaker than that of the all-band visible polarized light, which is more conducive to the observation of the surface characteristics of the rough target. Compared with 0° and 90° linearly polarized light, linearly polarized light with incident light of ±45° has a greater polarization contrast and is more conducive to target recognition under water background.

No abstract

With the wide application of image fusion technology in target detection and other fields, the fusion of polarization images and other intensity images is becoming a research focus. Traditional polarization image fusion includes intensity, degree of linear polarization (DOLP), and angle of polarization (AOP). However, images of DOLP and AOP fusion cannot meet the requirements of outstanding positive characteristics. Therefore, we propose a method to calculate the polarization characteristics image that can reflect the difference of polarization characteristics of different materials. The method and process are as follows: First, the polarization detection angle is divided into several angle intervals, and the orthogonal difference characteristics (ODC) image of each interval is obtained by weighting and accumulating the AOP probability density of the angle in the interval and the correlation between images. Second, the ODC images are reconstructed in the gradient domain, and the multi-angle orthogonal differential polarization characteristics (MODPC) image is obtained. The MODPC image is fused with the visible intensity image, and the fusion results are evaluated by using image evaluation indexes such as contrast (C), average gradient (AG), image entropy (E), and peak signal-to-noise ratio (PSNR). The experimental results show that the MODPC and S 0 fusion result images are superior to the DOLP and S 0 fusion results in terms of subjective visual perception and objective indicators among the six classical fusion algorithms. The proposed MODPC image can be applied in target detection.