Contrast optimization in broadband passive polarimetric imaging based on color camera

Polarization angle information enhancement method based on polarimetric array imaging

Polarization imaging, based on the measurement of polarization parameters containing specific physical information, has found extensive applications across various domains. Among these parameters, polarization angle information plays a crucial role in revealing texture details. However, in practical scenarios, noise during image acquisition can lead to significant degradation of polarization angle information. To address this issue, we introduce a novel, to the best of our knowledge, polarization angle information enhancement method based on polarimetric array imaging. Our proposed method utilizes the principles of polarimetric array imaging to effectively restore texture information embedded within polarization angle images. Through the deployment of a self-designed polarimetric array imaging system, we conducted experiments in diverse scenes to validate the efficacy of our approach. The acquired polarization angle data were subjected to our method for enhancement. The experimental outcomes distinctly illustrate the noise suppression capabilities of our method, showcasing its ability to faithfully reconstruct intricate details obscured by substantial noise interference.

BM3D-based denoising method for color polarization filter array

Color split-focal plane polarization imaging systems are composed of image sensors with a color polarization filter array (CPFA). The noise generated during image acquisition leads to incorrect estimation of the color polarization information. Therefore, it is necessary to denoise CPFA image data. In this study, we propose a CPFA block-matching and 3D filtering (CPFA-BM3D) algorithm for CPFA image data. The algorithm makes full use of the correlation between different polarization channels and different color channels, restricts the grouping of similar 2D image blocks to form 3D blocks, and attenuates Gaussian noise in the transform domain. We evaluate the denoising performance of the proposed algorithm using simulated and real CPFA images. Experimental results show that the proposed method significantly suppresses noise while preserving the image details and polarization information. Its peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) indicators are superior to those of the other existing methods. The mean values of the PSNR and SSIM of the degree of linear polarization (DoLP) color images calculated through CPFA image interpolation can be increased to 200% and 400%, respectively, by denoising with the proposed method.

Learning-based denoising for polarimetric images

Based on measuring the polarimetric parameters which contain specific physical information, polarimetric imaging has been widely applied to various fields. However, in practice, the noise during image acquisition could lead to the output of noisy polarimetric images. In this paper, we propose, for the first time to our knowledge, a learning-based method for polarimetric image denoising. This method is based on the residual dense network and can significantly suppress the noise in polarimetric images. The experimental results show that the proposed method has an evident performance on the noise suppression and outperforms other existing methods. Especially for the images of the degree of polarization and the angle of polarization, which are quite sensitive to the noise, the proposed learning-based method can well reconstruct the details flooded in strong noise.