Wavelet and Gradient Fusion Demosaicing Algorithm for Polarized Image Sensors

Polarization is a basic physical property of light, which plays an active role in artefact recognition and contrast enhancement. The division of focal plane (DoFP) polarized Image sensors have the ability of real-time detection, but its inherent focal plane structure inevitably leads to the reduction of spatial resolution. Many proposed demosaicing algorithms pay more attention to the reconstruction quality of intensity imaging, but ignore the sensitivity of polarization detection to edge and mutation information. In this paper, a demosaicing fusion algorithm is proposed, which uses gradient low-pass filter to reduce the trace of artificial interpolation and wavelet transform to preserve the polarization sensitive information such as edges. The results of ground and flight tests show that the proposed algorithm can improve the image resolution, maintain the polarization-sensitive information and meet the requirement of real-time polarization information acquisition under fast maneuvering state. It has a positive effect on the popularization and application of DoFP polarization sensors.