Super-wide-field-of-view long-wave infrared gaze polarization imaging embedded in a multi-strategy detail feature extraction and fusion network

This is the authors' abstract. We don't add key points for this paper.

Recently, infrared polarization imaging technology has become a research hotspot due to its ability to better resolve the physicochemical properties of objects and significantly enhance the target characteristics. However, the traditional infrared polarization imaging is limited to similar imaging mechanism restrictions, and it is difficult to acquire the polarization information of a wide-area posture in real time. Therefore, we report a combination of hardware and software for super-wide-field-of-view long-wave infrared gaze polarization imaging technology. Utilizing the non-similar imaging theory and adopting the inter-lens coupling holographic line-grid infrared polarization device scheme, we designed the infrared gazing polarized lens with a field-of-view of over 160°. Based on the fusion of infrared intensity images and infrared polarization images, a multi-strategy detail feature extraction and fusion network is constructed. Super-wide-field-of-view (150°×120°), large face array (1040×830), detail-rich infrared fusion images are acquired during the test. We have accomplished the tasks of vehicle target detection and infrared camouflage target recognition efficiently using the fusion images, and verified the superiority of recognizing far-field targets. Our implementation should enable and empower applications in machine vision, intelligent driving, and target detection under complex environments.