Research on infrared polarization enhancement technology for space targets

Le résumé fourni par la source

With the increasing complexity of space exploration missions, traditional infrared radiation intensity imaging faces limitations in high-precision space target identification due to susceptibility to environmental interference and difficulties in distinguishing surface material properties. While infrared polarization imaging effectively characterizes surface structures and material attributes by extracting polarization information from target radiation, single-modal imaging suffers from insufficient complementary information. To address the challenge of material identification caused by similar reflective characteristics in photoelectric detection of space targets, this study proposes a multimodal fusion method integrating infrared intensity and polarization images. A multiscale structural decomposition strategy is employed, combining low-rank sparse decomposition (SVD) and Non-subsampled Shearlet Transform (NSST) to hierarchically fuse global thermal radiation features from infrared intensity images and local material details from polarization images. Lowfrequency components are fused via averaging, while high-frequency details are enhanced through weighted fusion rules. Experimental results demonstrate that fused images of typical satellite surface materials (tungsten, aluminum, titanium) achieve a peak signal-to-noise ratio (PSNR) of 67.54 dB and a structural similarity index (SSIM) of 0.998, significantly improving edge contrast and material discriminability compared to single intensity images. Notably, mean squared error (MSE) remains below 0.03 at 0° and 180° observation angles, validating the algorithm’s effectiveness in noise suppression and detail preservation. This study provides a viable technical pathway for high-precision identification of space targets. Future work will focus on optimizing algorithm generalization and exploring engineering applications in real-time dynamic scenarios.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.