Optimized image fusion rule based on targeted image evaluation metrics for vehicle imaging systems

Vehicle assistance systems suffer from perception degradation in complex environments like haze. To address this, we develop an integrated visible polarization and infrared dual-band imaging system and propose an innovative image fusion rule based on target image evaluation metrics. Experimental results demonstrate that our method significantly enhances texture details and target discrimination in fused images. The proposed rule also shows strong versatility, supporting various fusion frameworks and multimodal imaging applications. This work provides valuable solutions for assisted driving and other fields like remote sensing and military reconnaissance.