Maintaining optimal vigilance in high-stakes operational environments requires effective integration of multimodal sensory inputs. This study investigates the neurocognitive mechanisms of audiovisual integration by combining computer-vision-based psychomotor vigilance tasks (PVT) with soundscape stimulation while monitoring 30 participants through synchronized EEG and behavioral recording. Our results demonstrate that Soundscape B significantly enhances multimodal perception, yielding a 12.3 ms improvement in PVT reaction times compared to 4.8 ms for Soundscape A ( \(p<0.01\) ), along with increased subjective relaxation ( \(p=0.0023\) ). Neurophysiological analysis reveals cross-modal enhancement effects through three key findings: (1) a 23% amplitude increase in frontal-central P300 components, indicating optimized attentional resource allocation during audiovisual integration; (2) characteristic spectral shifts featuring suppressed theta (4–7 Hz) and enhanced alpha/beta (8–30 Hz) power, consistent with alert cognitive states; and (3) strengthened functional connectivity between temporal and occipital cortices ( \(\varDelta \) clustering coefficient = 0.15, \(p<0.05\) ), suggesting improved cooperative perception across sensory modalities. These findings provide empirical evidence for bio-inspired multimodal fusion strategies in computer vision systems, with direct applications in driver monitoring, surgical robotics, and other scenarios requiring sustained visual attention. The proposed EEG-based evaluation framework further advances quantitative assessment of cross-modal interactions in perceptual systems.