Simulating Human Visual Perception in Nighttime Illumination

This paper presents an image-based algorithm for simulating the visual adaptation of the human visual system to various illuminations,especially in dark nighttime conditions.The human visual system exhibits different characteristics depending on the illumination intensity,with photopic vision in bright conditions,scotopic vision in dark conditions,and mesopic vision between these two.A computational model is designed to simulate multiple features of mesopic vision and scotopic vision,including the chromaticity change,luminance change,and visual acuity loss.The system uses a source image under bright illumination as input.Then assuming that the viewer has already adapted to the new conditions,the color spectrum of the input image is reconstructed to replace the source with modifications of the chromaticity and the luminance of the relighted scene.A bilateral filter is used to simulate the visual acuity loss.The model parameters have clear physical meanings and can be obtained from experimental data to achieve realistic results.The algorithm can be used not only for visual perception simulation,but also as a day-for-night tool to produce realistic nighttime images from daytime images.