The influence of visual complexity on the detection of targets investigated by computer generated images

摘要

In urban areas, assessing the visibility of the visual cues used by motorists to make driving decisions poses a number of methodological problems. To investigate these, the authors decided to conduct laboratory experiments to investigate the detection of pedestrians at night-time. The geometric and photometric characteristics of the computer-generated images used were derived from in situ measurements. A method for characterizing the visual complexity of the near background of a target was developed. The study has demonstrated that, when the visual complexity of the near background of an target increases, its visibility level must be increased by between 10 and 25 to achieve a detection rate of 90%. These values must be increased to 20 and 35, respectively in order to achieve a detection rate approaching 100%.