Crowdedness estimation in public pedestrian space for pedestrian guidance

A degree of crowdedness in public pedestrian space is obtained from sequences of greyscale images and still colour images in order to provide a route selection hint to pedestrians. Two measures are employed: amount of motion and colour entropy, and they are used selectively depending on the nature of target space. The pedestrian space is categorised into two: smooth and stagnated. The smooth space urges pedestrians to flow smoothly without pausing, while the stagnated space invites pedestrians to pause for socialising. A degree of stagnation, obtained from accumulated differentiated images, is used to distinguish these two types of space. The amount of motion and the colour entropy are compared against human´s sense of crowdedness for evaluation. The experiments have shown that the amount of motion represents the smooth space´s crowdedness well, while the colour entropy is suitable for evaluating the stagnated space. It has also shown tendency of significance between the degree of stagnation and the types of space.