An assessment method for landmark recognition time in real scenes

This study assessed landmark recognition time by investigating response time of participants in a visual search task. A simple predictive model was proposed to quantitatively estimate recognition time according to landmark features. Results revealed that size, shape and text (e.g., on billboard) features attracted attention of participants and led to shorter recognition time. Mean error of the predictive model was 0.42 s, which approximately described the time or cognitive burden needed for landmark recognition. These results improve our understanding of the relationships between landmark features and recognition efficiency, and facilitate the weighting of these features in landmark identification or selection methods. (c) 2014 Elsevier Ltd. All rights reserved.