Search Space Reduction in Pedestrian Detection for Driver Assistance System Based on Projective Geometry

Conference paper

DOI: 10.1007/978-3-642-35473-1_27

Part of the Smart Innovation, Systems and Technologies book series (SIST, volume 21)
Cite this paper as:
Dimza K., Su TF., Lai SH. (2013) Search Space Reduction in Pedestrian Detection for Driver Assistance System Based on Projective Geometry. In: Pan JS., Yang CN., Lin CC. (eds) Advances in Intelligent Systems and Applications - Volume 2. Smart Innovation, Systems and Technologies, vol 21. Springer, Berlin, Heidelberg


Vehicles are equipped with smarter and smarter driver assistance systems to improve driving safety year by year. On-board pedestrian detection system is a critical and challenging task for driving safety improvement because driving environment is very dynamic, where humans appear in wide varieties of clothing, illumination, size, speed and distance from the vehicle. Most of existing methods are based on the sliding window search methodology to localize humans in an image. The easiest and also the most popular way is to check the whole image at all possible scales. However, such methods usually produces large number of false positives and are computationally expensive because large number of inappropriate regions were checked. In this paper, we develop a method which reduce the search space in pedestrian detection by using properties of projective geometry in the case when camera parameters are unavailable. The simple user interaction with stochastic optimization is used to estimate projective parameters. We showed the efficiency of our method on public dataset with known camera parameters and self captured dataset without registered camera parameters. Experiment results show that the effectiveness of the proposed method is superior compared to the traditional uniform sliding window selection strategy.


Pedestrian Detection ADAS Vanishing Points 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  1. 1.Department of Computer ScienceNational Tsing Hua UniversityHsinchuTaiwan

Personalised recommendations