Abstract
Abstract. During the last decade, modern Pedestrian Detection Systems made massive use of the steadily growing numbers of high-performance image acquisition sensors. Within our naturalistic driving environment, a lot of different and heterogeneous scenes occur that are caused by varying illumination and weather conditions. Unfortunately, current systems do not work properly under these hardened conditions. The aim of this article is to investigate and evaluate observed video scenes from an open source dataset by using various image features in order to create a basis for robust and more accurate object detection.
Highlights
Due to the increasing interest of research in the field of Vulnerable Road User (VRU) protection, a series of Pedestrian Detection Systems (PDS) using in-vehicle sensors has been developed within the past years
The results of our first experiment show that Moving Pictures Experts Group (MPEG)-7 features ranked first place in the cross-validation test, whereas Histograms of Oriented Gradient (HOG) features achieved the first rank in classification accuracy on the dedicated testset
The combination of HOG and MPEG-7 features reached the second place in both categories
Summary
Due to the increasing interest of research in the field of Vulnerable Road User (VRU) protection, a series of Pedestrian Detection Systems (PDS) using in-vehicle sensors has been developed within the past years. Most of the current systems use Histograms of Oriented Gradient (HOG) features, since they yield good results in the field of pedestrian detection, but fail completely in some situations. Many approaches use additional features like Haar Wavelets (Papageorgiou et al, 1998) or Local Binary Patterns (Harwood et al, 1995) in order to improve the detection and to compensate the weaknesses of single feature systems. In this contribution, we present a novel multi-feature approach which combines HOG and MPEG-7 features.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
