Dynamic comparison of headlights

This web page contains additional material of the paper

Dynamic comparison of headlights.
J. Serrat¹, F. Diego¹, F. Lumbreras¹, J.M. Àlvarez¹, A. López¹ and C. Elvira².
Jourmal of Automobile Engineering, Proc. of the Institution of Mechanical Engineers, Part D.
¹ADAS Group, Computer Vision Center, Universitat Autònoma de Barcelona, Spain
²Electrical Engineering and Lighting Department, SEAT Technical Center, Martorell, Spain

Summary
Playing the video sequences
Headlights comparisons
Results

Summary

Headlights are an important active safety element of vehicles. As such, they are continuously improved by the automotive component manufacturers. This poses the problem of how to assess them. An important step in the evaluation of a headlights system consists in the actual experience of driving at night along a test track while observing their behaviour with regard features like the lateral and longitudinal reach of the low beams and the homogeneity of the light cast on the road surface. Dynamic headlights assessment is usually relative, that is, the performance of two headlights are compared. This comparison can be greatly facilitated by recording videos during the test and then synchronize and jointly visualize them. Moreover, it is possible, in addition, to spatially align the pairs of temporally corresponding frames so that a pixel level comparison can be performed. This web page contains the results of this work in the form of videos, since the still images in the manuscript can not convey the dynamic aspect of headlights comparison.

Playing the video sequences

All videos are compressed using the widespread DivXcodec, supported by most video players. The video format is unlikely to be a problem. However, we have observed that different players (like the official DivX player or the Windows Media Player) show differently the contrast and color intensites. The following link is a local copy of the Classic Media Player for MS Windows which we provide as reference player. It is a single executable file which does not require any installation at all. Note also that the visualization depends on the type of screen (CRT, TFT...) and its controls. Finally,in order to obtain an acceptable file size for a fast download, some noticeable color artifacts (false contours) may have been introduced by the codec.

Headlights comparisons

Two types of headlights comparisons have been performed, trying to answer two questions : how do compare different technologies of light sources? and given one same type of light source, which is the relative performance of two headlights on vehicles of different brands? Starting from 8 sequences, 7 headlitghs pairs have been compared :

Comparison	Types of light sources		Vehicle 1	Vehicle 2
1	simple^*	double^*	Seat Ibiza	Seat Ibiza
2	double	adaptive frontlighting	Seat Altea	Seat Toledo
3	double	high intensity discharge (Xenon lamp)	Seat Altea	Volvo XC90

^* Simple headlights are equiped with a double filament lamp and a unique parabolic reflector. Low and high beams are produced by the different filament position with respect the reflector focal point. Double headlights consists of two single filament lamps, each with its own reflector.

Comparison	Type of light source	Vehicle 1	Vehicle 2
4	simple	Seat Ibiza	Toyota Yaris
5	double	Peugeot 207	Seat Altea
6		Seat Ibiza	Seat Altea
7		Seat Ibiza	Citroën C3

The results are presented according to the following legend :

Original unsynchronized videos

Synchronized videos

Synchronized bird's views

Fusion

Original unsynchronized videos. In order to perform a fair comparison, the first frame of each video corresponds to the same camera position, that is, we have removed the relative time offset, so that the two videos "start at the same place" (the first reflective pole) at the same time.

Fusion of spatially aligned pairs of corresponding frames. The signed difference of intensities of the two registered frames is computed and encoded as a colour shade in the first sequence. If the difference is positive it is added to the red channel. If negative, it is subtracted from the green channel. Each frame has been cropped to the region of interest.