1

Www Lightsoffnow Tag Megapixel Ccd Lights Off Now Examinations of the Recognition Distance of Head-lamps

Www Lightsoffnow Tag Megapixel Ccd Lights Off Now

Www n1 searchr Lightsoffnow t Www csearchlsearchpsearchisearchtsearchoa Www is Ccd a Lightsoffnow Lightsoffnow x Ccd e Lightsoffnow i Ccd e Ccd tsearch Lightsoffnow s Ccd t Tag e1vsearchs Megapixel a Tag Ccd r Tag tr Ccd o dsearcht Lightsoffnow c1i Megapixel n Tag o rsearchco Megapixel n Lightsoffnow tsearchosearch)bc Tag ue isearch ha searchtsearchosearchg Lightsoffnow efcsearch Megapixel nt1esearchrsearchssearchlt Tag Ccd f Tag i Lightsoffnow u Www l1e Megapixel p Lightsoffnow r Megapixel m Ccd nsearchs Ccd Megapixel hsearch Lightsoffnow is Lightsoffnow a searchr Www tesearchisearchn Tag fr Ccd o Ccd rob Tag e Megapixel v Lightsoffnow r Tag Lightsoffnow as Www ¡°1u Megapixel tp Megapixel rcsearchiv Ccd d±,as1 Lightsoffnow h Ccd y Ccd di Tag n Tag ¯tsearchh Www dt Ccd Megapixel dsearchn Megapixel ify t1e v1sal Tag t Lightsoffnow r Lightsoffnow esearch. Tag T Ccd e Www r Tag trsearchon Tag ° Www u Ccd t psearchrsearcheived¡± corresponds to the detection of an object. According to common parlance we use the term recognition distance, even though we know that for the identification of objects we would get other results.

2 Previous research

Up to now the most common model for the prediction of the recognition distance is the illuminance-based model. Although it has been consistently emphasized that the illuminance is not a good predictor for recognition distance (/Kl04/, /Vö01/), it is used in the legislation to give minimal values for the recognition distance (ECE No. 20, No. 98). This common model has to be revisited.

A study of Locher and Völker (/LoVö02/) showed that for recognition distance the luminance in high distances (behind cut-off line) is the best predictor (other parameters were luminous flux, homogeneity, ¡­).

Eckert (/Ec93/) developed a model for the calculation of the recognition distance of pedestrians for the reconstruction of accidents. Therefore he uses the Threshold-Luminance-Model of Adrian (/Ad69/). He defines 13 pairs of measuring points on the pedestrian where the luminance should be measured. These measurements lead to 13 luminance differences, which will be compared with calculated threshold-luminance differences (based on the Adrian model). The object is visible if the existing values are bigger than the threshold values. With the size of an object and the existing differences of luminance, you are able to calculate a recognition distance. The application of this model to real luminance distributions with pedestrians (in /Kl04/) showed better relations between luminance difference and recognition distance as the illuminance.

In /KoGa00/ a model for the calculation of the drivers visual distance was developed. The basis of this model shows figure 1.

Figure 1: Model of FASIVAL

In /KoGa00/ a threshold contrast for every possible distance is calculated based on a model by Kokoschka (/Ko88/, solid line). Afterwards the mean of the edge contrast of a visual target is calculated for every possible distance (dashed line). The ratio of the edge contrast and the threshold contrast is the visibility level VL. ) The recognition distance is defined with a specified border (i.e. VL = 1, in figure 1 it is about 83m). Also this model was applied to existing luminance distributions (/Kl04/). The result was a high correlation between threshold contrast and recognition distance.

In a study, which was carried out last year, we measured the recognition distance of a grey panel and the luminance distribution at different distances. The result of this study was an attempt to predict recognition distances based on the courses of contrast of the headlamps (figure 2). More detailed information about this model can be found in /Kl04/.

Figure 2: Courses of contrast for different headlamps

The experiments were carried out with real headlamps, that means with real light distributions. The problem is, that located relations cannot be due to one isolated parameter because light distributions differ in a lot of parameters.

Therefore new experiments with controlled parameters are necessary. One of these experiments will be explained in the following chapters.

Measurement of the recognition distance

To measure the recognition distance we used a toy train which was set up in the light tunnel of Hella in Lippstadt, Germany. The track runs on the middle of the right roadway and on the train was a grey panel (30cm x 30cm, ¦Ñ = 4.9%) as the visual target. The observers were sitting in front of the light tunnel and had a free view over the road. A schematic view of the setup can be seen in figure 3.

Figure 3: Setup for the measurement of the recognition distance

The procedure of the measurement of the recognition distance is described as follows. The train with the visual target started at a high distance from the observer where the grey panel was not visible (position one). The train drove to the observers with a constant velocity of approximately 1m/s. When the train passed the first light barrier, a clock was started. When the observers were able to see the visual target, they had to push a button (position two). So for every observer the clock stopped separately. Out of these times and the velocity of the train (which was calculated with the signal of the second light barrier) the recognition distance (SEE) for every observer was calculated.

Test conditions

One main question or our studies is: Which parameters influence the recognition distance of headlamps in which way? Some of these possible parameters are listed in the following:

To get down to a new model for the prediction of the recognition distance, it is necessary to know which are the relevant and most important parameters. We also have to know the relations between these relevant parameters and the recognition distance. As it was said before, it is not possible to find out these relations by using real light distribution because they differ in a lot of these parameters. To control parameters, it was necessary to carry out our experiments with very easy light distributions. By using a slide projector to project the light distributions we were able to create every light situation we want and had the control over all parameters.

In the experiment we determined the effect of the visual angle of the observer on the recognition distance. This was not only carried out to identify the fundamental relation, but also to validate our procedure of measuring the recognition distance, because every observer has a different visual angle to the target. If there would be a relation between these quantities, we had to consider it in future measurements.

We used one very simple light distribution. It consists of a horizontal cut-off line, which divided the distribution into two regions. These regions had a homogeneous light distribution. The light distribution was adjusted in the way, that the cut-off line impinges the road in 65m and 75m respectively (this was done to estimate the influence of the position of the cut-off line). On a 10m wall in front of the projector the measured illuminance was 49.3lx for the area under the cut-off line and 5.8lx over the cut-off line. Figure 4 shows the geometrical data of the test setup.

Figure 4: Test setup

We had 15 observers (visus > 0.8, 21 ¨C 36 years) disposed in 3 groups per 5 observers according to 5 different observing positions. The eye position of observers sitting on position one, two and three was adjusted to a height of 1.25m. On position four and five the height of the eye position was 1.55m. Each observer had to sit on every position. On every position recognition distance was measured twice (to check the accuracy of measurement). We received 30 values of recognition distance for every position. Measurements of recognition distance started after an adaptation time of 10 minutes. The visual criterion for the observers was ¡°just perceived¡±, so didn¡¯t had to identify the card. After the observations the luminance distributions from every position were recorded in intervals of 5m.