Automobile manufacturers face a steep road in upleveling their vehicles systems to protect pedestrians and other vulnerable road users. A primary contributor to this challenge is imaging at night and other adverse weather conditions that impact and impede visibility.
When imaging pedestrians from a vehicle, two requirements contribute significantly to the likelihood of successful detection—sufficient detail to separate pedestrians from the background and a broad enough field of view to assure that all vulnerable pedestrians are seen. Combining these requirements defines the resolution required in the camera viewing the scene. This calculation is worth reviewing to facilitate selection of suitable cameras and lenses appropriate to pedestrian detection.
The standard Johnson Criteria - Detection, Recognition & Identification -shown here - provides sufficient levels of perception for the original defense applications it was created for, but the variety of objects present in traffic scenes prompts an additional level - Classification - where the system needs to know more detail than Recognition provides but does not need all the detail required for Identification.
In these images, the unit is pixels per meter (ppm) as measured at the object to be perceived so the requirement is to provide enough pixels at the location of the most distant pedestrian to be perceived so that the convolutional neural network (CNN) will reliably and accurately tag that pedestrian with a position in the image frame and a distance. This constraint produces a simple relationship between finding pedestrians and the width of the camera coverage, called the horizontal field of view (HFOV). Taking the minimum required resolution for Classification (an operation between Recognition and Identification) to be 10 pixels per meter, then a VGA camera with 640 pixels horizontally can cover 64 meters (211 ft).
Compared to the typical street width, somewhere in the 25-meter range, 64 meters seems like it should be sufficient for protection of pedestrians but, where improved safety is the goal, initial impressions should be verified by careful analysis. In this case, the necessary field of view must be carefully defined.
This isn't the only unique requirement for building pedestrian-safe automobiles. Download our latest white paper Requirements for Effective Pedestrian Imaging for a thorough review of the many unique requirements, and key watch-outs for designing pedestrian-friendly autonomous vehicles.