May 5, 2009
Vision Based Traffic Light Triggering for Motorbikes
Back in my college days at UC San Diego, I worked on a project using computer vision to solve the traffic light triggering problem. The general gist of problem is that a lot of traffic light sensors have a hard time detecting the presence of a motorcycle. This is a safety hazard as a motorcyclist may have to run a red light simply because the traffic light is not triggered. My idea was to detect and predict the trajectory of an object in video capture targeted at a traffic light.
Abstract
Current traffic light triggering is based on inductive loop sensors. Unfortunately, motorbikes (scooters, motorcycles, etc) have a difficult time triggering these sensors. In this paper, we propose an image processing algorithm to detect motorbikes at a traffic stop using a fixed camera. The algorithm tracks the trajectory of the objects in the footage by motion segmentation and connected component labeling. Classification can be created to categorize these objects as incoming traffic based on the object’s trajectory. To handle different lighting conditions in the motion segmentation, we take a dual approach by selecting RGB or Opponent colorspace. RANSAC is utilized to help trajectory creation. Experimental tests using real video footage exhibit robust results under varying conditions.
Video
In this video, you see the detection of all the light sources. Then we track the bike(note how it bike stays labeled #1) using RANSAC to differentiate the bike from the intersecting traffic.
Presentation
For more details about this idea, check out the project blog at Vision Based Traffic Light Triggering for Motorbikes