Project ID:2013s13
Year:2013
Students:Levy Rina , Leibovits Oren
Supervisor:Appleboim Eli
In Collaboration with:BrightWayVision

Lane Detection with Night Vision Camera

The evolution of the automobile industry is indispensably dependent on complementary means for vehicles safety. One of the main challenges to address is the need for real time fast and reliable lane identification technology enabling accident avoidance. Most of the systems currently available, achieve reliable results under relatively bright vision conditions. Unfortunately, significant number of the accidents occurs at the dark hours and under poor sight terms.  Research conducted in Israel shows that 50% of the lethal accidents occur at night although traffic at the dark hours is only 25% of the traffic at daylight.

The project is conducted in cooperation with BrightWay Vision Company which develops innovative leading night- vision systems for the automotive market. The project goal is to implement an algorithm for real-time identification the lane boundaries using a night vision video film supplied by the BrightWay Vision laser camera.

Among the major challenges faced implementing the solution are the following:

  • Coming out of night vision system, the input image is gray rather than colorful and is much harder to analyze.
  • The algorithm should identify curved lane boundaries regardless of any pre-existing marks
  • The algorithm is required to be robust and achieve reliable results filtering out all irrelevant environmental data

The implemented solution integrates comprehensive knowledge acquired researching numerous relevant academic and technological articles. Based on the information gathered we came up with a novel algorithm addressing the project requirements. The algorithm is based on our ability to build an accurate mathematical/geometrical model of the lane at real time and under poor sight conditions.

2013s13-2