traffic light triggering for motorbikes using computer vision presentation

Upload: tommy-chheng

Post on 30-May-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    1/8

    1

    Vision Based Traffic Light Triggering for Motorbikes

    Tommy Chhengtcchheng*at*ucsd.edu

    CSE 190: Projects In Computer VisionInstructor: Prof. Serge Belongie

    15 March 07

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    2/8

    2

    Vision Based Traffic Light Triggering for Motorbikes

    Goal: To create a computer vision algorithm to detect

    incoming motorbike traffic for traffic light triggering.

    Motivation: Current traffic light triggers do not work on motorbikes

    General Idea: From video footage, track motion of objects and

    determine if the object's trajectory is intended incoming

    traffic.

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    3/8

    3

    Vision Based Traffic Light Triggering for Motorbikes

    Motion Segmentation Background subtraction Defining the background image for subtraction:

    A still image from a set time- Not robust- Slight movement to the camera will produce noise.- Difficult to automate updates.

    Previous Frame- Difficult to pick up differences of slow moving objects.

    Sliding Temporal Average- Compute average frame from the previous n frames.

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    4/84

    Vision Based Traffic Light Triggering for Motorbikes

    Connected Component Labeling

    Label the foreground blobs for tracking.

    Accomplished by computing the distance fromprevious objects and attaching it to nearest one withina threshold. If not, create a new label.

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    5/8

    5

    Vision Based Traffic Light Triggering for Motorbikes

    Lighting Conditions for Motion Segmentation

    Adversely affect motion segmentation

    Use dual colorspace selection During night time, RGB gives optimal results due to the

    headlights of the motorbikes and low noise. During day time, opponent color blue/yellow channel.

    Reduce effects of luminance. L*A*B* too nonlinear for lowresolution images.

    Original RGB Subtract Opponent color B/Y Subtract

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    6/8

    6

    Vision Based Traffic Light Triggering for Motorbikes

    Tracking

    Connected Component Labeling process labeledobjects in each frame.

    Can save position information regarding each object.

    Objects labeled in frame

    Trajectories of objects in frame

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    7/8

    7

    Vision Based Traffic Light Triggering for Motorbikes

    Tracking Problems

    With simplified tracking determination, cross trafficcan easily steal the intended tracking(or vice versa).

    Even if the tracking does not entirely work, partialtracking maybe enough to do the job.

    Use RANSAC to determine the partial trajectories.

    Traffic mis-tracked RANSAC to get partial trajectories

  • 8/14/2019 Traffic Light Triggering For Motorbikes Using Computer Vision Presentation

    8/8

    8

    Vision Based Traffic Light Triggering for Motorbikes

    Steps I did not have time for: Actual classification system

    Now the traffic detection problem is reduced to a linefitting/classification problem.

    Ideas include feeding in annotated correct trajectories andtesting by a difference on the test data.

    Extensive training/testing set Difficult to get obtain varied footage

    Researched more on other methods at the varioussteps.

    Perhaps interest points moving in the same direction.