Project Outline

  Goals

  Project Plan

  Problems

  Proposed Solutions

  Constraints

Goals:

• To correctly recognize moving objects of interest
• To track those moving objects throughout their life spans.

Project Plan:

• Object Recognition steps:
  1. Obtain motion image sequence via temporal differencing
  2. Moving objects segmentation from moving pixels
  3. Templates Creation for object of interest

• Tracking steps:
  1. Template matching on locations as guided by motion detection
  2. Update the templates using the previous best match locations

Problems:

• Temporal Differencing results in too many noise. This is due to camera movement or just noisy background.
• When an object and background have close illumination intensity, region growing algorithm oftens make the object too big i.e., the background drifts onto the object.
• Region growing cannot grow a whole vehicle as one object but as many overlapping objects.

Proposed Solutions:

• Use size filter to eliminate noise. The idea is that the movements generated from the camera motion or noisy background will be a lot smaller than that of the objects of interest.
• Use another size filter to eliminate objects that are too big i.e., the objects that the background has drifted onto.
• Use equivalent table for the overlapping regions and assume them to be from the same object.

Constraints:

• Our algorithm cannot track objects that did not move in the beginning of the sequence.
• Because of our assumption of overlapping regions equivalence, the algorithm cannot track through severely occluded objects.
• Since our size filters are generated automatically they assume proportionate size between the objects' size and the image size. Hence when an object's size and the image size are way out of proportion the algorithm will fail miserably.

References:

1. Alan J. Lipton, Hironobu Fujiyoshi, and Raju S. Patil. "Moving Target Classification and Tracking from Real-time Video". Proc. IEEE Workshop Application of Computer Vision. Pages 8 - 14. 1998
2. Ismail Haritaoglu, David Harwood, and Larry S. Davis. "W4: Real-Time Surveillance of People and Their Activities". IEEE Transactions on Pattern Analysis and Machine Intelligence. Pages 809-830. 2000.
3. L. Wixson. "Detecting Salient Motion by Accumulating Directionally-Consistent Flow". IEEE Transactions on Pattern Analysis and Machine Intelligence. Pages 774-780. April 2000.
4. Alan J. Lipton and Hironobu Fujiyoshi. "Real-time Human Motion Analysis by Image Skeletonization". Proc. IEEE Workshop Application of Computer Vision. Pages 15 - 21. 1998
   
   
   
   HOME