Modified Optical Flow Morphology for Multiple-Object Tracking with Autonomous Driving Applications

Abha Choubey; Manpreet Kaur; Siddharth Choubey

doi:10.37591/joarb.v2i1.298

Abstract

In this study we have presented a real-time applicable multiple object tracking algorithm for autonomous driving applications. The proposed method is based on the modified morphology of the optical flow technique. Here the method allows the non-prior training of optical flow constraints for the detection of the objects. Additionally, it let the system execute itself in parallel rather than those sequential computing object tracking algorithms giving huge computing time for massive process. We are able to effectively deal with the occlusion and deforming feature points of the moving objects. The buffer vectors so introduced in the study gives stable results of predicting and tracking the moving objects in less number of computing steps.Keywords: object detection, multiple-object tracking, optical flow, object classification Cite this Article:Choubey A, Kaur M, Choubey S. Modified Optical Flow Morphology For Multiple-Object Tracking With Autonomous Driving Applications. Journal of Advancements in Robotics. 2015; 2(1): 1–7p.