Tracking of a Missile in a Video using Accelerated Chan-Vese Model
Divya Nemidoss1, Muthaiah Rajappa2, Jaikanth Jayakumar3
1Divya Nemidoss, M.Tech. Student, VLSI Design, SASTRA Deemed to be University, Thanjavur (Tamil Nadu), India.
2Muthiaiah Rajappa, Professor, Department of Computing, SASTRA Deemed to be University, Thanjavur (Tamil Nadu), India.
3Jaikanth Jayakumar, B.Tech. Student, Department of CSE, SASTRA Deemed to be University, Thanjavur (Tamil Nadu), India.
Manuscript received on 07 April 2019 | Revised Manuscript received on 20 April 2019 | Manuscript published on 30 April 2019 | PP: 1686-1691 | Volume-8 Issue-6, April 2019 | Retrieval Number: F5066048619/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The target of tracking is to find the trajectory of an object of interest in motion and it is one among the significant applications in image and video processing. Some of the applications of tracking are video path planning, motion based recognition and automated surveillance. The two phases involved in tracking of object of interest in motion are object detection and tracking of detected object of interest. One of the technique used for tracking of object of interest (missile) in a missile video is Chan-Vese model based tracking. Chan-Vese model based tracking consists of the following steps: Segmentation of object of interest in first frame by using deformable model (Chan-Vese model) and tracking that segmented object in consecutive frames by giving the segmentation output of previous frame as the initial contour to the successive frame. The advantage of Chan-Vese model is that it works even with noisy input image but its drawback of slow convergence to optimum solution restricts its usage to tracking application since tracking demands quick tracing of object of interest in successive frames of a video. This paper focuses on accelerating the convergence of Chan-Vese model using first order optimization scheme called Hybrid method. Hybrid method is a fusion of Nesterov accelerated gradient descent method and Barzilai and Borwein gradient descent method. The fastened algorithm has been tested with static missile image (both clean and noisy image) and also in missile video. The metrics noted were processing time (in seconds) and number of iterations. The efficiency of the proposed optimization technique has been proved through these metrics.
Keyword: Chan-Vese Model, Segmentation, Tracking.
Scope of the Article: Advanced Computing Architectures and New Programming Models