Detecting and tracking the position of suspicious objects using vision system.

Muthukrishnan, Ramkumar; Kannan, Somasundar; Prabhu, Radhakrishna; Zhao, Yafan; Bhowmick, Parijat

Abstract

Vision-based object tracking is crucial for both civil and military applications. A range of hazards to cyber safety, vital infrastructure, and public privacy are posed by the rise of drones, or unmanned aerial vehicles (UAV). As a result, identifying suspicious drones/UAV is a serious issue that has attracted attention recently. The key focus of this research is to develop a unique virtual coloured marker based tracking algorithm to recognise and predict the pose of a detected object within the camera field-of-view. After detecting the object, proposed method begins by determining the area of detected object as reference-contour. Following that, a Virtual-Bounding Box (V-BB) is developed over the reference-contour by meeting the minimum area of contour criteria. In order to track and estimate the precise location of the detected object in two-dimensions during observations, a Virtual Dynamic Crossline with a Virtual Static Graph (VDC-VSG) was constructed to follow the motion of V-BB, which is considered as a virtual coloured marker. Additionally, the virtual coloured marker helps to avoid issues linked to ambient lighting and chromatic variation. To some extent, it can function efficiently during obstructions like rapid position fluctuations, low resolution and noises etc. The efficacy of the developed algorithm is evaluated by testing with significant number of aerial sequences, including benchmark footage and the outputs were outstanding, with better results. The suggested method will support future industry of computer vision-based intelligent systems. Potential applications of the proposed method includes object detection and analysis applied to the field of security and defence.

Citation

MUTHUKRISHNAN, R., KANNAN, S., PRABHU, R., ZHAO, Y. and BHOWMICK, P. 2023. Detecting and tracking the position of suspicious objects using vision system. In Bouma, H., Dijk, J., Prabhu, R., Stokes, R.J. and Yitzhaky, Y. (eds.) Artificial intelligence for security and defence applications: 2023 proceedings of the SPIE (International Society of Optics and photonics) Security and defence conference, 4-5 September 2023, Amsterdam, Netherlands. Proceedings of the SPIE, 12742. Bellingham, WA: SPIE [online], article ID 127420C. Available from: https://doi.org/10.1117/12.2679801

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2023 SPIE (International Society of Optics and photonics) Security and defence conference
• Start Date: Sep 4, 2023
• End Date: Sep 5, 2023
• Acceptance Date: Apr 5, 2023
• Online Publication Date: Sep 8, 2023
• Publication Date: Oct 17, 2023
• Deposit Date: Nov 14, 2023
• Publicly Available Date: Nov 14, 2023
• Publisher: Society of Photo-optical Instrumentation Engineers
• Peer Reviewed: Peer Reviewed
• Series Title: Proceedings of the SPIE
• Series Number: 12742
• Series ISSN: 0277-786X; 1996-756X
• Book Title: Artificial intelligence for security and defence applications: 2023 proceedings of the SPIE (International Society of Optics and photonics) Security and defence conference, 4-5 September 2023, Amsterdam, Netherlands
• ISBN: 9781510667136
• DOI: https://doi.org/10.1117/12.2679801
• Keywords: Image processing; Cascade classifier; Object detection; Object tracking; Aerial images
• Public URL: https://rgu-repository.worktribe.com/output/2146552