A Novel Algorithm for Night Time Vehicle Detection Even with One Non-functional Taillight by CIOF (Color Inherited Optical Flow)

• Published in: International Conference on Pattern Recognition Systems : 20-22 April 2016, Talca, Chile
• Main Authors: Dave, P, Gella, NM, Saboo, N, Das, A
• Format: Conference Proceeding
• Language: English
• Published: Stevenage The Institution of Engineering & Technology 20.04.2016
• Subjects: Advanced driver assistance systems; Algorithms; Automobiles; Color; Containment; Developing countries; Headlights; LDCs; Night; Object recognition; Optical flow (image analysis); Optimization; Patent applications; Vehicles; Visibility
• ISBN: 9781785612831; 1785612832
• DOI: 10.1049/ic.2016.0031

Advanced Driver Assistance System (ADAS) is the first step towards achieving the dream of autonomous vehicle with a lot of re-usable building blocks. As far as the sensing of the environment is concerned in terms of object detection in night, ADAS needs to depend on tail and head light of vehicle due to partial or zero visibility of the vehicles. It is a challenge to implement ADAS in mid and low segment cars in developing countries like India as the traffic is far from ideal both in terms of minimum infrastructure available and abiding traffic rules. The current paper has addressed the aforementioned scenarios where even in absence of dual tail light and with partial visibility of vehicle, night time vehicles are detected. Color neighborhood correspondence between central bright blob and red halo is targeted as attributes and modified for optimization in terms of both accuracy and performance. Containment of vehicles even with one non-functional tail light has been determined through the non-divergent optical flow clustered hierarchically inherited from color maps. Technology disclosed in this paper is subject matter of pending patent application.