Recent advancement in the processing power of onboard computers has encouraged engineers to impart visual feedbacks into various systems like mechatronics and internet of things. Applications ranging from CCTV surveillance to target detection and tracking using UAVs, there is a wide variety of demand on image processing techniques in terms of computational time and quality. In this scenario, developing generalised algorithms which gives a freedom to user in choosing the trade-off between quality and quick response is a challenging task. In this paper a novel boundary detection algorithm for segregating similar coloured objects in an image is presented, which accommodates a degree of freedom in choosing resolution of object detection to the detection time. This method uses colour based segmentation as preprocessing technique to reduce overall computational complexity. It is independent of the shape (convex or non-convex) and size of the object. Algorithm is developed using Open-CV libraries and implemented for separating similar coloured vehicles from an image of different vehicles on road. Implementation results showing different choices of boundary tightness and computation times are showcased.
A. Alexander and Meher Madhu Dharmana, “Object detection algorithm for segregating similar coloured objects and database formation”, in 2017 International Conference on Circuit ,Power and Computing Technologies (ICCPCT), Kollam, India, 2017.