In this current world, unmanned aerial vehicles (UAVs) have been increasing their popularity in both the public and the private sector due to its high and seemingly limitless potential. They have been experimented rigorously and are known to provide great benefits in a wide range of applications such as surveillance and reconnaissance missions, and delivery of parcels and medical equipment. Having UAVs surveying the area instead could reduce the physical workload of humans, keeping them safe from hostile environments that could put their own lives at risk. As the UAV does not require a pilot to be on board, it is not constrained by a pilot’s physical limitation. They can be autonomously programmed to complete repetitive yet precise missions, and have the capability to enter hostile environments.
Object detection and tracking are the primary tasks of computer vision. As humans observe the world around them every day, the image processing in a human brain should be fast and accurate. The human vision may be considered the most complex among the human senses and the advancement in engineering has brought this ‘vision’ to artificial intelligence. Computer vision is the science and technology of how a machine can see and learn information from images and videos. Being able to provide real-time vision-based information during emergencies such as locating a stranded group of people, the possibilities of ensuring a high standard of safety and mission success are endless. In medical applications, computer vision is can extract information from medical images of patients such as used in X-Ray. Surveillance also aids in crime prevention, area monitoring, and other applications. From recorded videos or a live video feed, detecting and tracking a target of interest can provide key information. The use of computer vision in UAVs is highly useful, as the imagery of people and key objects, can be easily captured, detected and possibly tagged with a geolocation.
Computer vision applications such as aerial surveillance have been an active research topic as it can be widely used for numerous purposes and has created a huge amount of interest worldwide in object tracking algorithms. From recent research, it is found that detecting and tracking an object in a video sequence is a complex task in which the system continuously tries to detect, recognize and track targets. There are many different algorithms available which work efficiently on the given mission. Object detection is usually the first task in surveillance, followed by the extraction of background information for further processing. Images captured by a moving camera may be of low resolution, thus requiring appropriate image processing techniques to refine the quality of the imagery received.