Intelligent Person Behaviour Analysis in Low Resolution Beach Video Imagery
Green, Steven Paul · Griffith Research Online · 2011
Object detection in complex scenes plays an important role in many real world applications, such as scene analysis, security, traffic violation detection, and medical research. The Object Detection and Behaviour Classification (ODBC) system created from this PhD research, is ground-breaking and concentrates solely on the domain of people and behaviour detection in low resolution beach scene imagery. The video source for this research was obtained from Coastalwatch beach webcams, which are located throughout Australia. Current object detection systems, overall, utilise a similar methodology, following four main image processing procedures; 1) remove unwanted noise, 2) background extraction from an image scene, 3) segmentation of the objects from the scene, and 4) object classification into different class types. To my knowledge, there is no other past or current research studying behaviour detection in low resolution beach imagery, and therefore, my research leads the way in this domain. Information from beach imagery can provide important data for tourism promotions and beach safety. Analysis of people behaviour could assist in making decisions about promotions and beach amenities. Automated beach systems may provide constant monitoring of a beach, by detecting people entering the ocean in a flagged or non-flagged area of a beach. This may also provide an early warning system to assist lifeguards in protecting swimmers in a beach area.