The paper describes a system, known as TULIP (Traffic analysis Using Low Cost Image Processing). It is based on an IBM personal computer, which acts as the host to an image processing sub-system. The image processor consists of a frame grabber and frame store along with programmable look up tables, and can capture images and perform primitive processing of them in real time, i.e. the time it takes a video camera to transmit one full frame. Software has been written in 80286 assembler code and the C language to extract useful information from the video images, again in real time. The system has been designed for use by non-computer specialists. The operator is required only to position observation windows on the image screens. The passage of vehicles through the windows is monitored and basic information such as flow or speed is determined. Potential applications of this sort of system include temporary traffic flow and speed censuses. While other types of vehicle detector may be used for traffic censuses, image processing techniques obviate the need to attach inductive loops or pneumatic tubes to the surface of the road. It also permits a permanent record of the traffic (in the form of a video) to be kept for later analysis. The paper reports on results obtained with the TULIP system when applied to traffic monitoring. The article concludes with an assessment of the potential impact of anticipated advances in image processors and computer technology on the scope for the application of such systems.(A)
Abstract