The ever increasing use of video surveillance equipment for traffic monitoring together with rapid advances in digital signal and data processing technology and steady full in hardware costs provide a clear opportunity for the introduction of reliable and cost-effective automatic image analysis systems. The full potential of video image sensor systems can be realised if the system is capable of analysing and understanding the spatial and temporal characteristics of the image sequence. This paper describes a versatile system for the development of image analysis algorithms for road traffic data analysis applications. The model of a machine vision system (TRANSVISION) including its components and applications is described. In particular the implementation of a novel method for the extraction of partial 3-D information (a planar surface classifier) is presented followed by a discussion of initial test results and modifications to the classification subsystem. The objective of this work is to develop adaptive, robust and reliable image analysis algorithms for traffic monitoring and control. (A)
Abstract