Application of Image Mining for Knowledge Discovery of Analyzed Traffic Images

Abstract

This paper presents a new application of image mining systems for knowledge discovery of traffic/people behaviors. Our proposed image-mining framework analyzes the traffic image sequences by using background subtraction, image segmentation, and object tracking in the domain of traffic monitoring. The spatio-temporal relationships of the vehicle objects in each frame are discovered and accurately captured and modeled. Then trajectory of moving objects is determined and normality and abnormality of people/vehicle behaviors is recognized. In our system a fuzzy Decision Maker (FDM) is used to modeling and extracting knowledge from traffic behavioral patterns.