Image analysis and rule-based reasoning for a traffic monitoring system

Abstract

The paper presents an approach for detecting vehicles in urban traffic scenes by means of rule-based reasoning on visual data. The strength of the approach is its formal separation between the low-level image processing modules (used for extracting visual data under various illumination conditions) and the high-level module, which provides a general-purpose knowledge-based framework for tracking vehicles in the scene. The image-processing modules extract visual data from the scene by spatio-temporal analysis during daytime, and by morphological analysis of headlights at night. The high-level module is designed as a forward chaining production rule system, working on symbolic data, i.e., vehicles and their attributes (area, pattern, direction, and others) and exploiting a set of heuristic rules tuned to urban traffic conditions. The synergy between the artificial intelligence techniques of the high-level and the low-level image analysis techniques provides the system with flexibility and robustness.

The Sakbot System for Moving Object Detection and Tracking

“…These systems obviously differ in the approach proposed, but also in various assumptions about the operational environment. One first main distinction is between systems adopting a single, fixed camera [1][2][3][4][5][6] with respect to systems adopting either multiple cameras [7] or an airborne camera [8]. In this work, we focus on a single fixed camera scenario, since it still captures a wide spread of applications.…”

“…In this work, we focus on a single fixed camera scenario, since it still captures a wide spread of applications. Another relevant distinction is made between systems oriented to monitoring of traffic scenes, where the main targets are vehicles and pedestrians (see for instance [1,5,6,9]), and systems for video surveillance of unattended areas such as metro platforms,parking areas (see [3,4]) and unmanned railways environments [10]. In the two cases, different a-priori knowledge about objects in the scene can be exploited in order to improve detection and/or tracking.…”

“…Once the input image has been corrected, the still background is subtracted to detect MVOs. Differently from many other proposals (see for instance [1,6]), we exploit the information given by the pixel chrominance to better discriminate foreground from background.…”

A comparison of two Monte Carlo algorithms for 3D vehicle trajectory reconstruction in roundabouts

“…One approach to traffic monitoring is the visual vehicle tracking [3], i.e., the process of recognizing moving objects and estimating their trajectory from a video sequence. Most of the existing visual vehicle tracking systems propose a 2D approach (2D tracking hereafter): these systems identify moving vehicles on the image plane, e.g., by identifying their blobs (see Fig.…”

Moving Vehicle Tracking for the Web Camera

“…This approach requires heavy computation for calculating optical flow vectors. Another method infers the moving information by computing the difference images and edge features for complementary information to estimate plausible moving tracks [4] [9]. This method may be very sensitive to illumination and noise imposed on video stream.…”