Real time classification and tracking of multiple vehicles in highways
Introduction
High speed processing of image frame sequences is highly important for many real time computer vision algorithms. Image sequence analysis provides intermediate results for conceptual description of events in a scene. In vehicle tracking applications that uses image sequences, many methods are suggested. At first we can name model-based tracking methods in which a 3D model of vehicle is extracted (Coifman et al., 1998, Malik and Russell, 1997). One of the advantages of these methods is their high accuracy in determining the vehicle type and their detail geometric model. In fact, model-based tracking methods because of their high calculation cost, can be used only for free-flowing traffic with small number of vehicles.
In some other methods that are called feature-based (Roberts, 1994), a few features such as distinguishable lines or corners are extracted for each vehicle. Some of these features are grouped together to label a vehicle (Shi and Tomasi, 1994). One of the most important advantages of this type of methods is that, even in presence of partial occlusion, some of these features remain to be visible. On the other hand, they face problems in detecting features for individual vehicles that run close to each other. In region-based methods (Coifman et al., 1998, Setchell, 1997), vehicles are presented as blobs. In these methods, at first, connected components are extracted and then regions are merged or split ted if needed. The most serious weakness of these approaches is that merging and splitting regions can cause some inaccuracy in vehicle detection.
In addition, there are some methods in which the contour of vehicles is extracted. Although contours can be detected by simple edge detection methods, but these simple methods sometimes detect false edges of the background too. However, if more complex algorithms of edge detection such as active contours or snakes (Paragios and Deriche, 2002) are used, one should find a way to optimize the coding to make them usable for real time applications with commercially available processors. In practice there are many applications such as our system, that often one does not need to know the exact detail of vehicle type, but a general type category would be sufficient. In our system the surveillance CCD camera is installed in a relatively far distance from a highway and the vehicles are visualized as small objects with minimum detail on their geometrical model.
In this paper we introduce a novel real time machine vision system for classification and tracking of multiple vehicles and also determining some traffic parameters such as lane change and counting the number of vehicles passing the highway during a desired time interval. For tracking we used Kalman filter (Grewal and Angus, 1993) and background differencing techniques. Our algorithm takes advantage of region based and contour based methods by combining their ideas in order to detect a “Shape of Interest”, that in practice it is a bounding box around the vehicle. By using the bounding box and region boundary, the occlusion and overlapping of two regions are detected by examining the object shape and determining if it was the result of merging two or more vehicles and then deciding upon a proper split point to separate the merged vehicles.
Our system was implemented in Visual C++ using Matrox Meteor II frame grabber on a Pentium II 800 MHZ CPU. The input images were gray scale with eight bits per pixel resolution and of size 320 × 320. Our experimental results showed an accuracy of 96% when it was compared with the measurements done by a human expert. The initial version of this work is given in (Rad and Jamzad, 2003).
The rest of this paper is organized as follows: After a review of related works, our algorithm is described in three main sections, Change detection, Vehicle recognition (where our ideas for occlusion removal and vehicle classification are discussed) and Vehicle tracking. In Section 6, the experimental results is presented and finally the conclusion remarks is given in Section 7.
Section snippets
Related works
Many works have been reported for vehicle tracking from image sequences in machine vision and related topics literature. Vehicle detection is a fundamental component of image-based traffic monitoring system. Here we take a brief look at some of them. One such approach is to use background subtraction or optical flow for detection of moving objects (Javed and Shah, 2002, Gupte et al., 2002) and then tracking them. Methods based on background subtraction followed by object tracking do not suffer
Change detection
Detection of changes between sequences of frames is a major task in many machine vision applications. Methods that are based on sequence frame differencing and moving edge detection, have aperture problem and are sensitive to vehicle speed. In the following we show how to use background differencing method to group pixels in moving and non-moving category.
In this method, first we construct a background reference image of the road that has no moving vehicles in it. In order to avoid the problem
Vehicle recognition
As seen in Fig. 1(a), this binary image has several small noises in it. Our vehicle recognition algorithm assumes to receive as input a binary image that only has two groups of pixels. Pixels belonging to the background, and those belonging to the moving objects. This means that we have to modify Fig. 1(a) in such a way that it becomes completely noise free. For noise removal, we used Closing and Opening morphological operators. It is known that Closing fills little apertures and Opening
Vehicle tracking
For tracking objects in a sequence of frames, the relation between objects in two consecutive frames must be found and recorded. Doing this we developed three modules. The first one is a complete image search, in which the whole image is searched. In the second module, only the area on the road (i.e. excluding the background) is searched. And in the last module, a small area in which the vehicle might be seen in next frames was determined and the trajectory of each vehicle was tested.
These
Experimental results
In order to test our algorithm, we used image sequences of about 400 frames of a video tape captured by a traffic surveillance CCD camera. This camera was installed on a height and far distance from a wide highway in the city of Tehran. The recorded video showed two sides of the highway that has three lanes in each side. The average number of vehicles in a frame was measured to be 27.2. The mean processing speed of our algorithm measured on 400 consecutive frames is summarized in Table 1.
As
Conclusion
In this paper, we presented an algorithm for real time detection of vehicles, classification of their types, and tracking. Our system was implemented on commercially available PC and a frame grabber. Its processing speed is 10.99 fps. In this application, since video recording is done from a relatively far distance, the field of view of camera is large enough so that running vehicles remain in the field of view in such a period of time that the processing speed of about 11 frames per second
Acknowledgments
We would like to thank the Control Traffic Company of Tehran for their cooperation and providing the traffic video tapes.
References (19)
- et al.
A real-time computer vision system for vehicle tracking and traffic surveillance
Transport. Res.: Part C
(1998) Tracking multiple vehicles using foreground, background and motion models
Image Vis. Comput.
(2004)- et al.
Geodesic active regions: A new paradigm to deal with frame partition problems in computer vision
J. Visual Commun. Image Represent.
(2002) - Bennett, B., Magee, D., Cohn, A.G., Hogg, D.C. Using Spatio-temporal Continuity Constraints to Enhance Visual Tracking...
- Cheung, S.-C., Kamath, C. Robust techniques for background subtraction in urban traffic video, Video Communications and...
- et al.
Digital image processing
(2002) - et al.
Kalman filtering and practice
(1993) - et al.
Detection and classification of vehicles
IEEE Trans. Intell. Transport. Syst.
(2002) - Javed, O., Shah, M. Tracking and object classification for automated surveillance, Proceedings ECCV,...
Cited by (85)
A longitudinal scanline based vehicle trajectory reconstruction method for high-angle traffic video
2019, Transportation Research Part C: Emerging TechnologiesCitation Excerpt :Compared with tracking a single object, the task for tracking multiple objects imposes more difficulties. A popular technique for vehicle tracking is the Kalman Filter (KF), which is considered an efficient approach for estimating the dynamic state of a system for tracking (Rad and Jamzad, 2005; Hsieh et al., 2006; Kim and Cao, 2010). Many researchers have tried to add some prior knowledge, such as movement constraints, locations, or past tracking histories to better approximate the object locations (Shi and Tomasi, 1994; Coifman et al., 1998; Azevedo et al., 2014).
Multi-vehicle detection algorithm through combining Harr and HOG features
2019, Mathematics and Computers in SimulationCitation Excerpt :In fact, vehicle environment perception is based on primarily computer vision technology which can provide warning and auxiliary driving instructions to drivers, such as vehicle safety warnings, anti-collision warnings, autopilot and tracking assistance, through acquiring real-time information on the surrounding environment. Considering the importance of vehicle operations in transportation systems, vehicle environment perception has become a central issue internationally for improving the efficiency and safety of vehicle operations [15,25]. Computer vision technology based moving target detection is one of the basis of vehicle environment perception.
Machine learning and computer vision-enabled traffic sensing data analysis and quality enhancement
2018, Data-Driven Solutions to Transportation ProblemsEvaluating the accuracy of vehicle tracking data obtained from Unmanned Aerial Vehicles
2016, International Journal of Transportation Science and TechnologyCitation Excerpt :In their study, they estimate vehicle lengths of within 10% in each instance, deriving coordination mapping functions of a calibrated camera model. Rad and Jamzad (2005) performed a similar work developing an application able to count and classify vehicles and to identify the lane-changing events by tracking them. In their study, the background subtraction method combined with morphological operations has been applied in order to identify moving vehicles in regions.
Intelligent Traffic Surveillance through Multi-Label Semantic Segmentation and Filter-Based Tracking
2023, Computers, Materials and Continua