Multiple Car Detection, Recognition and Tracking in Traffic

Multiple Car Detection, Recognition and Tracking in Traffic

Table of Contents


Multiple Car detection, recognition, and tracking is one of the most rapidly growing systems in image processing used for a video surveillance systems. This has been designed with an intention to reduce the traffic congestion in highways, decrease street misshape, Identify people in crime scène, reduce road accidents, vehicle theft detection, living zone, etc. Detecting the shape of a moving car in a video sequence is very difficult in the present environmental conditions. In order to overcome the effect of this problem, The paper proposed a number of methods like the video is converted into frames in segmentation then followed by preprocessing and tracking of multiple cars. This paper provides a brief about multiple car detection, recognition, and tracking of approaches.

  • Author Keywords

    • Multiple car ,
    • detection ,
    • recognition ,
    • tracking ,
    • background subtraction ,
    • foreground ,
    • frame differencing ,
    • feature extraction ,
    • temporal differencing and optical flow ,
    • current frames
  • IEEE Keywords

    • Automobiles ,
    • Tracking ,
    • Feature extraction ,
    • Object detection ,
    • Image color analysis ,
    • Shape ,
    • Motion segmentation


Information and technology has not only play a significant role in modern civilization but also created a threat in the field of the modern transport system such as increase in number of roads, highways, and traffic in each developing nation. The major technique introduced for this purpose is the traffic monitoring and controlling system by an image processing designed to implement the multiple car detection and tracking in traffic. The static camera is operational for the purpose of the traffic surveillance system. It is useful in finding out the variations in effect of illumination, shades, and presence of shadow of a moving car. The process of finding multiple cars in a video which is divided into a number of frames .The aim of moving car detection, recognition and tracking is the process of determining the foreground image from a background image. Cars are detected by its shapes and recognized it by tracking multiple cars. The various methods proposed forthwith the object of detection and tracking in relation to video surveillance system. They are Image sequence, motion segmentation, preprocessing, background subtraction, clear objects and tracking. All the modules referred to above have been explained clearly in the subsequent text proposed.

Literature Survey

Sheeraz Memon et al[1]. This paper is designed to implement on python using openCV bindings. This system involves recognition, detection of categorizing the various vehicles in accordance with video frames and differentiate as per the size .In the process of background subtraction which is determined by counting and detecting the vehicles by using Gaussian Mixture Model(GMM).Foreground Extraction method by which background image is subtracted there by clear image of foreground can be obtained and then image enhancement technique which are essential for filtering, dilation and erosion in order to obtain a clear foreground objects. classification is done by two methods. They are Contour Comparison (CC) and Bag of Features (BoF), Support Vector Machine(SVM). Hongsheng He[2]. This system is a process of obtaining an image through traffic camera in a metropolitan cities. The view points and variations of lighting conditions at a given traffic area which images are captured by cameras are differentiated and rectified as model detectors. In the initial stage, the frame work of cars are first detected and extracted by using part-based model .These model consists of Model initialization, Deformation detector model, root detector and component detector. Feature region rectification is done taking into account of License-plate anchor and Headlamp anchor and finally, vertical and horizontal histograms of texture features. Photometric Feature Extraction is further classified as Illumination Normalization by two columns ,namelyfirst column is obtained in different shades and weather conditions. second column is demonstrated by shadow compensations and histogram equalization. The Car Model Recognition is determined by neural network classifiers. Ahmad Arinaldi et al[3]. The system has been technically designed for implementation of effective traffic of real time management and also to gather important statics, regarding lane usage monitoring, taking into account ,the classification such as counting ,speed and type of vehicle. This will generally help the regulator and policy makers to quickly respond to traffic situation. In this system two methods are proposed. A pipeline model consists of Gaussian background subtraction in which moving vehicles are detected in video frames. From there area of detection, abounding base of image is formed to classify the type of vehicle. Vehicle classification though this method are cars, delivery cars, trucks, large trucks and buses. The second one is Region Conventional Neural Network(RCNN) is used to classify, bounding box detection are captured in a video frames and tracking can be done on the position of the detected vehicle to get the speed and lane position of the vehicle.

Poonam A. et al[4]. The main proposal put, forth in this paper is to enumerate the importance of accurate and effective moving vehicles by detection methods, projected in different environmental conditions. The process of detection is subject, to two main applications, appearance-based, and motion-based techniques. Visibility such as shape, color, and texture of vehicles comprise in appear based technique, whereas the motion-based technique is the moving characteristics of vehicles from the stationary background image. The system contains two frames, enabling to differentiate in static and moving positions. Frame differencing is also detecting form the existing object and that, of a moving object in its position. After identifying the two different positions in frames. This is to further identify and define the position of objects and its movements. The author has contended that the outcome of results, after the experimental frame differencing method as a drawback because of its changes in contour. Shiva kamkar[5. Vehicle detection is generally performed based on a video monitoring system. By using an active base model(ABM) identified by the candidates in a video frame. These model benefits obtained in form of edge information and candidate template matching and verify it by reflection symmetric metrics, rotating it in a clockwise and anticlockwise direction and its symmetric value. If the condition is correct then, it is treated as accepted or otherwise it rejects. Counting and classification of vehicles are streamlined by using only a single counting line. There are two consecutive frames, in which the center of each vehicle frame is before and after the line simultaneously and the counting of the vehicle is done at this stage. The procedure for a random trained method is used for the classification of vehicles. Taking into account, the images of vehicles is calculated and correlated in TSI and that of the GLCM matrix is used to train in an RF classification and categorizing as small, medium, and large.

A.P.Shukla[6]: An analysis in various techniques of the On-Road vehicle System is the main part on the motion model technique. Tracking of vehicle detection is applied in the area such as traffic analysis and incident detection and approaches in relation to detection, segmentation, and tracking of vehicles. The flow of traffic is considered as very important, for analysis along with background conditions such as road signals, people, animals, weather conditions, etc for a clear-cut understanding with minimal traffic congestion problems. object detection is categorized under homogeneous and heterogeneous. The first one is termed as the hypothetical synchronized flow of traffic following speed and time, whereas the second one is termed as unsynchronized and unregulated. The main approaches are put, forth in the detection and segmentation of vehicles areas: background subtraction, feature-based method, frame differencing, and motion-based method. Vehicle tracking is a challenging task to identify the vehicle in its dynamic scene in videos. In order to overcome these obstacles, the following tracking methods are proposed. They are region-based,3D model-based, feature-based, color and pattern-based methods.


Techniques used in this paper is designed to process the tracking of cars, mainly based on the different modules namely conversion of frames, preprocessing, motion segmentation, feature extraction, and clear objects and tracking. In the beginning, videos are taken as an input to the system which is further converted into a number of frames, then preprocessing is performed by employing a method of the grayscale image in order to rectify the color of the RGB image, In the next process, the background subtraction technique is used to get the foreground image. By using blob analysis, the region and area of the image are identified in a clear object, so that extra noises are removed in the foreground object. The location of the moving object is done through tracking. Further target objects are obtained by consecutive video frames. Here the type of the object is identified as car and object detection in each frame are displayed.

About KSRA

The Kavian Scientific Research Association (KSRA) is a non-profit research organization to provide research / educational services in December 2013. The members of the community had formed a virtual group on the Viber social network. The core of the Kavian Scientific Association was formed with these members as founders. These individuals, led by Professor Siavosh Kaviani, decided to launch a scientific / research association with an emphasis on education.

KSRA research association, as a non-profit research firm, is committed to providing research services in the field of knowledge. The main beneficiaries of this association are public or private knowledge-based companies, students, researchers, researchers, professors, universities, and industrial and semi-industrial centers around the world.

Our main services Based on Education for all Spectrum people in the world. We want to make an integration between researches and educations. We believe education is the main right of Human beings. So our services should be concentrated on inclusive education.

The KSRA team partners with local under-served communities around the world to improve the access to and quality of knowledge based on education, amplify and augment learning programs where they exist, and create new opportunities for e-learning where traditional education systems are lacking or non-existent.

FULL Paper PDF file:

Multiple Car Detection, Recognition and Tracking in Traffic



S. Sushmitha, N. Satheesh, and V. Kanchana,




Multiple Car Detection, Recognition, and Tracking in Traffic

Publish in

2020 International Conference for Emerging Technology (INCET), Belgaum, India, 2020, pp. 1-5,



PDF reference and original file: Click here


+ posts

Somayeh Nosrati was born in 1982 in Tehran. She holds a Master's degree in artificial intelligence from Khatam University of Tehran.

Website | + posts

Professor Siavosh Kaviani was born in 1961 in Tehran. He had a professorship. He holds a Ph.D. in Software Engineering from the QL University of Software Development Methodology and an honorary Ph.D. from the University of Chelsea.

Website | + posts

Nasim Gazerani was born in 1983 in Arak. She holds a Master's degree in Software Engineering from UM University of Malaysia.