REAL-TIME RECOGNITION AND TRACKING OF MOVING OBJECTS

In recent years, surveillance systems have gained increased importance in order to increase the safety and security of people. These systems have applications in various domains like home or bank security, traffic monitoring, defense; and in public places like railway stations, malls, airports, etc. Our goal is to develop an intelligent real-time surveillance system that can help in increasing the efficiency of the system. In order to cover a large area, we need to install more number of cameras that leads to more number of videos that are to be monitored simultaneously. This in turn increases human intervention and makes it error prone. Therefore, it is of utmost importance to automate the complete system. In the proposed system, cameras are placed in such a way that there is a significant overlap between the field of view of the cameras. This helps in establishing an association between the cameras. The proposed real time surveillance system detects and tracks the objects in motion and provides automatic warning in case of suspicious activities such as unidentified object and restricted zone monitoring. The system provides different views in different cameras, thereby helping in resolving the occlusion in a simple and novel way. The algorithm has been tested on a CPU platform (Intel dual core with 4 GB RAM) with four wireless IP cameras (1.3 MP EDiMAX camera with night vision support) with a processing rate of 10 fps. This system can handle up to eight cameras on a GPU platform (NVIDIA GeForce GTX 480) with frame rate of 25fps.

International Journal of Computer Applications, 2017

Computer Vision (CV) concentrates on the automatic extraction, examination and comprehension of valuable data from a solitary image or a group of images. Object tracking, one of the key areas in CV has received a lot of attenstion in recent times. Tracking objects is a systematic process of monitoring the movement of a target object from its initial state to the nth state over a period of time using a camera. This technique is usually employed as a security feature in both military and civilian systems. However, prior studies has shown that tracking objects in motion is a very difficult task and is a hot research hotspot in the field of computer vision and machine learning. In this review paper we discuess various techniques in detection, tracking and some other related works of moving objects in video streams.

In the presented paper Real time object is tracked from real time video. So, many algorithm are available to track an image but KLT tracker which is proposed by Kanade, Lucas and Tomasi is widely used for object tracking. To track an Object in a video that has been selected by the user in the first frame, the user may select a particular region or an object with a rectangular box in the first frame of the video and that object has to be used as reference object or image and algorithm will track the object or an image for rest of the video sequence.

IOSR Journal of Engineering, 2013

Video surveillance[1] is becoming increasingly important for the security-sensitive areas such as airports, banks, casinos, and parking lots. Many efforts have been made in this field,to detect and track the human and human activities in scene and recognize simple motion such as walking and running. In this paper we present a autonomous real-time surveillance[3] system framework for recognizing complex activities in. In our system we use javacv and opencv to detect moving objects in the scene, and use tracking algorithm to record activities for further analysis. Compared with the traditional surveillance system, an autonomous system surveillance offers much better flexibility in video content processing and transmission[5]. At the same time, it, also, can easily implement advanced features such as real time transmission of video and audio if connected through known ip address I.

2006

This paper presents the multi-camera tracking component of a complete surveillance system that was developed as part of the AVITRACK project, involving 10 academic and industrial partners. The aim of the project is to automatically recognise activities around a parked aircraft in an airport apron area to improve the efficiency, safety and security of the servicing operation. The multi-camera tracking module takes as input per-camera tracking and recognition results and fuses these into object estimates using a common spatio-temporal co-ordinate frame. The multicamera localisation and tracking of objects is evaluated for a range of test data.

2016

Tracking objects in video sequences is a central issue in many areas, such as surveillance, smart vehicles, human-computer interaction, augmented reality applications, and interactive TV etc. It is a process that always involves two steps: detection and tracking. A common approach is to detect the object in the first frame and then track it through the rest of the video. Video Tracking is the process of locating a moving object over time using a camera module. The objective of video tracking is to associate target objects in consecutive video frames. The association can be especially difficult when the objects are moving fast relative to the frame rate. Another situation that increases the complexity of the problem is when the tracked object changes orientation over time. For these situations video tracking systems usually employ a motion model which describes how the image of the target might change for different possible motions of the object. In this project an algorithm is propo...

The International Conference on Electrical Engineering

A robotic vision system has been designed and analyzed for real time tracking of maneuvering objects. Passive detection using live TV images provides the tracking signals derived from the video data. The calibration and orientation of two cameras is done by a bundle adjustment technique. The target location algorithm determines the centroid coordinates of the target in the image plane and relates it to the aim point in the object plane. The stereoscopic images provide the information, from which the range, r of the object can be determined. The azimuth, and elevation, of the target with respect to a certain origin are determined by correlating the x-y displacements of the centroid in the image plane with the angular displacement of the target in the object plane. The servo drive signals for both the robot motion and the angular positioning of the cameras are derived from the image processing algorithm that keeps the centroid of the target image in the center of the frame and the target in line with the axis of the optical system. Hence, the spherical coordinates of the target are defined and updated with every TV frame. The time development of the centroid in successive TV frames represents the real time trajectory of the target path. A non-linear prediction technique keeps the target within the aim zone of the tracking system. In order to minimize the image processing time, i.e. kept within the demand of real time operation, one TV frame time, an image segmentation process is made to subtract nearly all redundant background details.

Multisensor Fusion, 2002

The problem of detecting and tracking of non-rigid objects, as well as classifying their behavior in complex and cluttered environments, can take advantage from using multiple cameras to collect information to be fused either at the remote control center or at intermediate, local hubs. In this paper, multicamera configurations and processing schemes of surveillance systems will be explored by paying attention to their influence with respect to the choice of appropriate fusion methods and algorithms oriented to scene interpretation. In particular, signal representation levels will be described at which data fusion can occur in a system oriented to detect objects and their behavior. Moreover, two different categories of fusion methods will be identified together with their requisites and technical constraints: fusion methods necessary for cameras with partially overlapped field of views and fusion methods for cameras monitoring different local areas of a global environment. Multisensor fusion methods will be presented that are based either on computationally efficient search for correspondences in multisensor data or on global statistical modeling. Possible synergies of fusion algorithms with other components of a surveillance system will be highlighted: constraining effects on representation levels at which fusion may occur.will be shown, which are coming from the choice of transmission modalities over open multimedia networks as well as from the strategies of distribution of intelligence among multiple computational units. Examples will be presented of specific multisensor fusion systems used for detecting objects and their behaviors in surveillance systems operating in the transport field. Finally, possible trends in distributing fusion algorithms among available computational resources will be introduced with particular attention to home surveillance.

Real-Time Imaging, 1998

Real-Time Tracking of Moving Objects with an Active Camera T his article is concerned with the design and implementation of a system for real-time monocular tracking of a moving object using the two degrees of freedom of a camera platform. Figure-ground segregation is based on motion without making any a priori assumptions about the object form. Using only the first spatiotemporal image derivatives, subtraction of the normal optical flow induced by camera motion yields the object image motion. Closed-loop control is achieved by combining a stationary Kalman estimator with an optimal Linear Quadratic Regulator. The implementation on a pipeline architecture enables a servo rate of 25 Hz. We study the effects of time-recursive filtering and fixed-point arithmetic in image processing and we test the performance of the control algorithm on controlled motion of objects.

2013

Recent investigations have shown the advantages of keeping multiple hypotheses during visual tracking. In this paper we explore an alternative method, which presents the concepts of histogram matching technique and absolute frame subtraction to implement a robust automated object tracking system. The object is later tracked using discrete kalman filter technique. Such a tracking system reduces the computation time by a factor of 10 w.r.t. using other methods of segmentation.