Robust visual tracking using a fixed multi-camera system

2008

Vision based tracking of an object using the ideas of perspective projection inherently consists of nonlinearly modelled measurements although the underlying dynamic system that encompasses the object and the vision sensors can be linear. Based on a necessary stereo vision setting, we introduce an appropriate measurement conversion techniques which subsequently facilitate using a linear filter. Linear filter together with the aforementioned measurement conversion approach conforms a robust linear filter that is based on the set values state estimation ideas; a particularly rich area in the robust control literature. We provide a rigorously theoretical analysis to ensure bounded state estimation errors formulated in terms of an ellipsoidal set in which the actual state is guaranteed to be included to an arbitrary high probability. Using computer simulations as well as a practical implementation consisting of a robotic manipulator, we demonstrate our linear robust filter significantly outperforms the traditionally used extended Kalman filter under this stereo vision scenario.

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.

A multi-camera monitoring system for online recognition of moving objects is considered. The system consists of several autonomous vision subsystems. Each of them is able to monitor an area of interest with the aim to reveal and recognize characteristic patterns and to track the motion of the selected configuration. Each subsystem recognizes the existence of the predefined objects in order to report expected motion while automatically tracking the selected object. Simultaneous tracking by two or more cameras is used to measure the instant distance of the tracked object. A modular conception enables simple extension by several static and mobile cameras mechanically oriented in space by the pan and tilt heads. The open architecture of the system allows additional subsystems integration and the day and night image processing algorithms extension.

VIPSI-2006 VENICE, 2006

In this paper we address the problem of tracking features efficiently and robustly along image sequences. To estimate the undergoing movement we use an approach based on Kalman filtering. The measured data is incorporated by optimizing the global correspondence set based on an efficient approximation of the Mahalanobis Distance (MD). Along the image sequence, to deal with the incoming and previously existing features a new management model is considered, so that each occluded feature may be kept on tracking or it may be excluded depending on its historical behavior. This approach handles adequately occlusion, disappearance and (re)appearance of features while tracking efficiently movement in the image scene. It also allows feature tracking in long sequences at low computational cost. Some experimental results are presented.

It is important to maintain the identity of multiple targets while tracking them in some applications such as behavior understanding. However, unsatisfying tracking results may be produced due to different real-time conditions. These conditions include: inter-object occlusion, occlusion of the ocjects by background obstacles, splits and merges, which are observed when objects are being tracked in real-time. In this paper, an algorithm of feature-based using Kalman filter motion to handle multiple objects tracking is proposed. The system is fully automatic and requires no manual input of any kind for initialization of tracking. Through establishing Kalman filter motion model with the features centroid and area of moving objects in a single fixed camera monitoring scene, using information obtained by detection to judge whether merge or split occurred, the calculation of the cost function can be used to solve the problems of correspondence after split happened. The algorithm proposed is validated on human and vehicle image sequence algorithm proposed achieve efficient tracking of multiple moving objects under the confusing situations.

Robotica, 2005

In this paper an algorithm for real-time estimation of the position and orientation of a moving object using a video camera is presented. The algorithm is based on the extended Kalman filter which iteratively computes the object pose from the position measured in the image plane of a set of feature points of the object. A new technique is proposed for the selection of the optimal feature points based on the representation of the object geometry by means of a Binary Space Partitioning (BSP) tree. At each sample time, a visit algorithm of the tree allows pre-selecting all the feature points of the object that are visible from the camera in the pose predicted by the Kalman filter. A further selection is performed to find the optimal set of visible points to be used for image feature extraction. Experimental results are presented which confirm the feasibility and effectiveness of the proposed technique.

Proceedings of the 2005 IEEE International Conference on Robotics and Automation, 2005

In this paper, two real-time pose tracking algorithms for rigid objects are compared. Both methods are 3D-model based and are capable of calculating the pose between the camera and an object with a monocular vision system. Here, special consideration has been put into defining and evaluating different performance criteria such as computational efficiency, accuracy and robustness. Both methods are described and a unifying framework is derived. The main advantage of both algorithms lie in their real-time capabilities (on standard hardware) whilst being robust to miss-tracking, occlusion and changes in illumination.

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.

Infrared Physics & Technology, 2007

In this paper, we deal with the problem of real-time detection, recognition and tracking of moving objects in open and unknown environments using an infrared (IR) and visible vision system. A thermo-camera and two stereo visible-cameras synchronized are used to acquire multi-source information: three-dimensional data about target geometry and its thermal information are combined to improve the robustness of the tracking procedure. Firstly, target detection is performed by extracting its characteristic features from the images and then by storing the computed parameters on a specific database; secondly, the tracking task is carried on using two different computational approaches. A Hierarchical Artificial Neural Network (HANN) is used during active tracking for the recognition of the actual target, while, when partial occlusions or masking occur, a database retrieval method is used to support the search of the correct target followed. A prototype has been tested on case studies regarding the identification and tracking of animals moving at night in an open environment, and the surveillance of known scenes for unauthorized access control.

Multiple object tracking in image sequences is an emerging topic of research in the field of image processing and computer vision. This paper presents a robust tracking algorithm for multiple moving object detection and tracking them in dynamic scenes. Our algorithm consists of two important steps. Firstly adaptive GMM background modeling with noise reduction is used for foreground object segmentation in the noisy scenes. Also a background updating method is applied to correct the detection of foregorund objects that bends to the background instantly. Secondly, our object tracking framework uses Extended Kalman Filter to model the nonlinear dynamics and measurement models. Also a new approach for solving data association problem is introduced to determine appropriate association between objects and Kalman Filters which is necessary in multipleobject scenario. The adaptive GMM with noise reduction greatly reduced the false detections in the scenes. The experiment of our proposed algorithm shows good results for multiple object motion tracking in complex dynamic background.