Television – Object tracking
Reexamination Certificate
1998-06-22
2002-08-27
Britton, Howard (Department: 2713)
Television
Object tracking
C348S157000, C348S159000, C700S091000, C702S166000
Reexamination Certificate
active
06441846
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method and apparatus for tracking moving objects in real time whereby trajectories corresponding to the movement of the objects are determined. More particularly, the present invention relates to a method and apparatus for tracking moving objects, such as athletes, in connection with sporting events and exploiting such tracking to derive information corresponding to the movement of the objects being tracked.
2. Description of Related Art
Tracking systems for moving objects typically generate trajectories corresponding to the motion of an object within the view of a camera. The trajectories or tracks typically consist of a sequence of x, y (location) coordinates and time coordinates. The information from these trajectories has a variety of applications. For example, the information can be used to count the number of objects, such as a people or vehicles, crossing a reference line and to associate a particular direction with each crossing. In addition, such trajectories may be used to determine the number of people present within the field of view of a camera at any instant, which information is useful, for example, for product marketing such as determining the effectiveness of a particular advertisement or advertising technique in a store. Tracking systems may also be employed for measuring consumer traffic throughout, for example, the aisles of a store, etc., including the length of time that particular persons spend in specific aisles.
One field in which real time tracking would be particularly desirable, but is not currently greatly utilized, is in the field of sports.
For example, most sports activities are measured in terms of statistics which highlight various aspects of the sport, such as personal performance (number of points scored, fastest time, etc.), comparative performance (e.g., number of points scored versus opponent), and playing strategy (e.g., number of perimeter baskets scored in a basketball game versus number of inside baskets scored).
Conventionally, most such statistics can be readily measured (e.g., number of points scored, time for running a race, number of errors committed). However, other statistics are of interest, but are relatively difficult to obtain readily, and are certainly difficult to obtain in connection with real time tracking of an event. Such statistics may include distance traveled by a player per an interval of time (e.g., a game, set, quarter, etc.), instantaneous and average player speed and acceleration, shot selection (e.g., in hockey or basketball), or areas of a playing field (court, rink, etc.) covered by a player.
Generally, real time tracking of athletes in a sporting event is challenging, especially because it is difficult to obtain a clean segmentation of an athlete from the background in view of changing lighting conditions, variations in clothing worn by athletes (especially with regard to color), differences in the visual characteristics (for example, reflectivity) of playing surfaces (grass, clay, hardwood, ice, etc.), and the fast and dynamic movement of athletes. Another factor complicating tracking is the simple presence of other moving objects or moving people, in addition to the athlete being tracked.
Several methods or systems have been developed for the tracking of moving objects, including people. However, these conventional systems do not yield a single motion region or even a consistent set of motion regions, which deficiencies are exacerbated when tracking athletes in the midst of highly dynamic movement.
For example, in Rashid, R. F., “Towards A System For The Interpretation Of Moving Light Displays”, 2 IEEE Transactions on Pattern Analysis and Machine Intelligence, 574-581 (1980), a method is described for interpreting moving light displays (MLD). In general, Rashid teaches segmenting out from MLD images individual points corresponding to moving people. The individual points are grouped together to form clusters based on, inter alia, the positions and velocities of the individual points; the formed clusters represented individual objects. Tracking is performed by matching points between consecutive frames based on the relative distances between the location of points in the current frame and the location of predicted points in a previous frame. The predicted position is based on the average velocity of the point in the previous frame and the relative distance, which is calculated using a Euclidean function.
The technique described by Rashid has several drawbacks. Specifically, the MLD system requires several frames before a good object separation is obtained, and no criteria is provided for determining when satisfactory object separation has occurred. In addition, no mechanism is provided for propagating the generated clusters to prior and subsequent frames for continuity in the motion representation. This undermines real time operation.
In another tracking system described in Rossi, M. and Bozzoli, A., “Tracking And Counting Moving People”, Proceedings Of The Second IEEE International Conference On Image Processing, 212-16(1994), a vertically mounted camera is employed for tracking and counting moving people. This system operates under the assumption that people enter a scene along either the top or bottom of the image where altering zones are positioned for detecting people moving into the scene. In reality, however, people can also appear in a scene, inter alia, from behind another object or from behind an already-identified person. In other words, people may be wholly or partially occluded upon initially entering a scene and would not be identified by this system. The problem of identifying occluded persons is also present in the system described in Rohr, K., “Towards Model Based Recognition Of Human Movements In Image Sequences”, 59 Computer Vision, Graphics And Image Processing: Image Understanding, 94-115(1994). Such problems are clearly pertinent to real time tracking of athletes during a sporting event.
In addition, the systems described in Smith, S. M., and Brady, J. M., “A Scene Segmenter: Visual Tracking of Moving Vehicles”, 7 Engineering Applications Of Artificial Intelligence 191-204(1994); and “ASSET-2: Real-Time Motion Segmentation And Shape Tracking”, 17 IEEE Transactions On Pattern Analysis And Machine Intelligence, 814-20 (1995), are designed specifically for tracking objects such as moving vehicles, and accordingly identify features representing corners or abrupt changes on the boundaries of the vehicles. This approach is based on that tracking objects which are rigid with unchanging contours, and, thus, permits the use of constant velocity or constant acceleration models, techniques clearly unavailable for tracking of people, particularly when tracking people in motion, particularly the dynamic motion of sports where hands and feet may all move with different speeds and in different directions.
In the related U.S. Pat. No. 5,764,283, an apparatus and method are disclosed for tracking moving objects in real time. In particular, an apparatus and method are disclosed in which local features, such as extrema of curvature on boundary contours, are tracked, and trajectories of motion are derived by dynamically clustering the paths of motion of the local features.
SUMMARY OF THE INVENTION
The present invention provides, most generally, a method and apparatus for tracking moving objects, particularly athletes engaged in sporting activities. More particularly, the invention provides an apparatus and method for obtaining information corresponding to the athlete(s) and/or sporting activity being tracked. In one embodiment of the present invention, such information is performance statistics for an athlete derived from real time tracking of that athlete. In another embodiment, such information is an occupancy map, which may be visually manifested, corresponding to the frequency with which an athlete occupies particular regions of a playing field, or the time spent in particular regions of the playing field. In ye
Carlbom Ingrid B.
Jean Yves D.
Pingali Sarma V G K
Britton Howard
Yacura Gary D.
LandOfFree
Method and apparatus for deriving novel sports statistics... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for deriving novel sports statistics..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for deriving novel sports statistics... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2894768