Method for tracking a video object in a time-ordered sequence of image frames

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In the following description, the present invention will be described by way of preferred exemplary embodiments, including a software program.
Those skilled in the art will recognize that an equivalent hardware embodiment may be readily constructed to provide an equivalent apparatus and method.
The user is considered to be provided with a digital sequence of frames which are shot by a moving camera and then stored in digital format.
The camera can zoom in and out and translate in space.
In addition, the objects in the scene that is being shot can move forward or backward with respect to the camera.
Such movements of the camera and the object cause the object size to become smaller and bigger.
FIG.
1 shows 10 steps of a method in accordance with the present invention.
Briefly, the steps, which will be explained in further detail below, are as follows.
Step 100 defines acquiring object boundaries in the first and last frames of the subject image sequence and computing a model color histogram.
Step 110 defines an initialization step.
Step 120 defines choosing the current frame and two reference frames.
Proceeding from Step 120 to a decision box on the question of whether the current frame equal to the last or the first frame: if YES, the flow proceeds to Step 190 which defines piecewise linearly interpolating the closest shape and location estimates for finding the shape and the location of the object in the unprocessed frames in between the first and last frames.
If NO, the flow proceeds to Step 130 which defines predicting the location and marker of the object and the confidence level of tracking in the current frame from those of the two reference frames.
Step 140 defines selecting a reference template for matching from the reference frames.
(Step 150) defines the step of obtaining a second prediction on the location and marker of the object and setting the search window size in the current frame.
Step 160 defines identifying the location, marker and shape of the object in the current frame based on template matching