초록 ▼

A component association tracker is disclosed. The tracker detects, identifies and tracks a target or targets using components extracted from an image. The tracker extracts the components from a window in the image according to edge directions of connected pixels. The tracker associates the component

A component association tracker is disclosed. The tracker detects, identifies and tracks a target or targets using components extracted from an image. The tracker extracts the components from a window in the image according to edge directions of connected pixels. The tracker associates the components with existing tracks from a track file. The tracks are updated according to information from the associated components.

대표청구항 ▼

What is claimed: 1. A method for tracking a target, comprising: receiving an input image including a target having a target position; determining components in a window of the image according to an edge direction of connected pixels, the window being associated with the target position; for all det

What is claimed: 1. A method for tracking a target, comprising: receiving an input image including a target having a target position; determining components in a window of the image according to an edge direction of connected pixels, the window being associated with the target position; for all determined components in the window, determining possible track-to-component associations and the target position that would be derived from each association using tracks in a track file; assigning a weight to each track-to-component association in the window; determining a best set of track-to-component associations in the window and using this to determine current target position; and updating the tracks in the track file based on the determined current target position. 2. The method of claim 1, wherein said determining components in the window is based on said connected pixels satisfying a predetermined threshold. 3. The method of claim 1, wherein said updating the tracks in the track file includes determining velocity of the target. 4. The method of claim 1, wherein updating the tracks in the track file includes setting any track to an established track when a predetermined threshold is satisfied, and deleting said any track when said predetermined threshold is not satisfied. 5. The method of claim 4, wherein setting said any track to an established track occurs when a number of times said any track is associated with a component exceeds a first threshold in a time period, and deleting said any track when the number of times said any track is associated with a components does not exceed a second threshold in the time period. 6. The method of claim 1, wherein said updating the tracks includes deleting a track if there is failure to associate the track with a component for a predetermined amount of time in determining possible track-to-component associations. 7. The method of claim 1, further comprising: generating a track to associate with a component when there is failure to associate the component with a track in the track file in determining possible track-to-component associations. 8. A device for tracking a target, comprising: a processor for receiving an input image including the target having a target position; and a track file including a plurality of predetermined tracks, wherein said processor determines components in a window of the image according to an edge direction of connected pixels, the window being associated with the target position; for all determined components in the window, determines possible track-to-component associations and the target position that would be derived from each association using tracks in a track file; assigns a weight to each track-to-component association in the window; determines a best set of track-to-component associations in the window and using this to determine current target position; and updates the tracks in the track file based on the determined current target position. 9. The device of claim 8, further comprising a memory to store instructions accessible by the processor. 10. A computer program product comprising a recording medium having stored thereon a computer readable program executable by a computer for tracking a target by carrying out: receiving an input image including a target having a target position; determining components in a window of the image according to an edge direction of connected pixels, the window being associated with the target position; for all determined components in the window, determining possible track-to-component associations and the target position that would be derived from each association using tracks in a track file; assigning a weight to each track-to-component association in the window; determining a best set of track-to-component associations in the window and using this to determine current target position; and updating the tracks in the track file based on the determined current target position. 11. A method for tracking a target, comprising: receiving an input image including a target having a target position; determining a plurality of components in the image according to an edge direction of connected pixels within the component; associating the plurality of components with a plurality of predetermined tracks based on the edge direction of said component, to generate a plurality of sets of track-to-component associations, wherein at least one track is associated with the target position and each component is associated with no more than one track in a set; assigning a weight to each track-to-component association in a set based on the distance between each track and associated component as related to the target position; determining the best set of track-to-component associations based on the total weight, calculated by adding the assigned weight for each track-to-component association in the set, for one of the sets summing up to a minimum value, wherein the best set determines the current target position. 12. The method of claim 11, wherein said associating includes generating at least one set of track-to-component associations wherein at least one track in the set fails to associate with a component.