Tracking Apparatus And Tracking Method

A tracking apparatus characterized by comprising:

• image dividing means for dividing input moving image data to generate divided image data;
• extracting means for extracting background image data and different non-background image data from the divided image data;
• feature data calculating means for calculating color higher-order local auto-correlation feature data for the non-background image data;
• norm calculating means for calculating a norm which is the length of a feature data vector from the color higher-order local auto-correlation data;
• target presence/absence determining means for determining whether or not at least part of a target to be tracked is included in divided images based on an output of said norm calculating means;
• target position determining means for collecting adjacent divided images determined to include the target to determine the position of the target;
• adding means for adding the feature data of the collected non-background image data;
• and
• identifying means for identifying the target to be tracked based on a distance between the feature data of the registered target to be tracked and an output of said adding means.

A tracking program characterized by causing a computer to function as:

• image dividing means for dividing input moving image data to generate divided image data;
• extracting means for extracting background image data and different non-background image data from the divided image data;
• target presence/absence determining means for determining whether or not at least part of a target to be tracked is included in the divided images based on an output of said extracting means;
• feature data calculating means for calculating color higher-order local auto-correlation feature data for non-background image data determined to include the target;
• a target position determining means for collecting adjacent divided images determined to include the target to determine the position of the target to be tracked;
• an adding means for adding the feature data of the collected non-background image data;
• and
• identifying means for identifying the target to be tracked based on a distance between feature data of the registered target to be tracked and an output of said adding means.

A tracking program characterized by causing a computer to function as:

• image dividing means for dividing input moving image data to generate divided image data;
• extracting means for extracting background image data and different non-background image data from the divided image data;
• feature data calculating means for calculating color higher-order local auto-correlation feature data for the non-background image data;
• norm calculating means for calculating a norm which is the length of a feature data vector from the color higher-order local auto-correlation data;
• target presence/absence determining means for determining whether or not at least part of a target to be tracked is included in divided images based on an output of said norm calculating means;
• target position determining means for collecting adjacent divided images determined to include the target to determine the position of the target;
• adding means for adding the feature data of the collected non-background image data;
• and
• identifying means for identifying the target to be tracked based on a distance between the feature data of the registered target to be tracked and an output of said adding means.

A tracking method characterized by comprising the steps of:

• dividing input moving image data to generate divided image data;
• extracting background image data and different non-background image data from the divided image data;
• determining whether or not at least part of a target to be tracked is included in the divided images;
• calculating color higher-order local auto-correlation feature data for non-background image data determined to include the target;
• collecting adjacent divided images determined to include the target to determine the position of the target to be tracked;
• adding the feature data of the collected non-background image data;
• and
• identifying the target to be tracked based on a distance between feature data of the registered target to be tracked and the outputted sum.

A tracking method characterized by comprising the steps of:

• dividing input moving image data to generate divided image data;
• extracting background image data and different non-background image data from the divided image data;
• calculating color higher-order local auto-correlation feature data for the non-background image data;
• calculating a norm which is the length of a feature data vector from the color higher-order local auto-correlation data;
• determining whether or not at least part of a target to be tracked is included in divided images based on the norm;
• collecting adjacent divided images determined to include the target to determine the position of the target;
• adding the feature data of the collected non-background image data;
• and
• identifying the target to be tracked based on a distance between the feature data of the registered target to be tracked and the outputted sum.