Visual Object Tracking Based on Adaptive Background-Awareness and Spatial Constraint

Algorithm 1

BSCF Tracker.

	Input: Image sample , filter model , filter spatial reference template , object box position and scale information.
	Output: The scale and the position information of the target in each frame, the updated filter model , and the updated appearance model.
(1)	initialization:
	The filter space reference template as a Gaussian shape distribution.
(2)	: repeat
(3)	: Extract the image sample of the frame i to form a sample matrix ;
(4)	: Extract the background information samples by using equation (6);
(5)	: The background information samples form a matrix ;
(6)	: Solve the filter model of the current frame according to equation (5);
(7)	: Obtain the model peak response of the filter representing the object location;
(8)	: Obtain the position and scale information of the tracking object;
(9)	: According to equation (13), the filter model parameter is obtained;
(10)	: Update filter model and filter space template .
(11)	: end
(12)	：until End of the video sequence of frame n;