Robust Visual Correlation Tracking

Algorithm 1

Proposed tracking algorithm.

Input: The th frame video sequence , Initial target position and scale .

Output: Detected target position and scale .

Repeat:

Crop out the searching region in frame according to and , and extract the sample ;

//Position Detection:

() Compute the response with , and using (12);

() Find the target position which maximizes .

//Scale Prediction:

() Extract a sample from at and ;

() Compute the response with , and using (16);

() Find the target scale which maximizes .

//Model Online Update:

() Extract samples and from at and ;

() Update using (13), (14) and update using (9);

() Update using (13), (14) and update using (9).

Until the End of the Video Sequence