The Augmented Lagrange Multiplier for robust visual tracking with sparse representation

Sparse representation has been used to visual tracking to find the object with the minimum error from the target template subspace, which leads to the L1 trackers as it needs to solve an L1 norm related minimization problem. While L1 trackers showed high tracking accuracies, they are expensive computational costs owing to numerous calculations for L1 minimization. This paper aims at researching an L1 tracker that not only runs in real time but also has better robustness than others. First, we develop a new method based on Augmented Lagrange Multiplier. This convex solvers provide a solution to real-world, time-critical applications such as visual tracking. Second, a new L1 norm related minimization model is proposed. The tracking accuracy is improved. The improved tracking accuracy and time efficiency of the proposed tracker is validated with several state-of-the-art trackers on challenging benchmark sequences.