提出了一种新的结构稀疏跟踪算法，它不仅利用候选目标及其局部块之间的内在关系，学习它们的联合稀疏表示，而且保留了各候选目标内的局部块中的空间布局结构。(A new structure sparse tracking algorithm is proposed. It not only uses the inherent relationship between candidate targets and their local blocks to learn their joint sparse representation, but also retains the spatial layout structure in the local blocks of each candidate target.)