中文摘要：

利用局部约束协同表示法改进最小二乘回归子空间分割方法，提出局部强化最小二乘子空间分割方（LSLSR）。该方法通过近邻样本的协同作用强化重构系数使得LSLSR有更好的抗噪能力，结果表明该方法有较高的聚类准确率。

英文摘要：

Subspace segmentation segments the data set into different clusters, which is broadly used in pattern recognition. Authors improve least squares regression subspace segmentation by using locality-constrained collaborative representation. Local strengthen least square regression subspace segmentation is proposed. The proposed method strengthens the reconstruct coefficient by using neighbor data samples, so that LSLSR has the ability to robust against various corruptions and occlusions. The experimental results show that the method can achieve a high clustering accuracy.