<p>Multi-view subspace clustering seeks to enhance clustering performance by optimizing and integrating the graph structure information derived from each view. In recent years, anchor-point-based clustering methods have been significantly developed to address large-scale real-world scenarios. These methods capture data distribution from different views by sampling anchor points. But they typically construct the anchors independently, failing to effectively utilize complementary multi-view information. To overcome this limitation, we propose an innovative approach to subspace clustering under a Laplace rank constraint, which integrates consistency anchors and sparse constraint. Our approach integrates anchor selection with subspace graph construction within a single optimization framework, where the consensus matrix and similarity matrix are optimized each other to enhance clustering performance. Connectivity constraint is incorporated to ensure that connected components directly represent clusters, improving the accuracy of the clustering process. Additionally, to eliminate redundant information hidden in the data, we iteratively update the similarity matrix under sparse constraint. This process allows us to construct a more robust bipartite graph, effectively leveraging the duality between samples and features. Comprehensive experimental results on artificial and real-world datasets show the superior performance of our proposed algorithm.</p>