<p>Incomplete multi-view clustering aims to utilize the correlation and complementarity between multiple perspectives to fill in the missing information, thereby improving the accuracy and robustness of clustering. Nevertheless, current researches often need to pay more attention to the divergences and high-order correlated information between views. They also have high computational complexity and memory requirements. To tackle these problems, we propose a novel method called Dynamic Anchor-based Tensor Learning for Incomplete Multi-view Clustering (DATL-IMC). Specifically, DATL-IMC designs a consensus anchor graph learning algorithm to obtain the consensus anchor graph by exploiting dynamic anchor learning and applies the low-rank tensor to optimize the anchors further. Meanwhile, DATL-IMC exploits projection learning to learn the latent representation of each view to get the consensus latent representation. Then, DATL-IMC utilizes graph regularization to collaborate the consensus latent representation and the consensus anchor graph to improve the clustering results. Finally, experiments conducted on nine real-world datasets demonstrate that DATL-IMC surpasses fifteen state-of-the-art methods.</p>