Multi-view　clustering,　which　aims　to　discover　the　complex　correlations　among　multiple　views　and　to　partition　multi-view　instances,　is　an　important　issue　in　both　the　data　mining　and　machine　learning　areas.　It　has　received　great　interest　over　the　past　decades,　and　a　large　number　of　effective　multi-view　clustering　methods　have　been　proposed,　inspired　by　classical　low　rank　theory.　However,　in　most　existing　methods,　sample　data　are　always　vectorized　as　a　column　in　linear　space.　This　conventional　approach　may　lose　its　efficacy　for　some　complex　data,　such　as　video　or　imageset.　This　is　because　these　data　are　often　treated　as　samples　on　a　nonlinear　manifold.　Besides,　traditional　low　rank　based　clustering　methods　always　adopt　a　uniform　representation　matrix　constrained　by　a　single　nuclear　norm.　These　methods　could　potentially　reduce　clustering　accuracy,　as　they　may　result　in　suboptimal　solutions.　In　this　paper,　we　propose　a　new　multi-view　clustering　method　based　on　Multi　Low　Rank　Representation　and　Heterogeneous　Manifolds.　In　the　proposed　method,　multi-view　data　are　constructed　as　sample　data　on　Heterogeneous　Manifolds.　Meanwhile,　we　extend　the　bilinear　low　rank　representation　on　tensors　and　propose　a　tensor-based　multi-linear　low　rank　representation　to　achieve　better　clustering　results.　Experiments　on　four　public　benchmark　datasets　show　that　our　method　achieves　superior　clustering　performance.　©　2024　Technical　Committee　on　Control　Theory,　Chinese　Association　of　Automation.