The definition of perceptual objects is one of the most central but still controversial questions in cognitive science. By using the motion-induced blindness (MIB) paradigm, we found that: a. when two targets were connected, enclosed, or contained the same number of holes, they intended to be grouped together and co-disappeared; b. the targets and background figures with different topological properties significantly weakened MIB effects; c. once observers lost awareness of the targets, the topological changes of the target compared with the non-topological changes, lead it return to awareness much quickly. These results revealed an important role of the topological invariants in the MIB, and provided further support for the topological definition of perceptual objects. Furthermore, current study suggested that the representation of the perceptual new object could be processed in the absence of visual awareness.