<p>In this paper, the problem of 3D low-rank tensor completion within the tensor singular value decomposition framework is investigated. We propose a new model with a novel nonconvex tensor rank surrogate function measure under limited sample constraint and bound constraint, where the nonconvex terms have a difference of convex functions structure and can better explore the global low-rank characteristics. A proximal difference of convex functions algorithm is also developed to solve the nonconvex model, whose explicit solution can be obtained under our framework. Using variation analysis tools, we prove that the sequence generated by the proposed algorithm converges to a stationary point under very mild conditions. Comprehensive experimental results demonstrate that the new method is effective.</p>