In this study, we explore the projective synchronization of quaternion-valued competitive neural networks with multiple time scales (QVMTSCNNs), while analyzing the impacts of discontinuous activation functions and time delays. To achieve the control goal, two novel quaternion controllers are designed, which do not depend on the ratio of the fast and slow time scales. By applying the nonsmooth analysis and quaternion inequality techniques, two novel theorems for projective synchronization of QVMTSCNNs are derived by non-separating methods. The obtained results in this study are relatively simpler and straightforward, extending some previous findings. Lastly, numerical analyses are executed to substantiate the theoretical conclusions.