Sub-bandgap defect energy levels (SDELs) introduced by the point defects located in surface defect areas are considered the main factors in decreasing laser-induced damage thresholds (LIDTs). The suppression of SDELs could greatly increase LIDTs. However, no available method could detect SDELs, limiting the characterization and suppression of SDELs. Herein, a self-designed photo-luminescence detection system is developed to explore the weak transient-steady photo-luminescence properties of machined surfaces. Based on the excitation laser wavelength dependence of photo-luminescence properties, a sub-bandgap energy-level structure (SELS) containing SDELs is unveiled for the first time. Based on the developed mathematical model for predicting LIDTs, the feasibility of the detection method was verified. In summary, this work provides a novel approach to characterize SDELs on machined surfaces. This work could construct electronic structures and explore the transition behaviors of electrons, which is vital to laser-induced damage. Besides, this work could predict the LIDTs of the machined surfaces based on their PL properties, which provides convenience for evaluating the LIDTs of various optical elements in industrial production. Moreover, this work provides a convenient method for raising the LIDTs of various optical elements through monitoring and suppressing the SDELs on machined surfaces.