Abstract: Laser direct imaging (LDI) technology is the core of digital lithography equipment. The imaging accuracy depends on the precise matching between the surface of the printed circuit board (PCB) and the focal plane of the projection lens. Unfortunately, during the long-term operation of the equipment, multiple factors, such as substrate warping, thermal deformation, and mechanical vibration, collectively cause the imaging surface to deviate from its intended surface. The problem seriously deteriorates the imaging quality and alignment accuracy. In this research, a visual self-calibration method for out-of-plane disturbance suppression is proposed. A self-calibration system based on a bionic compliance mechanism is formulated to achieve Z-direction micro-motion focusing of the camera. A front-end dynamic focusing algorithm, integrating the compliant mechanism and target recognition, is designed to achieve local adaptive focusing on the PCB surface. A back-end focal plane calibration model is quantitatively established to analyze the mapping relationship between defocus and imaging accuracy. The development and performance test of the self-calibration system prototype is completed. The experimental results show the performance of the proposed method. The defocus is within ±0.5 mm, and the co-axiality error is below 5 μm. The proposed approach significantly improves the imaging accuracy and alignment reliability of the LDI machine under multiple out-of-plane disturbances.