To improve the rendering quality of light field images, a high-quality light field image rendering method based on optimized optical flow is proposed and validated. First, an adaptive off-axis camera is constructed using computer graphics technology to acquire texture and depth data (RGBD) of the model. Then, an RGB image pyramid is built through down-sampling, and optical flow is iteratively calculated based on this pyramid. Subsequently, a disparity threshold mechanism is established according to depth information to clip the optical flow, thereby improving the estimation accuracy under large displacements. Meanwhile, an optical flow similarity filtering matrix (SFM) is constructed to select the optimized flow with minimal reprojection error, thereby enhancing both the precision and continuity of the optical flow. Finally, the occlusion relationship is calculated using depth information and optical flow, and the light field image is generated by combining interpolation weights. Based on this method, an integral imaging-based light field display system is implemented to achieve 3D display. Experimental results confirm that within a ±18.9° field of view, the proposed method achieves a 20.7% improvement in structural similarity (SSIM) and a 67.5% increase in peak signal-to-noise ratio (PSNR) compared to conventional optical flow rendering techniques.