This study aims to improve the measurement accuracy of liquid film thickness using a liquid film sensor with an optical waveguide film (OWF). The measurement principle of employing the OWF is based on the detection of light reflection at the liquid film surface with high spatial resolution. Because the curved surface of the liquid film reflects light and increases measurement error, we propose a signal processing method to remove the error factor in the calculation of the time-averaged thickness. This method requires prediction of the surface curvature, and we numerically investigated the characteristics of the output signal related to the reflected light intensity. The analysis results showed that the effect of the curved surface up to the surface curvature of 5.0 mm−1 was negligible because the liquid film thickness showed good agreement with that of the flat liquid film surface within 7% accuracy. Furthermore, we consider the applicable range of liquid film thicknesses under the operating conditions of boiling water reactors (BWRs). We estimated the surface curvature of the liquid film based on the calculation of the critical Weber number and confirmed that the curvature caused under the BWR operating conditions was covered by the analysis conditions of this study. Therefore, our proposed method for signal processing via the OWF enabled us to improve the measurement accuracy of the time-averaged thickness with respect to the base film thickness by extracting accurate surface curvature data.