It is desirable to be able to estimate the light trapping performance of glass in photovoltaic applications prior to depositing the active semiconductor layer onto it. However, conventional methods such as haze, root-mean-square roughness (Rms) are not very effective. Additionally, decoupling the geometries of two surfaces ("surface decoupling") is known to enhance light trapping effectively. In this investigation, clear surface decoupling is observed between the inner surface in contact with an aluminium induced textured (AIT) glass surface and the exposed Si film surface grown on such glass by using cross-sectional focused ion beam (FIB) and scanning electron microscope (SEM) imaging. Results suggest that light travelling from a texture on glass to a texture on the exposed Si film surface growing from the same glass texture gets randomised to some extent depending on the decoupling degree between the two textures. The surface decoupling degree can be characterised by a calculated correlation coefficient (CC) between decoupled surfaces. The CC calculation requires the topographies of both AIT glass and the Si film grown on such glass, with the former extracted from atomic force microscopy (AFM) and the latter calculated by a self-coding model based on the underlying AIT glass topography. This CC is found to form a stronger correlation with light trapping capability of AIT glass compared to the conventional light trapping estimation methods. (C) 2012 Elsevier B.V. All rights reserved.