A method for measuring and modeling a bidirectional reflection distribution function (BRDF) of the inner surfaces of holes based on endoscope images is proposed, which provides a theoretical basis for hole inner surface defect detection based on machine vision. In this study, an endoscopic imaging reflection model was first established to analyze the relationship between the endoscopic image brightness and its three-dimensional topography. With stainless steel and galvanized threaded holes used as examples, BRDF values of hole inner surfaces were measured by image brightness in the normal direction. Based on the hybrid reflection model, three types of reflection models of the inner surfaces of holes were established. A genetic algorithm was then used to determine the optimal model parameters of different materials and to analyze the model errors. Experimental results show that the inBRDF model based on the Cook-Torrance model can effectively reveal the reflection characteristics of the inner surfaces of holes, and the calculation results of the model are in good agreement with the measurement results. The average fitting error of the two materials is 6.22%, which can accurately describe the reflection characteristics of hole inner surfaces and can be applied to the detection and morphological reconstruction of inner surface defects.