A reconstruction method is proposed based on the reflection characteristics to measure the defect profile and depth of the inner surface of the hole. In addition to capturing geometric information of the scene, this method decomposes the distortion in the endoscope image into circumferential and axial distortions, thereby improving correction accuracy and enabling accurate measurement of the defect contour. The bidirectional reflection distribution function (BRDF) is established based on reflection characteristic theory, allowing for mapping between image brightness and normal declination. By accumulating tangent values of normal deflection angles, the height of axial contour lines inside the hole can be reconstructed to achieve defect shape reconstruction and depth measurement. This approach overcomes convex-concave ambiguity in shape from shading (SFS), as demonstrated by experimental results. The proposed detection method offers equipment versatility and high measurement efficiency, making it suitable for machine vision-based detection of internal surface defects in holes.