Rough surface description using photometric stereo

We present a framework for the measurement, description and discrimination of rough surfaces. The measurement technique, photometric stereo (PS), is a fast, inexpensive, non-contact technique for the estimation of shape. We assess its suitability for the shape estimation of rough surfaces and discuss the consequences of using surface derivatives for surface description. Surface models from the literature are adapted and compared with test surfaces. A method that uses frequency and directional information to discriminate surfaces is presented and demonstrated. We found that PS is an accurate method of measurement for Lambertian surfaces and that, provided the optical properties of the camera are accounted for, the estimated surface derivatives are useful for surface description and discrimination.