The field of rough surface classification is largely focused on optical solutions. There are situations however, where an acoustic method for obtaining information about a surface's texture is required. This thesis investigates the relationship between acoustic reflections from rough surfaces and the resulting signals at a microphone array, and develops methods to find information about the reflecting surface from the array signals in order to build up the knowledge of the object's characteristics for identification. Two methods are explored. The first method exploits the relationship between the distribution of visible image-sources and the surface roughness to achieve an estimation of its parameters. The method requires knowledge of image-sources, so an investigation is carried out into methods for localising image-sources of reflections from a rough surface when presented with microphone array signals and geometry and assuming knowledge of the source signal and geometry. The second method uses an interferometry based method with a tailored source signal in order to probe a surface for a roughness. The contrast of signal power at the microphone array as the centre frequency is varied is explored, and the ability of the resulting contrast profiles in classifying surface roughness is assessed. The methods are assessed using a dataset of reflections from rough surfaces, modelled using BEM, from which microphone array signals can be extracted.