An experimental approach for surface roughness measurement based on the speckle images caused by a laser beam on the milled surfaces and the white-light images of the same surfaces is presented. Since the surface slope at every point of the surface influences the speckle pattern, the surface roughness parameters Rda and Rdq were used for comparison. A CMOS camera, LASER and LED light sources were used for capturing speckle and white-light images of the milled surfaces. From the image pixel intensity matrix, a signal vector was generated and was used for the image metric. It is found that standard deviation and mean of the image signal vector correlate well with Ra, Rda and Rdq values measured by a standard Taylor Hobson surface roughness tester. The correlation was found to be better for speckle images than the white-light images. © Springer Nature Singapore Pte Ltd. 2019.