This dissertation deals with roughness measurement by using light scattering for in-process use. Light scattering behaviors form various engineering surfaces are investigated based on numerical analysis. Inherent scattering patterns are then described quantitatively and qualitatively according to surface characteristics to be measured. Main emphasis is given to developing design guidelines for optical roughness sensors and also to identifying practical application limits.
In the first place, an optoelectronic measuring hardware system is proposed in which application flexibility can be maximized, i.e., single-hardware/multi-algorithms scheme can be realized for practical roughness measurement. A mathematical scattering model is then established to analyze the scattering behaviors from machined surfaces. The model is based upon Beckmann's electromagnetic wave scattering theory and actual profile data of machined surfaces. Numerical computation is then performed for five basic representative engineering surfaces: sinusoidal, turned-like, Gaussian(normally distributed), non-gaussian, and lapped-like surface. Scattering characteristics from the basic surfaces are identified and pattern recognition algorithms are developed from which actual roughness values can be obtained. Effects of surface waviness on light scattering are also investigated qualitatively. Finally, basic design requirements are discussed for the optoelectronic measuring system proposed.