Development of an Optical-Based Vision System for Surface Finish Assessment Using Fractal Geometry

Assessment of finish quality of machined components has been a major concern of the machine tool industry in an effort to improve quality and productivity. The research presented in this thesis focuses on assessing the finish quality of machined parts. A new approach is formulated to apply fractal geometry to characterize the surface texture formed during machining. Research efforts have been devoted to understanding the need and importance of moving fractal geometry from a purely mathematical domain into a manufacturing domain. In addition to using fractal dimension, a second fractal parameter called lacunarity is introduced to characterize the space filling, or the slope distribution of surface texture. This addition represents a unique contribution of this thesis research for improving the precision of surface characterization.A vision system, in which optical devices are employed, is designed and built to implement the surface finish assessment using fractal geometry. The developed prototype serves as a testbed for this thesis research. The testbed employs a CCD (Charged Coupled Device) camera to capture the image of a machined surface. A computer software tool is developed and implemented to process the image data and extract information characterizing the surface condition by using the proposed fractal geometry approach. The system provides 3-D visualization of surface topography and displays numerical values of the surface characterization indices including the fractal dimension and the lacunarity for the purpose of performing diagnostics. Special efforts have been made to investigate effects of the reflectivity of part material on the optical area-based surface characterization technique. Significant findings include the effectiveness of using light filters in image data acquisition and the utilization of light filters to facilitate the system calibration for machining a variety of materials through compensation