A Surface Roughness Measurement Method Based On The Spatial Filtering Of Laser Speckle Pattern Textures

The surface roughness is the mostly used measure to describe the micro-profile of part surface in machining. The surface roughness of part surface has significant impact on their friction coefficient, anti-fatigue, anti-corrosion properties, positioning accuracy and so on. Consequently evaluation of surface roughness is significant for the control of product manufacturing quality, analysis of mechanical properties and improvement of manufacturing conditions. Therefore, surface roughness measurement is very important for production quality control.Most existing measurement techniques, such as the contact stylus method, optical interferometry, microscopy, optical scattering method, laser speckle method and so on, are often used in off-line measurement. A surface roughness measurement method using machine vision imaging, which may will be suitable for on-line inspection, is presented in this thesis. A texture analysis method based on spatial filtering is used to analyze laser speckle patterns of grinding surface to extract roughness information embedded in the surface speckle pattern texture images. Spatial filtering is used to extract three types of vectors based on fractional Brownian motion model (FBM) of window speckle images, which are the normalized scale range vector NSR, the normalized pixel pair number vector NPN, and the normalized multiscale intensity difference vector NMSZD, then speckle patterns are transformed, and finally both the transformed images and the transformed images without zero gray pixels are statistically analyzed. The results show that both texture features energy and new entropy of the transformed images have a good monotonic relationship with surface roughness value Ra.An experiment system for the surface roughness measurement based on the method proposed in this thesis has been set up. The system configuration consists of a laser and a CCD camera only. The experimental results show that the measurement system has the advantage of not only requiring simple measurement, a lower requirement for the measurement environment and conditions than existing systems, but also being no-contact measurement with some anti-interference. If the algorithm could be optimized, most probably, it would be used for an on-line surface measurement. After the measurement system is calibrated by a standard surface roughness component, the surface roughness of a part surface made of the same material and machined by the same method as the standard component surface can be evaluated from a single speckle pattern texture image. The surface roughness measurement method presented in this thesis is a potential approach for on-line surface roughness measurement inspection.