Surface Roughness Measurement Based On Co-occurrence Texture Features Of Speckle Pattern Images

The surface roughness is the mostly used parameter to describe the micro-profile of surface in machining. Reasonable evaluation of surface roughness is of significant for the control of products manufacturing quality, analysis of mechanics properties and improvement of manufacturer conditions.In this paper, a new method of surface roughness is presented. This method is developed based on the speckle image processing and texture analysis. Surface roughness emerged in speckle images is extracted using a kind of texture analysis technique rather than the traditional statistic methods. The texture analysis is called Gray-level co-occurrence matrix (GLCM), the Haralick features of which can be used to indicate the surface roughness.To compute the Haralick texture features, a GLCM matrix is firstly to be created, and then the Haralick texture features are computed using matrix algorithms. To improve the computing speed, this paper proposes a novel fast computational method of Haralick texture features, taking the feature extraction of speckle pattern images as an example. The computational method is to use chained hash table to work out the Haralick texture features of speckle pattern images without directly access GLCM matrix. In the paper, the new algorithm speed is compared with the conventional matrix algorithm. The results show that the speed is much improved.The Gaussian weighting idea is used in this paper. According to the different distance between the pixels, the texture feature depended on the different distance between the pixels of the same speckle pattern image is Gaussian weighted. Then the final texture features can describe the nature of speckle pattern more appropriately.Utilizing the Microsoft Visual C++ 6.0 and MATLAB 7.0, the technique of mixed-language programming is used in this paper. By the MATLAB engine, the Microsoft Visual C++ 6.0 can call the MATLAB 7.0 easily. The benefit is that the advantages of both and the resources of network can be made full use of in the scientific research.An experiment system for the surface roughness measurement based on the method proposed in this paper is set up. The result of the experiment indicates that the measurement system has not only the advantage of simple measurement, a lower requirement to the measurement environment and conditions, but also no-touch measurement,fast measurement speed, high-precision etc.