Research On Key Issues In The Automated Defects Detection Of Sphere Surface

It is challenging to evaluate the surface defects since the defects are of microns while the optics are usually of hundreds of millimeters, which made defect detection is the constraint of surface detection. In the paper, an exploration about the sphere surface defect evaluation system (SSDES) is proposed, and to verify, some simulations and experiments are conducted. The research including 2 parts, auto-alignment of sphere optics and automated classification between dust particles and digs.A conceive about the SSDES is proposed, including the system layout and working flow. And some key blocks are introduced briefly, auto-alignment, dark-field illumination and imaging, multi-axis motion, defect evaluation software included.An auto-alignment system of sphere optics is proposed. Working in the reflected and rotate mode, the system can detect the center of the sphere automatically. The image entropy algorithm, hill-climbing method and curve fitting are combined for auto-focus. Then the error of the system is analyzed, the error in X-axis position is 15μm, Y is 18μm, Z is 8μm.Dark-field scattering microscopy and pattern recognition methodology are combined to classify digs and dusts. The SSDES is employed for dark-field images acquisition of optical samples. Grayscale, texture and morphology analyses are then conducted on each image to extract raw feature data, which are compressed with the PCA (principal component analysis). Based on the compressed feature data, the SVM (support vector machine) is used to construct the classification model.Besides, experiments are conduct to verify the techniques. The comparison between the images before and after alignment proves the validity of the auto-alignment system. And the repetitive experiments show the stability and repeatability. In the dust and dig classification experiment, the success discrimination rates are 96.56% for the training set and 93.90% for the prediction set, respectively. Besides, the classification results are presented to show the potential of this method to be used for practical digs and dusts discrimination o n the actual optical samples.