Research On Concentricity Detection Method Of Micro-speakers Based On Support Vector Machine

Measurement technology is the base for industrial development. Precision machining and measurement technology of development are closely related. Geometric measurement technology is interrelated to the development of science.Now, concentricity parameters measured has been widely used in a number of fields. Concentricity parameter of speaker is a key index of its performance. The traditional measuring instrument can not satisfy the measurement requirements of small size. It can not be directly used for the micro-speakers. Accordingly, this paper design a set of methods can be used for micro-speakers concentricity detection.In the process of concentricity detection, if the edge of inside and outside circles obtained, concentricity parameter can be easily attained. So the focus of this paper is to distill edges of inside and outside circles of micro-speakers.The object of study is micro-speakers. The speakers are made in the same material and have the same color. The inside circles rely on bumpy three-dimensional differences to obtain. The image is vulnerable to light and background. The traditional edge detection operator can not be extracted edge effectively. Support vector machine theory is used to detect the image edge and completes distilling edge of micro-speakers. Finally compute concentricity parameters.The main research can be summarized as follows:Firstly, introduce the development of SVM theory. For the small sample, non-linear and high-dimensional pattern recognition problems, SVM show a lot of unique advantages. Besides, SVM applied to image processing were introduced. Secondly, use SVC to distill edge. Simulations show that the method can be only applied to some image whose gray-value chart is suitable for bimodal distribution. Then, combine with edge detections and SVR image features, this paper design edge detection algorithm based on SVR, The simulation results show that the algorithm can extract edges of both inside and outside circles. The research is completed in the Matlab platform, uses Matlab and Libsvm to realize all of image processing algorithm. The experiment has been proved that the algorithm of concentricity detection is reliable.