In order to improve measurement precision and measurement speed of small size parts

a new kind of precision measurement method based on Support Vector Regression(SVR) for micro size parts was put forward. Taking the tooth shape chain board as the study object

its main parameters were measured. Using a transillumination in this system

the digital image of tooth shape chain board was collected by A102FCCD device

and was input into computer by IEEE1394 digital card. Then

the original gray level image was processed by reducing noise with median filtering

creating a binary image with threshold method

extracting contour and rotating the image. The image rotation can be realized using Hough transform to detect the longest line. After the image processing

the original gray level image with noise was changed into edge information with single-pixel width. Based on the theory parameter calculation of the tooth shape chain and SVR principle

the detected regions were determined according to the proportions of the length and width oftooth shape chain board. By pixel points on edges in the detected regions

the training set of corresponding line section was madeup for the support vector regression to obtain the regression function of every detected line section

which is the expression of the sub-pixel

and then the main parameters of the tooth shape chain board were measured. Finally

the measurement error was analyzed. The measurement results meet tolerance requirement of the part. The measurement precision can reach 2 μm. Theory analysis and experimental results show that the proposed method significantly reduces the effect of the discreteness of CCD and the system noise on the measurement precision

and is characterized by high speed and high precision. At the same time

this method is very robust to image in plane rotation

scale varation and noise.