Dynamic deformation measurement of machine parts in fatigue strength test is studied by using machine vision technique. Considering the uncertainty of parts surface, we adopt circular mark to locate the object profile in order to obtain high quality images. Through some image pre-processing with linear filtering, continuous contour searching method and circular detection based on random Hough transform (RHT), the real-time deformation can be measured with image characteristic parameters. In the practical application, the deformation of the loaded bicycle handle-bar is calculated. The test results show that the machine vision measurement is very effective; measurement resolution attains 0.1mm/pixel; the discrete degree of measurement data is low and the system meets the requirement of real-time measurement. The study proves that the measurement method of dynamic deformation based on machine vision is feasible, which can give some help for fatigue strength test of machine part and other structure deformation.