Taking the geometric size of the mobile phone tail plug parts as detection object, a precise measurement method and system of small size, irregular shape parts based on machine vision is proposed. The method includes image acquisition, image enhancement, image registration, edge detection, target line extraction, camera calibration and computational measurement. Aiming at the problems that the traditional scaleinvariant feature transform(SIFT) matching algorithm completely ignores the geometric relationship among different feature points and is prone to more mismatches when searching for matching feature points in the workpiece images with smooth gray change, an improved image registration algorithm is proposed, the contour matching is introduced to acquire image geometric information and constrain the SIFT feature point matching, and the random sample consensus(RANSAC) algorithm is used to remove the influence of noise point pair and precisely estimate the parameters of geometric transformation array. Aiming at the facts that the existing Hough transform fitting line algorithm can easily form pseudopeaks in Hough space for nonlinear edges and affect edge detection accuracy, a new strategy of spatial voting weight allocation in Hough space is designed to suppress the pseudopeak generation. The experiment results show that compared with traditional method, the proposed method improves the accuracy of feature point matching by 12% and the accuracy of line detection by 22%, and the measurement accuracy of the proposed system reaches 0015 mm。