This paper presents a three-dimensional (3D) measurement method for discontinuous specular objects based on feature matching, which can not only avoid the integration errors between discontinuous surfaces, but ensure the high accuracy. First, the image segmentation method and stereo deflectometry technique are applied to separate the discontinuous specular object into several continuous surfaces and calculate the 3D shape of each area independently by slope integration. Then, the binocular vision technology is utilized to obtain the absolute spatial coordinates of the feature points to accurately evaluate the relative positions between continuous surfaces. Meanwhile, a feature matching method combining epipolar constraint and feature descriptor is proposed to overcome the problem of poor matching rate of non-texture specular objects. Finally, the 3D shape of the measured object is reconstructed by combing the topography and absolute spatial position of each surface. Experimental results show that the method can obtain better results without texture matching and can achieve high precision measurement of discontinuous specular object.