In this paper, the control of strand width uniformity in extrusion-based additive manufacturing process based on machine vision is studied. Firstly, the images of the strand width are collected frame by frame by a CCD camera. Secondly, through a series of processes of images acquisition, images processing including images filtering, images binarization and information extraction, the useful information of strand width is obtained. Then, the theoretical relationship between the strand width and printing speed is obtained through experimental research, and a control model is obtained. Finally, by using the control model and the strand width obtained from images processing, the printing speed is adjusted to an appropriate value, which eventually led to the stabilization of the strand width. The uncontrolled and logarithmic controlled, are studied in this work. The results show that the logarithmic controlled strand width is more stable than the uncontrolled strand width. Therefore, the instability of strand width in material extrusion-based additive manufacturing process can be effectively solved by machine vision control.