Single cell studies represent a cutting-edge field in life science. For cells growing on a soft substrate, a precise micromanipulation is indispensable in order for high-quality single-cell sampling. The foundation of micromanipulation is the accurate triaxial movement control, of which the major challenge lies in the z-axis positioning. Conventional methods rely on manual operation but are hampered by low efficiency. Resorting to additional sensors and cameras might help, whereas the system configuration becomes inevitably complicated. In this work, a vision-based approach utilizing interface refraction color is reported. By monitoring the color change induced by refraction during the contact process, we realized an accurate movement control of the end-effector. The reported method is capable of detecting the z-axis contact between the capillary tip and the substrate in a fully automated manner. Tested by 70 batches of a total of 3840 images, the proposed technique achieves an overall accuracy of 94.28% with precision of 2.73 μm, sufficing in the applications of single cell sampling. Detecting z-axis contact by interface refraction color marks an innovative computer vision method and it is proposed and validated in this work. Our method holds a promise of being a potent part of a close-loop control mechanism for automated single cell micromanipulation.