Presently, teaplucking machine has a disadvantage that it cuts indiscriminately without identification of the tender tea. In order to solve this problem, a kind of teaplucking machine was designed based on machine vision. A method was put forward to cut intelligently fused with position servo, visual servo and levelness servo. The cutting line was kept consistently with tea ridge and the header of machine was consistent with horizontal plane by levelness servo. The initial height of the cutter was set by position servo. In order to make the cutting more precise, PID algorithm was used to obtain highly subtle measurements. In terms of visual servo inspection, firstly, tea images of picking surface were taken and the threshold of B component in RGB was used to eliminate background and segment the range of tea. Secondly, the thresholds of G and G-B components were analyzed to distinguish tender leaves from the image by improved OSTU (the algorithm of threshold automatically extracted according to the maximum deviation). Template matched method and threshold of R component were useful to identify cutter line. Finally, the proportion of tender leaves area above cutter line in the image was calculated and its height was adjusted to ensure the ratio above 70%. Experimental result shows that the proposed method solves present disadvantages of teaplucking machine effectively. Also, the efficiency of picking was improved with reduced labor cost.