Abstract: The visual measurement accuracy requirements for quadruped robot platforms in multi-objective collaborative scenarios are relatively high. This paper addresses the challenge of coordinate system unification in a three-camera vision system composed of a pan-tilt binocular camera and a monocular hand-eye camera, and proposes a global calibration method for the three-camera vision system based on planar target tensor constraints. Firstly, the Zhang calibration method is used to calibrate the internal and external parameters of the three cameras respectively. Then, a unified coordinate system is established with the left camera of a pan-tilt binocular camera as the base coordinate system. The external parameters of the three cameras are optimized according to the three-view tensor constraints. Finally, a three-camera vision measurement system is established based on the principle of three-camera positioning. Experiments have proved that the maximum relative error of three measurements is only 3.18%, and this method is practical and effective in operation.