With the rapid development of the new generation of Human Computer Interaction(HCI), more and more methods for body feeling gesture control are also developed. But most of the control modes are limited to specific control scenes with a single control function, and their convenience, the success rate and accuracy rate are lower. To solve these problems, two basic instruction algorithms, continuous instruction algorithm model and Boolean instruction algorithm model, are proposed. A complex control task under various scenarios will be divided into two types of instruction algorithm models mentioned above through three steps: function decomposition, instruction extraction, and model matching. The model takes the 3-D coordinates of the hand and the hand posture as the input conditions to realize the state jump and the instruction output. After practical work and comparison, the proposed control strategy can be implemented in many cases, and the algorithm proposed in this paper has high success rate, accuracy and robustness.