Abstract: To meet the needs of three-dimensional posture measurement under complex environments, using the non-contact camera measurement method and the specially designed observation equipment for winter sports, the data of 8 ski jumpers during the training process were collected. Firstly, the joint points of the athletes were identified and tracked based on a deep learning improved method. And then, the studies on athletes' attitude measurement and movement analysis were performed. Finally, the influence of the take-off speed of the athletes on the flight distance and the law of change of the main joint angle were analyzed. The results show that the take-off speed is an important factor affecting the flight distance. At the same time, the law of limb angle changes from takeoff to the initial flight stage has been obtained. The method and equipment can provide a new way for the measurement of athletes' posture and equipment motion parameters of various sports, and the observation results can provide the coaches with refined data to guide the athletes' technical movement.