The invention discloses a vision-based user specific behavior
analysis method and a self-service medical terminal, and is characterized in that video images of a plurality of
human body posture samples are sampled, corresponding comparison
feature data of each
human body posture is obtained through calculation and learning, and the comparison
feature data is stored as a comparison feature
database; the method comprises the following steps: calculating
user space-time characteristic data of a
human body posture in an image, comparing the data with data in a comparison characteristic
database, and searching comparison characteristic data with a minimum difference value with the
user space-time characteristic data in the comparison characteristic
database; and when the difference value between the space-time characteristic data of the user and the found comparison characteristic data is smaller than a preset threshold value, judging that the
current user belongs to a human
body posture corresponding to the comparison characteristic data. Specific behaviors of users are taken as priori knowledge through training, and user behavior analysis is realized based on comparison of the prioriknowledge. Through the
system, monitoring is more intelligent, user interaction is more user-friendly, and development of man-
machine interaction is promoted.