The invention relates to the technical field of computer vision, and particularly discloses a gesture recognition method based on a fixed background. The gesture recognition method comprises the following steps that S1, a gesture recognition background is fixed as a black background, and black background modeling is carried out; S2, an image generated when a hand of people extend into is obtained,and based on an established background model, a foreground image of the hand of people is obtained by using background subtraction; S3, convex hull and convex defect detection is carried out on the hand of people in the foreground image, and a current general gesture of the hand of people is obtained; S4, a framework of the hand of people is obtained by utilizing an image thinning algorithm, a basic structure of the hand of people is obtained, and thus the current closing degree of the hand of people is estimated; and S5, a palm of a robot is controlled to close according to the current closing degree of the hand of people. The gesture recognition method has the beneficial effects that system calculation amount is low, the requirement for hardware configurations is low, fine recognition and gesture estimation of hands of people are effectively achieved, and the closing degree of hands of people can be recognized.