Robot vision positioning method based on feature matching

The invention discloses a robot vision positioning method based on feature matching. The method comprises the following steps: collecting and storing color images of a robot motion scene, and preprocessing the color images of the robot motion scene to obtain image sequence data cut and subtracted by a mean value; calculating characteristic values of all the color images; in a set motion space, a robot using a picture shot by a monocular camera as a query image, cutting an image to be queried, subtracting the mean value of the image to be queried in the step S1, and calculating a feature valueof the query image; and comparing the feature value of the query image with the feature value of the color image, finding out two images with the closest distance and the space coordinates thereof, and obtaining the space position of the robot at the moment. According to the method, only image positioning is used, extraction of artificially designed feature points is avoided, the hardware cost isgreatly reduced, the workl