Vision processing method and device for welding line auto-tracking

摘要

A vision processing method and device which processes a photographed-object image to provide information about a position and shape of the object, and perform the steps of forming a shape-memory contour model having a tendency to mention its original shape, the model being capable of changing its shape and position by an exterior operation; giving an outline of the shape-memory elastic contour model on the basis of an image of a photographed object; displacing the outlined contour model to an area expected to be occupied by the image of the photographed object; modifying the shape-memory elastic contour model in response to an exterior force generated in the image of the photographed object; and determining an image coordinate of the photographed object on the basis of the modified model.