摘要 |
PROBLEM TO BE SOLVED: To provide a seal impression character recognition device by a self-learning type fuzzy inference model using a two-dimensional laser displacement sensor, which can reduce the influence of a stain, texture, colors, and ambient light variation of a seal impression part on a recognition result by picking up an image of the raised shape of the external surface of an object and processing the image, can screen object characters to be identified by reflecting the feature quantities of image picked-up characters on a membership function like self-learning, and actualize efficient template matching and is highly robust and speedy. SOLUTION: The three-dimensionally raised shape of a fixed-form character carved in the external surface of an object 1 is picked up as a distance image by using the two-dimensional laser displacement sensor; and three-dimensional shape features of image pickup information are registered in the membership function like self-learning and a character candidate to be identified is screened by using a fuzzy inference model to carry out efficient template matching, thus automatically recognizing carved characters on the external surface of the object 1.
|