| CPC G06V 10/469 (2022.01) [G06V 10/72 (2022.01); G06V 10/761 (2022.01)] | 14 Claims |
|
1. A geometric pattern matching method comprising:
determining, by a geometric pattern matching device, information on reference geometric pattern contour points for a learning object on a learning image;
determining, by the geometric pattern matching device, information on detection object contour points for a detection object on a detection image; and
performing, by the geometric pattern matching device, geometric pattern matching between the learning object and the detection object on the basis of the information on the reference geometric pattern contour points, the information on the detection object contour points, and a priority of learning object contour points,
wherein the information on the reference geometric pattern contour points is determined based on the learning object contour points selectively extracted by removing noise from initial contour points extracted from the learning image using change rate magnitude information or change rate direction information of the learning object, the noise comprising connection pixels having a relatively small change rate magnitude as compared to the learning object contour points, the noise removal including connecting connection pixels having a change rate magnitude greater than or equal to a threshold value, determining a connection pixel group having a size less than or equal to a noise determination reference value and removing the connection pixel group,
wherein the information on the detection object contour points is determined based on a plurality of images determined by overlapping the change rate magnitude information, the change rate direction information, or contour point information of the detection object,
wherein the information on the reference geometric pattern contour points is determined by setting, after removing the noise, the priority of the learning object contour points in consideration of a degree of spreading of the learning object contour points and according to change rate magnitudes of the learning object contour points, the spreading including dividing the learning object into a plurality of regions such that a learning object contour point having a largest change rate magnitude is sequentially extracted from each region of the plurality of regions without overlapping the region, and
wherein by increasing a degree of the spreading of the learning object contour points, the learning object contour points reflect an overall geometric structure of the learning object arranged in order of the priority.
|