Pattern recognition system utilizing a plurality of partial standard patterns

• US 4,014,000 A
• Filed: 03/29/1976
• Issued: 03/22/1977
• Est. Priority Date: 03/28/1975
• Status: Expired due to Term

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Ueda, Hirotada, Uno, Takeshi, Ikeda, Sadahiro
• Primary Examiner(s): Thesz, Joseph M.
• Application Number:
  US05/671,048
• Time in Patent Office: 358 Days
• Field of Search: 340/146.3 Q, 340/146.3 H, 340/146.3 AQ, 340/146.3 MA
• US Class Current: 382/203
• CPC Class Codes: G06V 10/443 by matching or filtering

Abstract

A pattern recognition system comprises a picture tube for picking up an object to be recognized, a partial image pattern cutting out apparatus which, in turn, outputs a plurality of partial image patterns cut out from an image pattern obtained from the picture tube, the partial image patterns being output in synchronization with the scanning of the picture tube and a memory storing a plurality of partial standard patterns. Each of the partial image patterns is compared with the respective partial standard patterns and the respective partial image patterns similar to each partial standard pattern are grouped every partial standard pattern. A group of the partial image patterns, each of which is closely spaced from each other in a position of two-dimensional coordinate is selected from each group of the grouped partial image patterns and the respective representative positions of the respective elected groups of the partial image patterns are calculated. The shape and the position of the object to be recognized can be detected from the fact that the respective representative positions are closely spaced from each other in a position of a two-dimensional coordinate.

Claims

1. A pattern recognition system comprising:

image pickup means for sequentially scanning an image of an object to produce first signals representative thereof;

first means, responsive to said first signals, for generating second signals representative of partial image patterns of a two-dimensional pattern of said object as sequentially scanned by said image pickup means;

second means for forming a plurality of two dimensional substandard patterns, each of which approximately corresponds to selected portion of the two-dimensional pattern of said object;

third means, responsive to the scan of said object by said image pickup means, for generating third signals representive of the respective sequential positions of said partial image patterns;

fourth means, coupled to said first means and said second means, for comparing said each partial image pattern with said respective substandard patterns, and for extracting said each partial image pattern which is within a predetermined degree of coincidence with said substandard pattern;

fifth means, coupled to said third means, for storing positions representative of said respective partial image patterns in response to the outputs of said fourth means; and

sixth means, coupled to said fifth means, for grouping said respective partial image patterns to groups of said partial image patterns, in each of which the distances between the positions representative of said partial image patterns are within a predetermined distance from each other in a two-dimensional coordinate system, and for calculating the coordinate of specific positions of said object in the two-dimensional coordinate system from the groups of said partial image patterns.

2. A pattern recognition system according to claim 1 wherein said second means comprises memory means for storing signals representative of the plurality of said substandard patterns.

3. A pattern recognition system according to claim 1 wherein said first means comprises first and second groups of shift registers, said first group of the shift registers coupled to said image pickup means, each of which is serially connected, said second groups of the shift registers coupled to said first group, each of which has output terminals, and being so designed that said partial image patterns, in turn, are cut out from said two-dimensional pattern, and said second and fourth means comprise connecting means coupled to said output terminals of said first grouped shift registers with a relationship such that said respective substandard patterns are formed by connections between said connecting means and said output terminals, resistors coupled to said connecting means and comparing means coupled to said resistors, said comparing means providing threshold levels each of which corresponds to the predetermined degree of coincidence between said partial image pattern and said each substandard pattern.

4. A pattern recognition system according to claim 1 wherein said first means outputs said partial image patterns as a plurality of picture elements, the number of which is equivalent to a number of picture elements of said each substandard pattern, and said fourth means compares the picture elements of said each partial image pattern with the picture elements of said respective substandard patterns in corresponding positions, said fourth means comprising;

signal generating means for generating third signals corresponding to the number of the picture elements which the picture elements of said each partial image pattern disagree with those of said respective substandard patterns; and

comparing means, coupled to said signal generating means for outputting fourth signals in a case where said number of the picture elements is smaller than a predetermined number N_max.

5. A pattern recognition system according to claim 4 wherein said sixth means comprises:

seventh means, coupled to said fifth means, for grouping said partial image patterns to first groups of partial image patterns every substandard pattern, in each of which the distances between the positions representative of said partial image patterns in the coordinate system are within a predetermined distance L_max from each other, and for calculating the average positions representative of said respective first groups; and

eighth means, coupled to said seventh means, for further grouping said grouped partial image patterns to second groups of partial image patterns, in each of which the distances between the positions representative of said partial image patterns in the coordinate system are within a predetermined distance M_max from each other, and for calculating the average positions representative of said respective second groups.

6. A pattern recognition system according to claim 4 wherein said fifth means further includes memory means, coupled to said signal generating means for said third signals, for storing said third signals in response to said fourth signals from said comparing means.

7. A pattern recognition system according to claim 6 wherein said sixth means comprises:

ninth means, coupled to said fifth means, for grouping said partial image patterns to first groups of partial image patterns every substandard pattern, in each of which the distances between the positions representative of said partial image patterns in the coordinate system are within a predetermined distance L_max from each other, and for extracting the respective positions representative of said first groups, each of said respective position representative of said first groups being the representative position of the partial image pattern which has the minimum number of said picture elements corresponding to disagreement; and

tenth means, coupled to said ninth means, for further grouping said grouped partial image patterns to second groups of partial image patterns, in each of which the distances between the positions representative of said partial image patterns in the coordinate system are within a predetermined distance M_max from each other, and for calculating the average positions representative of said respective second groups.

8. A pattern recognition system according to claim 7, wherein said ninth means sequentially compares the positions of said partial image patterns from said third means with the stored positions of the previous partial image patterns in order to discriminate whether the distances therebetween in the coordinate system are within the predetermined distance L_max, and at the same time, discriminate whether the number of the picture elements corresponding to disagreement, in said each partial image pattern is smaller than that in said each previous partial image pattern, whereby said stored positions and said numbers of the previous partial image patterns are renewed.

9. A pattern recognition system accordng to claim 1, further including discriminating means coupled to said sixth means, said discriminating means comprising:

eleventh means for storing predetermined X and Y positions in the two-dimensional coordinate system in order to recognize the portions of said object, and

twelfth means, coupled with said eleventh means, for discriminating that said specific positions are of said object to be recognized in comparison with said predetermined X and Y positions.