Method for recognizing a printed circuit board fiducial mark in order to decide origin point in chip mounter

• US 5,953,447 A
• Filed: 02/27/1997
• Issued: 09/14/1999
• Est. Priority Date: 02/28/1996
• Status: Expired due to Term

• Current Assignee: Bellwave Co., Ltd.
• Original Assignee: Daewoo Electronics (Government of South Korea)
• Inventors: Jin, Sang-Hun
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): PATEL, KANJIBHAI B
• Application Number:
  US08/808,503
• Time in Patent Office: 929 Days
• Field of Search: 382/145, 382/147, 382/151, 348/87, 348/126, 250/559.3, 356/375
• US Class Current: 382/145
• CPC Class Codes: G06V 10/42 Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
• CPC Class Codes: G06V 10/752 Contour matching
• CPC Class Codes: H05K 1/0266 Marks, test patterns or identification means

Abstract

Disclosed is a method for recognizing a printed circuit board fiducial mark in order to decide an origin point in a chip mounter so that electronic parts can be mounted on a desired position of the printed circuit board. Since the method can be used for a distorted image, it can exactly recognize a damaged object as well as a general object. The method can also reduce a calculation amount relative to a polarity number of a candidate object when the candidate object has the polarity number. The method can recognize a candidate object of the same pattern fiducial mark regardless of its size.

Claims

1. A method for recognizing a fiducial mark on a printed circuit board in order to decide an origin point in a chip mounter, said method comprising the steps of:

a) storing fiducial mark recognition data including at least reference distance data from a center point to each boundary pixel of the fiducial mark, a total number of the reference distance data, a polarity number of the fiducial mark and a similarity threshold value, to form a reference database in a data storing means;

b) taking an image of a test area which is a portion of the printed circuit board, where the test area is wide enough for including one or more candidate objects and the fiducial mark is a member of the candidate objects, and transforming the image into digital data;

c) successively searching into the digital data of the image of the test area to meet all the candidate object;

d) calculating coordinates of boundary pixels of the candidate object which has been met during the searching step c);

e) calculating a coordinate of a center point of the candidate object from the coordinates of the boundary pixels of the candidate object, and then calculating distance data, whose number is equal to the total number of the reference distance data, from the center point to the boundary pixels;

f) calculating a similarity degree between the candidate object pattern and the pattern of the fiducial mark by means of a similarity function, where a particular similarity function is used for calculating the similarity degree according to the polarity number of the fiducial mark, and the distance data of the candidate object and the reference distance data of the fiducial mark are used by the similarity function for calculating the similarity degree; and

g) verifying whether a pattern of the candidate object is similar to a pattern of the fiducial mark by comparing the similarity degree which has been calculated with the similarity threshold value.

2. The method as claimed in claim 1, wherein the center point of the candidate object is calculated by following equations, ##EQU7## where (C_x, C_y) is the coordinate of the center point of the candidate object, (X_i, Y_i) is the coordinate of the boundary pixels of the candidate object, and N is a total number of the boundary pixels of the candidate object.

3. The method as claimed in claim 1, wherein the center point of the candidate object is calculated by following equations:

##EQU8## where (C_x, C_y) is the coordinate of the center point of the candidate object, (X_si, Y_si) is a coordinate of all the pixels forming the candidate object, and N_s is a total number of all the pixels forming the candidate object.

4. The method as claimed in claim 1, wherein the distance data from the center point to the boundary pixels of the candidate object is calculated by a following equation, ##EQU9## where each m_n is the total number of boundary pixels each of whose azimuth angle θ

_i viewed from the center point of the candidate object satisfies n.increment.θ

<

θ

_i ≦

(n+1).increment.θ

, .increment.θ

is a unit angle of the candidate object, n is natural number whose value is equal to or less than 360/.increment.θ

, each l_i is distance from the center point of the candidate object to each of the boundary pixels thereof, dist(n) is an average distance of the distance data l_i.

5. The method as claimed in claim 1, wherein, in the case where the fiducial mark is a circle mark whose the polarity number is infinite, the similarity function is as followings, ##EQU10## where SD1 is the similarity degree between the candidate object and the fiducial mark, OBJ(n) is the distance data of the candidate object, and DB(n) is the reference distance data of a circle fiducial mark.

6. The method as claimed in claim 1, wherein, in the case where the fiducial mark is a polygon whose the polarity number is equal or more to three and less than infinite, the similarity function is as follows,

7. The method as claimed in claim 1, wherein, in the case where the fiducial mark is a free mark whose the polarity number is one, the similarity function is as followings,

8. The method as claimed in claim 1, wherein in the verifying step when the calculated similarity degree is equal to or less than the stored similarity threshold value, the candidate object which has been met is taken as the fiducial mark to find, and when the calculated similarity degree is greater than the stored similarity threshold value, the candidate object which has been met is not taken as the fiducial mark to find.

9. The method as claimed in claim 1, further comprising the steps for preliminary test of:

storing an allowable range of pixel number in the data storing means;

calculating a total pixel number of the candidate object;

comparing the total pixel number of the candidate object is within the allowable range of pixel number to find out whether the candidate object is a suspected fiducial mark; and

discarding the candidate object not to calculate the similarity degree when the total pixel number of the candidate object is out of the allowable range of pixel number, and embracing the candidate object to calculate the similarity degree when the total pixel number of the candidate object is within the allowable range of pixel number.

10. The method as claimed in claim 1, wherein

storing an allowable range of a rectangular size and an allowable aspect ratio range of a rectangular in the data storing means;

calculating a width, a length and an aspect ratio of a rectangular within which the candidate object is just fitted;

comparing the width, the length and the aspect ratio of the rectangular is within the allowable range of the rectangular size and the allowable aspect ration to find out whether the candidate object is a suspected fiducial mark; and

discarding the candidate object not to calculate the similarity degree when the width, the length and the aspect ratio of the rectangular is out of the allowable range of the rectangular size and the allowable aspect ratio, and embracing the candidate object to calculate the similarity degree when the width, the length and the aspect ratio of the rectangular is out of the allowable range of the rectangular size and the allowable aspect ratio.