CN114441554B - Detection method - Google Patents

Abstract

The application provides a detection method, which comprises the following steps: performing binarization processing on the input image related to the printed circuit board; judging a plurality of contour lines belonging to black-white boundaries or image frames in the input image, and defining the contour lines as a plurality of line segment objects; screening each line segment object according to the perimeter threshold value and the times of passing through the image frame to obtain a plurality of candidate line breaking objects; and judging whether the broken line object belongs to a first group of broken lines or not according to the midpoint coordinates of the broken line ends of each candidate broken line object and the first set line distance. Thereby, a detection method capable of detecting a line disconnection position of a printed circuit board without using a standard template and without requiring a large amount of data is realized.

Description

Detection method

Technical field

The present invention relates to a detection method, in particular to a detection method for detecting the line disconnection position of a printed circuit board.

Background technique

Machine vision technology has been widely used in the manufacturing process of printed circuit boards to detect defects in semi-finished products. The existing printed circuit board defect detection methods mainly use the following two technologies: one requires a perfect and defect-free standard sample comparison circuit, which means that a circuit layout (Layout) data file needs to be prepared in advance for comparison and reference; and The other is to use machine learning to make judgments through an image database with a large amount of image data when circuit-related data cannot be obtained or information can only be obtained through images of existing circuit boards. However, whether there are other better detection methods has become a question to be resolved.

Contents of the invention

The object of the present invention is to provide a detection method of an image database that does not require the use of standard templates for comparison and does not require a large amount of data.

Therefore, the present invention provides a detection method, which is suitable for and implemented by a detection system. The detection method includes steps (A) to (D).

In step (A), the input image related to the printed circuit board is binarized, and the input image has an image frame.

In step (B), multiple contour lines belonging to the black and white boundary or the image frame are determined from the binarized input image and defined as corresponding line segment objects.

In step (C), each line segment object is filtered according to a perimeter threshold and the number of times it passes through the image frame, and a plurality of candidate line segment objects are obtained.

In step (D), it is determined based on the midpoint coordinates of the disconnected end of each candidate disconnected object and the first set line distance whether it is classified as belonging to the first group of disconnected objects.

In some implementations, in step (C), when the perimeter of any of the line segment objects is greater than or equal to the perimeter threshold, and the number of times the line segment object passes through the image frame is equal to At 1, the line segment object is determined to be one of the candidate line break objects.

In some implementations, in step (D), each of the candidate wire break objects has a design line width corresponding to the design value, and multiple actual line widths measured at different locations. When one of the When the actual line width is equal to the design line width multiplied by the proportion threshold, the position of the center point of the two endpoint coordinates corresponding to the actual line width on the contour line is equal to the midpoint of the broken end. coordinates, the scale threshold is greater than zero and less than 1.

In some implementations, in step (D), the midpoint coordinates of the disconnected end of each candidate disconnected object and the disconnection point of any other candidate disconnected object are calculated. The midpoint distance of the midpoint coordinates of the line end. When it is determined that the minimum of the at least one midpoint distance is less than the first set line distance, the corresponding two candidate line break objects are classified as Describe one pair of data in the first group.

In some implementations, the detection method further includes step (E), based on the midpoint coordinates of the broken ends of the candidate broken objects that are not classified into the first group. At least one midpoint distance, and a second set line distance, determine whether the candidate disconnected objects that are not classified as the first group are classified as belonging to the second group, the second set line distance greater than the first set line distance.

In some implementations, in step (A), the image frame is a quadrilateral. In step (E), when it is determined that any one of the at least one midpoint distance is smaller than the second set line distance, and the two candidate line break objects corresponding to the said one pass respectively When the image frames are different sides of the quadrilateral, the corresponding two candidate disconnected objects are classified as one pair of data of the second group.

In some implementations, the detection method further includes step (F) of determining that the at least one candidate disconnected object that is not classified as the first group or the second group is classified as Belongs to the third group.

In some implementations, in step (D), the proportion threshold is equal to 85%, and the first set line distance is equal to the minimum line distance of the printed circuit board process corresponding to the input image. The second set line distance is equal to the minimum line distance of the printed circuit board manufacturing process corresponding to the input image multiplied by 2 times.

In some implementations, in step (D), when it is determined that the minimum of the at least one midpoint distance is equal to the first set line distance, the corresponding two candidate line break objects One pair of data classified into the first group.

In other embodiments, in step (E), when it is determined that any one of the at least one midpoint distance is equal to the second set line distance, and two of the distances corresponding to the When the image frames passed by the candidate disconnected objects are different sides of the quadrilateral, the corresponding two candidate disconnected objects are classified as one pair of data in the second group.

The beneficial effect of the present invention is that: the input image of the printed circuit board is first pre-processed and contour lines are identified through the processing module, and then the midpoint distance of the midpoint coordinates is calculated based on Classifying the candidate disconnected objects to obtain the first group belonging to disconnected objects can implement a detection method that does not require the use of standard templates for comparison and does not require a large amount of data in an image database.

Description of the drawings

Figure 1 is a block diagram illustrating a detection system to which the detection method of the present invention is applicable;

Figure 2 is a flow chart illustrating an embodiment of the detection method of the present invention;

Figure 3 is a schematic diagram illustrating an aspect of an input image in this embodiment;

Figure 4 is a schematic diagram supplementing Figure 3 to illustrate a partial enlargement of the input image; and

FIG. 5 is a schematic diagram illustrating another aspect of the input image in this embodiment.

Detailed ways

The present invention will be described in detail below in conjunction with the accompanying drawings and examples:

Before the present invention is described in detail, it should be noted that similar elements are represented by the same numbers in the following description.

Referring to Figure 1, an embodiment of the detection method of the present invention is applicable to a detection system 100. The detection system 100 includes an image capture module 1, a processing module 3, and a storage module 2. The image capture module 1 is, for example, a camera or a photographic equipment, and is disposed on a printed circuit board inspection production line for capturing an image including the printed circuit board. The processing module 3 is, for example, at least one central processor of a computer system for implementing known or existing image recognition technology, and is electrically connected to the image capture module 1 to control the image capture module 1 to capture the image. image, and is also electrically connected to the storage module 2 to store the image and various data generated by the processing module 3 . The storage module 2 is, for example, a disk or other storage device.

Referring to Figures 1 and 2, the detection method includes steps S1 to S13.

In step S1, the processing module 3 controls the image capturing module 1 to capture and obtain the image including the printed circuit board. Next, step S2 is executed.

In step S2, the image is used as an input image through the processing module 3, and the input image is binarized. For example, the input image is a color image, and the binarization process is to convert the color image into a black and white image. The corresponding value of each pixel (Pixel) in the black and white image is 0 or 1. In this embodiment, the size of the input image is 240 pixels * 240 pixels, and has an image frame. The image frame is, for example, a square. Next, step S3 is executed. In other embodiments, the input image may also be of other sizes.

In addition, it should be noted that in other embodiments, the image frame can also be other quadrilaterals or other shapes. In addition, in step S2, the processing module 3 may also divide the image into multiple input images. Furthermore, the input images can also be stored in the storage module 2 in advance, and steps S1 and S2 are omitted.

In step S3, the processing module 3 determines multiple contour lines belonging to the black and white boundary or the frame of the image from the binarized input image, and defines them as corresponding line segment objects. Referring again to FIG. 3 , FIG. 3 illustrates an aspect of the input image, in which the outlines of the six line segment objects 41 to 46 correspond to the six outlines. Next, step S4 is executed.

In step S4, the processing module 3 determines whether a perimeter of each line segment object is greater than or equal to a perimeter threshold. When it is determined that the perimeter is smaller than the perimeter threshold, step S5 is executed. When it is determined that the perimeter is greater than or equal to the perimeter threshold, step S6 is executed. In this embodiment, the perimeter threshold is, for example, 10% of the perimeter, that is, 240*4*10%=96. In other embodiments, step S4 may also be to determine whether the perimeter is greater than the perimeter threshold.

In step S5, it indicates that the line segment object is too small and is not judged to be a wire segment of the printed circuit board, but may be bubbles, dirt, or stray spots, etc. Referring again to FIG. 3 , for example, the line segment object 43 .

In step S6, the processing module 3 determines whether the number of times each line segment object passes through the image frame is equal to 1 for filtering, and then the line segment objects corresponding to the number of times equal to 1 are used as multiple candidate line break objects. When it is determined that it is not equal to 1, step S7 is executed. When the judgment is equal to 1, step S8 is executed. Referring again to FIG. 3 , multiple line segment objects 41 , 42 , 45 , and 46 serve as multiple candidate line break objects.

In step S7, it indicates that the line segment object is too large or belongs to an independent object. Referring again to Figure 3, the degree corresponding to the line segment object 44 is equal to 0, and it is an independent object.

In step S8, referring again to FIG. 4, each candidate line break object has a design line width corresponding to the design value, and the design value is the theoretical width at design time. The processing module 3 calculates multiple actual line widths of each candidate disconnected object, that is, multiple actual line widths measured at different locations. The three distances 463 to 465 in Figure 4 are the three of them. Actual line width. When the processing module 3 determines that one of the actual line widths is equal to the design line width multiplied by a proportional threshold, the position of the center point of the two endpoint coordinates corresponding to the actual line width on the contour line is equal to the broken line end. The midpoint coordinate, the scale threshold is greater than zero and less than 1. For example, the proportion threshold is equal to 85%, and the actual line width (equal to the distance 465) is equal to 85% of the design line width, then the two endpoint coordinates 461 and 462 (or 451 and 452, or 421 and 422, The location of the center point of (or 411 and 412) is equal to the midpoint coordinate of the broken end of the line segment object 46 (or 45, or 42, or 41). The processing module 3 also calculates a midpoint distance between the midpoint coordinates of the disconnected end of each candidate disconnected object and the midpoint coordinates of the disconnected end of any other candidate disconnected object. Next, step S9 is executed.

In step S9, the processing module 3 determines whether each of the candidate disconnected objects is a first group. When the processing module 3 determines for each candidate disconnected object that the smallest one of the at least one midpoint distance from any other candidate disconnected object is less than a first set line distance, the corresponding two candidates The disconnected object is classified as a pair of data belonging to the first group of disconnected objects, and then step S10 is performed. On the contrary, the candidate disconnected objects that are not classified into the first group are executed in step S11. In this embodiment, the first set line distance is equal to a minimum line distance of the printed circuit board manufacturing process corresponding to the input image. In other embodiments, it can also be, for example, the line segment object in the input image. The minimum value of the design line distance, but not limited to this.

In step S10, the processing module 3 outputs data belonging to the first group. The data is, for example, the midpoint coordinates of the disconnected ends of one or more pairs of two candidate disconnected objects, or/and the endpoint coordinates corresponding to the midpoint coordinates.

In step S11, the processing module 3 determines whether the candidate disconnected objects that are not classified as the first group are a second group. The second group is a kind of classification data indicating that there may be a disconnection. . When the processing module 3 determines for each candidate wire-break object that any one of the at least one midpoint distance is less than a second set line distance, and the two candidate wire-break objects corresponding to the midpoint distance are When the image frames passing through respectively are different sides of the quadrilateral, the corresponding two candidate disconnected objects are classified as one pair of data of the second group, and then step S12 is performed. On the contrary, candidate disconnected objects that are not classified into the second group are executed in step S13. In this embodiment, the second set line spacing is equal to twice the minimum line spacing of the printed circuit board manufacturing process corresponding to the input image, but is not limited to this.

In step S12, the processing module 3 outputs data belonging to the second group. The data is, for example, the midpoint coordinates of the disconnected ends of one or more pairs of two candidate disconnected objects, or/and the endpoint coordinates corresponding to the midpoint coordinates.

In step S13, the processing module 3 determines that the candidate disconnected object that is not classified into the second group is a third group, and outputs data belonging to the third group. The data is, for example, the at least one midpoint coordinate of the at least one disconnected end of the at least one candidate disconnected object, or/and/or the endpoint coordinate corresponding to the at least one midpoint coordinate.

Referring again to FIG. 5 , FIG. 5 exemplarily illustrates an aspect of the input image, in which six midpoint coordinates 51 to 56 are defined to respectively correspond to six candidate disconnected objects A to F, and respectively correspond to the plurality of broken line objects A to F. The midpoint distances are respectively A to B (equal to B to A, and omitted below) = 2.0, A to C = 3.7, A to D = 4.1, A to E = 1.8, A to F = 0.6, B to C = 1.7 , B to D=3.0, B to E=1.1, B to F=2.3, C to D=2.2, C to E=2.0, C to F=3.8, D to E=2.3, D to F=3.8, E To F=1.8, the unit of the midpoint distance is, for example, the minimum line spacing of the printed circuit board manufacturing process. Then the candidate disconnection objects A and F are classified into the first group in pairs, the candidate disconnection objects B and E are classified into the second group in pairs, and the candidate disconnection objects C and D are classified into the second group. The third group.

In addition, it should be noted that in steps S9 and S11, when the processing module 3 determines for each candidate disconnected object that the minimum of the at least one midpoint distance from any other candidate disconnected object is equal to When the line distance is first set, the corresponding two candidate line break objects can be classified into the first group (or the second group) according to preset rules. Similarly, in steps S11 and S13, when the processing module 3 determines for each candidate disconnected object that the minimum of the at least one midpoint distance from any other candidate disconnected object is equal to the second setting When the line distance is determined, the corresponding two candidate broken objects can be classified into the second group (or the third group) according to preset rules.

In summary, the input image of the printed circuit board is pre-processed and contour lines are identified through the processing module 3, and then the candidate broken object is determined based on the calculated midpoint distance of the midpoint coordinates. Classification is performed to obtain the first group belonging to disconnections, the second group belonging to possible disconnections, and the third group belonging to the remaining parts, thereby achieving a comparison that does not require the use of standard templates. , and the detection method does not require a large amount of data in the image database, so the purpose of the present invention can indeed be achieved.

The above descriptions are only preferred embodiments of the present invention, but they are not intended to limit the scope of the present invention. Anyone familiar with the art can make further improvements based on this without departing from the spirit and scope of the present invention. Improvements and changes, therefore the protection scope of the present invention shall be subject to the scope defined by the claims of this application.

Claims (8)

1. A method of detection suitable for use in and implemented by a detection system, the method comprising:

a: performing binarization processing on an input image related to the printed circuit board, wherein the input image is provided with an image frame;

b: for the binarized input image, judging a plurality of contour lines belonging to a black-white boundary or the image frame, and defining the contour lines as a plurality of corresponding line segment objects;

c: screening each line segment object according to a perimeter threshold value and the times of passing through the image frame to obtain a plurality of candidate line segment objects; and

D: judging whether each candidate broken line object is classified as belonging to a first group of broken lines according to the midpoint coordinates of the broken line ends of each candidate broken line object and a first set line distance, judging whether each candidate broken line object has a designed line width corresponding to a designed value and a plurality of actual line widths measured at different positions, when one of the actual line widths is equal to the designed line width multiplied by a proportion threshold value, the position of a central point of two end coordinates corresponding to the actual line width on the contour line is equal to the midpoint coordinates of the broken line ends, the proportion threshold value is larger than zero and smaller than 1, calculating the midpoint distances between the midpoint coordinates of the broken line ends of each candidate broken line object and the midpoint coordinates of the broken line ends of any one of the candidate broken line objects, and when the smallest one of the at least one midpoint distances is judged to be smaller than the first set line distance, the corresponding two candidate broken line objects are classified as one pair of the first set line distances, and the corresponding data of the first set line distances of the first set image plates are equal to the minimum image distance of the first set image plates.

2. The detecting method according to claim 1, wherein in step C, when any one of the line segment objects has a perimeter greater than or equal to the perimeter threshold and the number of times the line segment object passes through the image frame is equal to 1, the line segment object is determined as one of the candidate line segment objects.

3. The detecting method according to claim 2, further comprising a step E of determining whether the candidate broken line object not classified as the first group is classified as belonging to a second group based on the at least one midpoint distance of the midpoint coordinates of the broken line end of the candidate broken line object not classified as the first group and a second set line distance, the second set line distance being larger than the first set line distance.

4. The detecting method as claimed in claim 3, wherein in the step A, the image frame is a quadrangle, and in the step E, when it is determined that any one of the at least one midpoint distance is smaller than the second set line distance and the image frames through which the two candidate broken line objects corresponding to the midpoint distance respectively pass are different sides of the quadrangle, the corresponding two candidate broken line objects are classified into one pair of data of the second group.

5. The method of claim 4, further comprising step F of determining that at least one candidate broken line object that is not classified as either the first group or the second group is classified as belonging to a third group.

6. The inspection method of claim 5, wherein in step D, the ratio threshold is equal to 85%, and the second set line spacing is equal to 2 times the minimum line spacing of the printed circuit board process corresponding to the input image.

7. The method of claim 6, wherein in step D, when it is determined that the smallest one of the at least one midpoint distance is equal to the first set line distance, the corresponding two candidate broken line objects are classified as one pair of data of the first group.

8. The detecting method as claimed in claim 6, wherein in the step E, when it is determined that any one of the at least one midpoint distance is equal to the second set line distance and the image frames through which the two candidate broken line objects corresponding to the midpoint distance respectively pass are different sides of the quadrangle, the corresponding two candidate broken line objects are classified as one pair of data of the second group.