RU2522870C1 - Method for range verification of printed-circuit boards - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: disclosed is a method for range verification of printed-circuit boards from binary images thereof, which, unlike the existing method, includes counting the number of compressed and defocused sections of a source image matrix; detecting sinks in conductor strips, assigning said sinks a conductor attribute; correcting the source image matrix; counting the number of compressed and defocused sections of the corrected matrix, as well as the total number of defects, the number of joints to the compressed and defocused regions, the number of associated defects of the corrected matrix corresponding to the number of metal coating sections; determining the number of unassociated defects as the difference between the total number of defects and sections adjoining to the compressed and defocused regions; generating a reference number as the difference between the number of associated sections in the corrected matrix and the number of unassociated defects; comparing the number of compressed and defocused sections of the source image matrix with the reference number; equality of said numbers indicates absence of unallowable constrictions in conductors; comparing the number of compressed and defocused sections of the corrected matrix with the reference number, wherein equality of said numbers indicates that sinks are either absent or are located on the periphery of conductor lines and do not lead to unallowable constrictions.

EFFECT: high efficiency of inspection by avoiding the procedure of determining the connection of each defective section and broader functional capabilities owing to determination of the number of sinks on conductor strips.

Description

The invention relates to automation and computer technology and can be used in systems with technical vision in the control of printed circuit boards.

Known methods of control of printed circuit boards based on comparison with standards (Automatic PCI Inspection Algoritms. / M. Moganti, F. Ercal, C. Dagli [etc.] // A Survey, Computer Vision and Image Understanding. - 1996. - No. 63 .- P.287-313; Automatic Visual Inspection of Printed Circuit Boards: An Experimental System. / B. Benhabib, CRCharette, KCSmith, AMYip. // International Journal of Robotic and Automation. - 1990. - No. 5 ( 2).

Despite the apparent simplicity, the methods are sensitive to the quality of alignment of the image and the standard, in particular to mutual rotation, parallel displacement, and to insignificant zooming and do not allow to differentiate the actual and false narrowings.

Known methods for monitoring printed circuit boards based on the use of structural-linguistic methods of image recognition (A. L. Drzhevetsky, V. N. Kontishev “Methods and means of monitoring the topological characteristics of many objects in systems with technical vision.” - Devices and control systems. 1993 . No. 3), allowing to control the acceptable values of the width of the conductors and the permissible distances between the conductors, regardless of their orientation. The disadvantages of these methods are: reduced accuracy in determining the narrowing of conductors in the presence of shells located on the periphery of the printed conductors, and the inability to classify the causes of narrowing.

Closest to the proposed one is a method of optical tolerance control (A.L.Drzhevetsky, A.V. Grigoriev. "Automated system of optical tolerance control of printed circuit boards and photo masks." -Metrology, monthly supplement to the scientific and technical journal "Measuring equipment". 1995 No. 4), which makes it possible to control almost the entire wide class of defects in printed circuit boards with an indication of their location and quantitative characteristics, including unacceptable narrowing of printed conductors, for the detection of which using complex algorithms are introduced that reduce the control performance and do not exclude the selection of false defects caused by the presence of shells located on the periphery of the printed conductors, as well as the difficulty of unambiguously specifying the causes of narrowing, which can be caused by a natural decrease in the cross section of the conductors due to the influence of shells in the conductor body, which is an important characteristic of defects, simplifying the analysis of the technological process and the choice of direction of its correction.

The technical result of the proposed method of tolerance control of printed circuit boards is to increase the control performance by eliminating the procedure for determining the connectivity of each defective area and expanding functionality by determining the number of shells of printed conductors, leading to unacceptable narrowing.

The proposed method of tolerance control of printed circuit boards is based on the fact that sections of the conductors of printed circuit boards are made that have a cross section that exceeds the permissible value. In this case, the region outside the conductors is analyzed, in which there may exist regions of defects whose cross section does not exceed the allowable value, and the region of defects such as sagging or chips that closely border the main body of the conductor, distorting its boundaries. In addition, in the body of the conductor there are shells that are mainly concentrated in the peripheral part of the conductor, reduce its cross section and form doubly connected defects, which are necessary but not sufficient conditions for narrowing. The narrowing of the conductor in the absence of shells also forms a doubly connected defect, and the presence of such a defect is a necessary and sufficient sign of narrowing.

For the correct diagnosis of narrowing, it is important to determine the nature of the formation of doubly connected defects, namely, to find out: they are caused by physical narrowing of the conductor or by shells inside the conductor. This is important for finer selection of defects. The problem is solved by forming one singular, which is a reference and reflects the absence of narrowing in the conductors of printed circuit boards in the field of view, and a comparison of the real number for a specific printed circuit board with a reference gives information about narrowing and the nature of these narrowing. The reference number is formed in several stages.

At the first stage, sinks are excluded in the images of conductors and the number of metallized areas, the total number of defective areas, which is equal to the sum of single and adjacent defects, are calculated. The reference number, which characterizes the conductors without narrowing, is the difference between the total number of metallized sections and the number of single defects. Count the number of images whose cross section exceeds the permissible value, and compare with the reference number. If the number of images, the cross-section of which exceeds the permissible value, exceeds the reference number, then we conclude that there are narrowed conductors in the field of view, and such a board is rejected. If the number of images whose cross section exceeds the permissible value is equal to the reference number, then we conclude that in the field of view biconnected defects are either absent or located on the periphery of the conductors and do not lead to narrowing.

Improving the accuracy and performance of the tolerance control of printed circuit boards during selection of defects is achieved due to the fact that the narrowing of conductors is evaluated without a specific consideration of the parameters of each defect, and is performed integrally for the entire image field, which includes the conductors themselves, defects outside the conductors and the shell inside the conductor body. This approach does not require the use of complex and detailed image processing, which naturally increases the productivity of the method and allows, at the same time, differentially distinguish narrowing caused by a physical change in the width of the conductors, which are less than the permissible value, and immediately draw a conclusion about marriage without further lengthy research. and also to eliminate false narrowing, which are manifested in the form of doubly connected defects, but are located on the peripheral sections of the conductor, which increases the accuracy of the tolerance control without additional time and the use of complex computational procedures. The proposed method makes it quite simple to form a reference number, a comparison with which gives an answer to the question about the nature of narrowing, and to eliminate false narrowing. The method can be used both independently and as part of others, improving their characteristics.

The proposed method of tolerance control of printed circuit boards is focused on the use of television-type readers that convert the image of a section of a printed circuit board into a video signal, which, after quantizing it into two levels, forms samples of a binary image of a section of a printed circuit board corresponding to the matrix of the original image in which the level “1” corresponds to the sections metallization, and "0" - the dielectric. As a result, an additional matrix is formed, obtaining the original image from the matrix by the operations of compression and defocusing.

The principles of performing compression and defocus operations and the implementation of digital filters are described in the article (A. L. Drzhevetsky “Methods for constructing non-linear spatial filters with controlled characteristics.” - Penza, 1991). The first number, which will later be compared with the reference, is obtained by counting the number of compressed and defocused connected portions of the matrix of the original image. The principle of counting the number of images and devices that are closest to the method described is given in (AS 1383413 USSR, G06K 9/00. A device for counting the number of images of objects in AS 1640720 USSR, G06K 9/00. Device to count images of objects). By logically subtracting the readings of the additional matrix from the matrix of the original image, a defect matrix is obtained, based on which the connected image areas corresponding to the defects and the defective areas adjacent to the compressed and defocused areas are extracted. Simultaneously with reading the matrix of the initial image, shells are isolated in the conductors and marked, that is, each shell is assigned a serial number and a connection is established with the marked conductor. For each shell, the area, perimeter and coordinates of the center of gravity are calculated. A method and apparatus for determining the coordinates of the centers of gravity of images of many objects that are maximally adapted for use in this invention are discussed in (Patent of the Russian Federation RU No. 2032218 C1, CL G06K 9/00 dated 03/27/95 Bull. No. 9).

The next important operation that must be performed before the reference number is formed is the correction of the matrix of the original image, the purpose of which is to assign the conductor sign to the selected shells (that is, level “1”). Since all sinks are marked and there is a library of sinks, only those that are excluded from the object are selected from all openings. This means that sinks with the corresponding numbers to be excluded are assigned a level of “1” corresponding to the level of the conductor. Technically, this operation is implemented during the reading of the second frame. For this, a random access memory (RAM) is introduced, the address space of which corresponds to the numbers of the sinks, and write “1” according to the corresponding numbers. Writing to RAM is done before reading the second frame, and after writing, the RAM address space is connected to the shell marking nodes that determine the shell numbers. During the reading of the second frame, elements with a level of “1” from the RAM output are mixed into the matrix of the original image. When reading the second frame, the matrix of the original image is replaced with the adjusted one and all operations are performed with the adjusted matrix. During reading of the second frame, all shells are excluded from the matrix of the original image.

As a result, the number of compressed and defocused areas of the adjusted matrix is calculated, as well as the total number of defects, the number of adjacencies to the compressed and defocused areas in the adjusted matrix, as well as the number of connected sections of the adjusted matrix, which correspond to the number of metallization sections. Based on these numbers, a reference number is formed, as the difference between the number of connected sections in the adjusted matrix and the number of disconnected defects, which, in turn, is determined as the difference between the total number of defects and the number of adjoining to the compressed and defocused areas. Based on a comparison of the reference number with the number of compressed and defocused sections of the adjusted matrix or with the number of compressed and defocused sections of the matrix of the original image, conclusions are made about the presence or absence of narrowing caused by the natural narrowing of conductors or shells in the body of the conductors.

Natural narrowing occurs if the number of squeezed and defocused sections of the adjusted matrix is greater than the reference number, which indicates unconditional marriage, and if equal, the narrowing is absent and the presence of narrowing due to shells is checked.

The equality of the reference number and the number of compressed and defocused portions of the matrix of the original image indicates either the absence of shells, or the fact that, if any, these shells are located on the periphery of the conducting tracks and do not lead to unacceptable narrowing.

This method can be expanded for other cases, for example, to exclude shells with predetermined properties, with a small area, and in these conditions compare the number of compressed and defocused sections of the matrix with a reference number. There are other options that do not affect the essence of the method.

Claims (1)

The method of tolerance control of printed circuit boards according to their binary images, which consists in the fact that from the matrix of the original image, by isotropic compression and defocusing, an additional matrix is formed in which the areas of the connected image elements correspond to image fragments whose size exceeds the minimum allowable value specified by the isotropic coefficients compression and defocusing, by logically subtracting the samples of the additional matrix from the matrix of the original image, a matrix of defects is obtained in which the regions of connected elements corresponding to defects and the defective regions adjacent to the regions of connected elements of the additional matrix are distinguished, characterized in that the number of compressed and defocused regions of the matrix of the original image is counted, shells are detected in the printed conductors, the conductor attribute is assigned to these shells, and corrected the matrix of the original image, count the number of compressed and defocused areas of the adjusted matrix, as well as the total number of defects, the number of adjacencies to compressed and defocused areas, the number of related defects in the corrected matrix corresponding to the number of metallization sections, determine the number of unbound defects, as the difference between the total number of defects and areas adjacent to the compressed and defocused areas, form a reference number, as the difference between the number of connected sections in the adjusted matrix and the number of unrelated defects, compare the number of compressed and defocused sections of the matrix of the original image with the reference number, with e of these numbers conclude that there are no unacceptable narrowings in the conductors, compare the number of compressed and defocused sections of the corrected matrix with a reference number, if they are equal, they conclude that the shells are either absent or located on the periphery of the conducting tracks and do not lead to unacceptable narrowing .