CN117115142B - Electronic element defect rapid detection method based on artificial intelligence - Google Patents
Electronic element defect rapid detection method based on artificial intelligence Download PDFInfo
- Publication number
- CN117115142B CN117115142B CN202311329968.4A CN202311329968A CN117115142B CN 117115142 B CN117115142 B CN 117115142B CN 202311329968 A CN202311329968 A CN 202311329968A CN 117115142 B CN117115142 B CN 117115142B
- Authority
- CN
- China
- Prior art keywords
- window
- signal
- windows
- image
- frequency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000007547 defect Effects 0.000 title claims abstract description 56
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 20
- 238000001514 detection method Methods 0.000 title claims abstract description 19
- 238000001228 spectrum Methods 0.000 claims abstract description 115
- 230000008859 change Effects 0.000 claims abstract description 53
- 230000002159 abnormal effect Effects 0.000 claims abstract description 52
- 238000000034 method Methods 0.000 claims abstract description 39
- 230000000903 blocking effect Effects 0.000 claims abstract description 9
- 230000008569 process Effects 0.000 claims abstract description 9
- 238000004458 analytical method Methods 0.000 claims abstract description 6
- ATJFFYVFTNAWJD-UHFFFAOYSA-N Tin Chemical compound [Sn] ATJFFYVFTNAWJD-UHFFFAOYSA-N 0.000 claims description 32
- 238000005070 sampling Methods 0.000 claims description 13
- 239000012634 fragment Substances 0.000 claims description 12
- 238000012163 sequencing technique Methods 0.000 claims description 10
- 238000010586 diagram Methods 0.000 claims description 9
- 230000005856 abnormality Effects 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 3
- 238000005314 correlation function Methods 0.000 abstract description 10
- 230000000875 corresponding effect Effects 0.000 description 20
- 230000011218 segmentation Effects 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000011179 visual inspection Methods 0.000 description 2
- 238000003466 welding Methods 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20056—Discrete and fast Fourier transform, [DFT, FFT]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image processing, in particular to an electronic element defect rapid detection method based on artificial intelligence, which comprises the following steps: graying the acquired electronic element image to obtain an electronic element gray image; performing tile blocking treatment on the window, and performing Fourier transform operation on the split window to obtain a spectrum image of the window; analyzing according to the spectrum signal change in the window to obtain the signal change degree in the window, and then setting a cross-correlation information function to obtain an abnormal spectrum template window; carrying out differential analysis on different windows according to the abnormal spectrum template window to obtain an abnormal spectrum image region; and carrying out positioning identification according to the abnormal frequency spectrum image area to obtain the defect position of the electronic element. According to the invention, the defect position area is detected and processed by adopting the fixed window according to the signal cross correlation function property in the frequency spectrum image by combining the regional characteristics of the electronic element image, and the defect position area is identified, so that the processing process of the electronic element is more intelligent.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an electronic element defect rapid detection method based on artificial intelligence.
Background
Electronic component defect detection is a key step in the production process of semiconductor factories to stabilize processing quality. In the processing process of electronic components in a digital factory, the surfaces of the electronic components are affected by welding spot defects caused by operations such as automatic packaging of articles, and the like, and the defects such as bad defects caused by scratches of welding spots of the electronic components are usually easily generated, and the scratches can seriously affect the reliability of electronic products and the quality of a subsequent whole hardware system. Therefore, electronic component defect detection is important in a digital factory.
The traditional electronic component defect detection is a detection method for evaluating the appearance and characteristics of components by manual visual inspection, particularly a detection method for observing and judging by human eyes, wherein the manual visual inspection needs to take a long time to inspect each component, and moreover, the visual ability and subjective judgment of people can be influenced by factors such as individual difference and fatigue, so that the detection result is inaccurate.
Disclosure of Invention
The invention provides an electronic element defect rapid detection method based on artificial intelligence, which aims to solve the existing problems.
The invention discloses an electronic element defect rapid detection method based on artificial intelligence, which adopts the following technical scheme:
an embodiment of the invention provides an electronic component defect rapid detection method based on artificial intelligence, which comprises the following steps:
acquiring a gray image of an electronic element;
performing tile blocking processing on the gray level image to obtain all windows, and performing Fourier transform operation on all windows to obtain spectrum images in all windows;
constructing a frequency power diagram of signal frequency and signal amplitude according to the frequency spectrum image in any window; presetting a plurality of signal amplitude intervals; acquiring all signal frequency intervals in each signal amplitude interval on a frequency power diagram; obtaining the waveform change of the window according to the number of the signal frequency intervals; sampling and sequencing the signal amplitudes of all the signal frequency intervals in the window according to the waveform change to obtain a waveform change sequence of the window; establishing a cross-correlation information function between windows according to waveform change sequences of any two windows; acquiring all abnormal spectrum template windows according to the cross-correlation information function;
and carrying out positioning identification according to all abnormal spectrum template windows to obtain the defect position of the electronic element.
Preferably, the step of acquiring the gray-scale image of the electronic component includes the following specific steps:
acquiring the electronic element graph by using an industrial camera to obtain an original electronic element graph; and carrying out graying treatment on the original electronic element image to obtain a gray image of the electronic element.
Preferably, the tile blocking processing is performed on the gray image to obtain all windows, which comprises the following specific steps:
performing tile blocking processing on the gray level image according to the position and the size of the tin spot to obtain all windows with equal size, and marking the windows as a window setOnly the area with tin spots is segmented, the size of the segmented area is consistent with the size of the tin spots of the gray level image,for window set->All windows in (1) are numbered, the obtained window set +.>Expressed as:
in the method, in the process of the invention,representing a set of all window compositions; />Representing the total number of windows in the window set; />Representing Window set +.>The%>A window.
Preferably, the constructing a frequency power diagram of signal frequency and signal amplitude according to the spectrum image in any window includes the following specific steps:
performing Fourier transform operation on any window to obtain a frequency spectrum image of the window, and performing sampling analysis on the frequency spectrum image of the window to obtain the frequency signal change degree of the frequency spectrum image in the window when performing Fourier transform; and establishing a frequency power diagram of signal frequency and signal amplitude according to the frequency signal change degree of the frequency spectrum image, wherein the abscissa is the signal frequency, and the ordinate is the signal amplitude.
Preferably, the step of obtaining all signal frequency intervals in each signal amplitude interval on the frequency power map includes the following specific steps:
the preset four signal amplitude intervals are respectively:,/>,/>,/>；
recording the signal frequency between any two adjacent wave troughs as a signal segment;
acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>；
Acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>；
Acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>；
Acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>。
Preferably, the step of obtaining the waveform change of the window according to the number of the signal frequency intervals includes the following specific steps:
will beMarked as->The method comprises the steps of carrying out a first treatment on the surface of the Will->Marked as->The method comprises the steps of carrying out a first treatment on the surface of the Will beMarked as->；
Further acquiring the waveform change of the signal in any window,/>And->。
Preferably, the sampling and sorting process is performed on the signal amplitudes of all the signal frequency intervals in the window according to the waveform change to obtain a waveform change sequence of the window, and the method comprises the following specific steps:
taking the signal amplitude 0 as a reference point, assigning 1 to the signal amplitude higher than the reference point, assigning-1 to the signal amplitude lower than the reference point, and assigning 0 to the signal amplitude equal to the reference point; amplitude sampling sequencing is carried out on all signal frequency intervals of any window from large to small; i.e. to waveform variations,/>And->Corresponding signal frequency interval +.>，/>，/>And->And performing amplitude sampling sequencing to obtain a waveform change sequence of the signal in any window.
Preferably, the establishing a cross-correlation information function between windows according to the waveform change sequences of the two windows includes the following specific steps:
the waveform change sequences for any two windows are respectively recorded as；/>The elements in (a) are marked as->，/>The elements in (a) are marked as->The method comprises the steps of carrying out a first treatment on the surface of the I.e. < ->、/>The method comprises the steps of carrying out a first treatment on the surface of the Window->The cross-correlation information function between is:
in the method, in the process of the invention,representing the cross-correlation degree of any two windows; />Element of (a)>Representation->Elements of (a) and (b); />Representation->Probability of simultaneous occurrence in signal waveform sequences within two windows simultaneously; />Representation element->At->Probability of occurrence of->Representation element->At->Is a probability of occurrence in the past.
Preferably, the step of obtaining all abnormal spectrum template windows according to the cross-correlation information function includes the following specific steps:
(1) All windows are combined pairwise to obtain all window pairs, and calculation is performedThe cross-correlation degree of each window pair, and the cross-correlation degree of all window pairs obtains sequences according to the sorting records from big to small；
(2) Respectively obtain the firstPersonal window->The degree of cross-correlation with all other windows is denoted as set +.>；
(3) If it isIn the sequence->15 percent later, then window->Is an abnormal spectrum template window; if->Not in sequence->15 percent later, then window->Not an abnormal spectrum template window;
all abnormal spectrum template windows are obtained.
Preferably, the positioning and identifying are performed according to all abnormal spectrum template windows to obtain the defect positions of the electronic elements, including the following specific steps:
preset parametersFor all abnormal frequency spectrumsTemplate Window set +.>Performing inverse Fourier transform to obtain window setThe position information in the gray level image of the electronic element is marked as a defect area of the electronic element, so that the defect area of the electronic element is intelligently identified; and then the number ratio of all abnormal spectrum template windows in all windows is marked as abnormal probability +.>If the abnormality probability->And judging the electronic element as a bad product.
The technical scheme of the invention has the beneficial effects that: compared with the conventional electrical detection method in the prior art, the method can conveniently carry out digital signal processing by adopting the signal waveform sequence of the electronic element, and the waveform sequence of the signal in the image is extracted by an image spectrum processing technology for detecting the abnormality in the signal. By comparing the change conditions of the normal waveform and the abnormal waveform sequence, the problems of the abnormal waveform sequence and the like can be found, and the correlation degree between the characteristics of the spectrum signals and the target variable can be estimated by adopting the cross-correlation information. Through calculating the cross-correlation information of the spectrum characteristic signals and the target variable signals, characteristic signal sequences highly correlated with the spectrum signals can be screened out, and further, characteristic signal sequences with lower frequency spectrums are distinguished to identify abnormal spectrum signals; the identification speed of the abnormal defect position in the image is improved, so that fault diagnosis and repair are convenient.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of a method for rapidly detecting defects of electronic components based on artificial intelligence according to the present invention;
fig. 2 is a schematic diagram of frequency power established for a spectral image within any one window of the artificial intelligence-based electronic component defect fast detection method of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following description refers to the specific implementation, structure, characteristics and effects of the electronic element defect rapid detection method based on artificial intelligence according to the invention by combining the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the electronic component defect rapid detection method based on artificial intelligence provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of steps of an artificial intelligence-based electronic component defect rapid detection method according to an embodiment of the invention is shown, the method includes the following steps:
step S001: and acquiring an electronic element image, and carrying out graying treatment on the electronic element image to obtain an electronic element gray image.
Specifically, an industrial camera is utilized to collect the electronic element graph, and an original electronic element image is obtained; carrying out graying treatment on the original electronic element image to obtain a gray image of the electronic element; and then carrying out Fourier transform processing on the gray level image of the electronic element to obtain the whole spectrum image of the electronic element.
Thus, a gray-scale image of the electronic component and an entire spectrum image of the electronic component are acquired.
Step S002: and performing tile blocking processing on the gray level image according to the positions and the sizes of tin points of the gray level image to obtain windows with all fixed sizes, and performing Fourier transform operation on all windows to obtain spectrum images of all windows.
Since a strong periodic phenomenon occurs in the whole spectrum image of the electronic component, if defects in the image cannot be well identified simply by analyzing the whole spectrum image, the spectrum image of the electronic component can be subjected to block processing of the same size window, a cross-correlation function is set by utilizing the change of the spectrum signal in the window, and an abnormal window is identified according to the cross-correlation function, so that an abnormal defect area is intelligently identified.
It should be further noted that, according to the distribution condition of tin point positions of the gray level image of the electronic element, a window is obtained, and then fourier transformation operation is performed on the obtained window, so as to obtain a spectrum image of the window. The abnormal window is obtained by comparing the signal change rule of the frequency spectrum in the frequency spectrum image of the normal window with the signal rule in the frequency spectrum image of other windows, so that the abnormal defect position of the electronic element can be intuitively found.
Specifically, a plane coordinate system is established by taking the upper left point of the gray image as the origin, and the gray image is subjected to coordinate system processing to obtain the positions and the sizes of all tin points in the gray image; performing tile blocking processing on the gray level image according to the position and the size of the tin spot to obtain all windows with equal size, and marking the windows as a window set；
Pair window setAll windows in (1) are numbered, the obtained window set +.>Expressed as:
in the method, in the process of the invention,representing a set of all window compositions; />Representing the total number of windows in the window set; />Representing Window set +.>The%>A window.
It should be noted that when the tile segmentation is performed on the gray image, the segmentation is performed according to the tin points in the gray image, only the region where the tin points exist is segmented, and the size of the segmentation is consistent with the size of the tin points of the gray image.
So far, all windows after the gray image blocking processing are obtained.
Step S003: and obtaining the signal change degree in the window according to the spectrum signal change of the spectrum image in the analysis window, and then setting a cross-correlation information function to obtain an abnormal spectrum template window.
The method is characterized in that according to the spectrum signal change condition of the spectrum image in the window after the gray image is segmented, then the cross correlation degree analysis of all the signals among the windows can obtain windows in different states. Because the spectrum signals of the spectrum images in the windows show the signal power of each window, when the windows perform Fourier transformation, the changes of the frequency, amplitude and phase of the signals are included, and the changes also reflect different states and changing conditions of tin points.
When the tin points of the electronic element are in a normal state, the image spectrum images of the windows are high in correlation degree with the spectrum images of other normal tin points, and when the abnormal tin points occur in the electronic element, the spectrum images in the windows are greatly different from the spectrum images of most of the normal tin points and the correlation degree between the windows is low, so that the correlation degree between the windows can be obtained according to the difference between the spectrum images of the windows, the abnormal windows can be identified by utilizing the cross correlation degree between the windows, and a defect area is obtained.
Specifically, pair aggregationPerforming Fourier transform operation on all windows to obtain spectrum images of all windows; and (3) utilizing the correlation of the spectrum images of the windows to sample and analyze the spectrum images of all the windows to obtain the frequency signal change degree of the spectrum images in the windows when the Fourier transformation is carried out on all the windows.
The signal of the spectrum image in the window can be subjected to power frequency decomposition processing according to the frequency signal change degree of the spectrum image in the window to obtain the signal change of the power frequency image in the window, so that the change condition of the defect tin point in the signal can be easily found. In the gray level image of the electronic element, the fluctuation degree of the power frequency signal change of the frequency spectrum image in the window corresponding to the normal tin point is not large, but when the fluctuation degree of the power frequency signal of the frequency spectrum image in a certain window is large, the possibility that the window is a defect is high is shown, and the cross correlation function can be set according to the frequency signal change degree of the frequency spectrum image in all the windows to rapidly identify the target defect.
The implementation sets a cross-correlation function according to the spectrum signal variation degree of the spectrum image in the window, and further performs defect identification on the image. The principle is as follows: when any tin point in the gray level image is a defect, when the spectrum image in the corresponding window is compared with the spectrum images of adjacent surrounding windows in the fluctuation change degree of the signals, the fluctuation change degree of the signals is found to be large, so that the abnormal spectrum image and the normal spectrum image can be identified by utilizing the cross-correlation function.
It is further noted that because the cross-correlation function is established based on the waveform variation of the signal in the frequency power plot of the frequency image within the window. It is necessary to count waveform variations and waveform sequence variations of signals of the frequency power maps of the frequency images within all windows. A cross-correlation function is then established by the waveform variations and the waveform sequence variations.
1. The waveform change of the signal within the window is acquired.
Establishing a frequency power diagram for the spectrum image in any window, as shown in fig. 2; in fig. 2, the abscissa indicates signal frequency, and the ordinate indicates signal amplitude;
in fig. 2, it can be seen that the direction of the signal waveform, the degree of difference between the wave front and the wave trough, and the degree of variation between the large wave crest and the small wave crest, are often subject to certain statistical characteristics. And obtaining the waveform change condition of the signal in the window according to the waveform trend of the signal in the power frequency image of the frequency spectrum image in the window.
Presetting four interval parametersAnd->Wherein the present embodiment usesAnd->Examples are described, the present embodiment is not particularly limited, wherein +.>And->Depending on the particular implementation.
Specifically, the signal waveform in any window is segmented to obtain four signal amplitude valuesDegree intervalThe intervals are respectively: />, /> ,/>,。
The signal frequency between any two adjacent valleys is denoted as a signal segment and from 0 to the first valley is also denoted as a signal segment.
Acquiring all signal frequency intervalsThese signal frequency intervals +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In all of these frequency intervals +.>The number of the internal signal fragments is marked as +.>；
Acquiring all signal frequency intervalsThese signal frequency intervals +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In all of these frequency intervals +.>The number of the internal signal fragments is marked as +.>Will->Is marked as；
Acquiring all signal frequency intervalsThese signal frequency intervals +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In all of these frequency intervals +.>The number of the internal signal fragments is marked as +.>Will beMarked as->；
Acquiring all signal frequency intervalsThese signal frequency intervals +.>Each of (3)The signal amplitudes corresponding to the signal frequencies belong to the interval +.>In all of these frequency intervals +.>The number of the internal signal fragments is marked as +.>Will beMarked as->；
Thus, the waveform change of the signal in any window is obtained,/>And->。
2. A sequence of waveform changes of the signal within the window is acquired.
Specifically, according to the waveform change condition of the frequency power of the spectrum image in the window, the waveform change sequence of the signal in the window is subjected to amplitude sampling sequencing, and the specific operation is as follows: with the signal amplitude 0 as the reference point, the signal amplitude higher than the reference point is given 1, the signal amplitude lower than the reference point is given-1, and the signal amplitude equal to the reference point is given 0.
Amplitude sampling sequencing is carried out on all signal frequency intervals of any window from large to small; i.e. for all signal frequency intervals，/>，/>And->Performing amplitude sampling sequencing to obtain a waveform change sequence of signals in any window of +.>。
So far, the waveform change sequence of any signal in a window is obtained。
3. And obtaining a cross-correlation information function according to the waveform change of the signal in the window and the waveform change sequence of the signal in the window.
It should be noted that, in the gray level image of the electronic element, the fluctuation degree of the power frequency signal change of the spectrum image in the window corresponding to the normal tin point is not large, but when the power frequency signal of the spectrum image in a certain window has a large fluctuation degree, the possibility that the window is a defect is large, and the cross correlation function can be set according to the frequency signal change degree of the spectrum image in all the windows to rapidly identify the target defect.
It should be further noted that, in the electronic component, the frequency power maps of the spectrum images in the windows corresponding to all the normal tin points are often correlated to a high degree, the signal waveform changes of the frequency power maps of the spectrum images in the windows corresponding to the abnormal tin points are similar, in the comparison of the signal waveform changes of the frequency power maps of the spectrum images in the windows corresponding to the normal tin points with the spectrum images in the windows corresponding to the normal tin points, the signal waveform changes are obvious, the frequency power images between the two frequency power maps are different, and the waveform sequences between the two frequency power maps are also different. The cross-correlation function may be set from the signal waveform sequence of the power frequencies of the spectral images within all windows. (mutual information is a measure of the correlation between two event sets)
Specifically, for any two windowsThe waveform change sequences of (2) are respectively recorded as；/>The elements in (a) are marked as->，/>The elements in (a) are marked as->The method comprises the steps of carrying out a first treatment on the surface of the I.e. < ->、/>The method comprises the steps of carrying out a first treatment on the surface of the Window->The cross-correlation information function between is:
in the method, in the process of the invention,representing the cross-correlation degree of any two windows; />Element of (a)>Representation->Elements of (a) and (b); />Representation->Probability of simultaneous occurrence in signal waveform sequences within two windows simultaneously; />Representation element->At->Probability of occurrence of->Representation element->At->Is a probability of occurrence in the past.
Based on cross-correlation informationAnalyzing the cross-correlation degree of signal waveform sequence changes in the two windows, and when the cross-correlation degree of the two windows is larger, indicating that the frequency spectrum correlation of the two windows is stronger; the weaker the degree of cross-correlation between the two, the weaker the correlation between the two windows.
Thus, a cross-correlation information function between windows is obtained.
4. And acquiring an abnormal spectrum template window.
When comparing the spectrum signals of the divided windows according to the cross-correlation information function, the frequency power patterns of the spectrum images in the windows corresponding to all normal tin points in the electronic element are always cross-correlated to a high degree, the signal waveform changes of the frequency power patterns of the spectrum images in the windows corresponding to the abnormal tin points are similar, in the comparison of the signal waveform changes of the frequency power patterns of the spectrum images in the windows corresponding to the normal tin points with the frequency power patterns in the windows corresponding to the normal tin points, the signal waveform changes are obvious, the frequency power images between the two frequency power patterns are different, and the waveform sequences between the two frequency power patterns are different.
And establishing a cross-correlation information function according to the spectrum signals of the normal tin points in the window and the spectrum signals of the normal tin points, wherein the correlation degree is higher, the defects are random, the spectrum signals presented by the defects in the window are also randomly distributed, and the cross-correlation information function is established by the spectrum signals of the normal window and the spectrum signals of the defect window, so that the smaller the correlation degree is, the weaker the correlation between the two windows is, and the probability of the defects is higher.
Specifically, the step of acquiring the abnormal spectrum template window is as follows:
(1) All windows are combined pairwise to obtain all window pairs, the cross-correlation degree of each window pair is calculated, and the cross-correlation degree of all window pairs is recorded according to the sequence from big to small to obtain the sequence。
(2) Respectively obtain the firstPersonal window->The degree of cross-correlation with all other windows is denoted as set +.>。
(3) If it isIn the sequence->15 percent later, the description window->Is an abnormal spectrum template window; if->Not in sequence->15 percent later, the subsequent window +.>Template matching is carried out on all windows to find abnormal frequency spectrum windows, and +.>Whether it can be an abnormal spectrum template window.
(4) If the subsequent windowThe degree of cross-correlation with all windows is in sequence +.>After 15 percent, find abnormal spectrum template window +.>。
So far, can judge the firstPersonal window->If the obtained value is the abnormal spectrum template window, all abnormal spectrum template windows are obtained by the same way and are marked as a set +.>。
Step S004: and carrying out positioning identification according to the abnormal spectrum template window to obtain the defect position of the electronic element.
Presetting a threshold parameterWherein the present embodiment is +.>Examples are described, the present embodiment is not particularly limited, wherein +.>Depending on the particular implementation.
Obtaining all abnormal spectrum template windows through the abnormal analysis of the spectrum image in the steps, and collecting all abnormal spectrum template windowsPerforming inverse Fourier transform to obtain window set->The position information in the gray level image of the electronic element is marked as a defect area of the electronic element, so that the defect area of the electronic element is intelligently identified; and then the number ratio of all abnormal spectrum template windows in all windows is marked as abnormal probability +.>If the abnormality probability->And judging the electronic element as a bad product.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (8)
1. The electronic component defect rapid detection method based on artificial intelligence is characterized by comprising the following steps of:
acquiring a gray image of an electronic element;
performing tile blocking processing on the gray level image to obtain all windows, and performing Fourier transform operation on all windows to obtain spectrum images in all windows;
constructing a frequency power diagram of signal frequency and signal amplitude according to the frequency spectrum image in any window; presetting a plurality of signal amplitude intervals; acquiring all signal frequency intervals in each signal amplitude interval on a frequency power diagram; obtaining the waveform change of the window according to the number of the signal frequency intervals; sampling and sequencing the signal amplitudes of all the signal frequency intervals in the window according to the waveform change to obtain a waveform change sequence of the window; establishing a cross-correlation information function between windows according to waveform change sequences of any two windows; acquiring all abnormal spectrum template windows according to the cross-correlation information function;
positioning and identifying according to all abnormal spectrum template windows to obtain the defect position of the electronic element;
the method for establishing the cross-correlation information function between windows according to the waveform change sequences of the two windows comprises the following specific steps:
the waveform change sequences for any two windows are respectively recorded as；/>The elements in (a) are marked as->，/>The elements in (a) are recorded asThe method comprises the steps of carrying out a first treatment on the surface of the I.e. < ->、/>The method comprises the steps of carrying out a first treatment on the surface of the Window->The cross-correlation information function between is:
in the method, in the process of the invention,representing the cross-correlation degree of any two windows; />Element of (a)>Representation->Elements of (a) and (b); />Representation->Probability of simultaneous occurrence in signal waveform sequences within two windows simultaneously; />Representation element->At->Probability of occurrence of->Representation element->At->Probability of occurrence of (a);
the method for acquiring all abnormal spectrum template windows according to the cross-correlation information function comprises the following specific steps:
(1) All windows are combined pairwise to obtain all window pairs, the cross-correlation degree of each window pair is calculated, and the cross-correlation degree of all window pairs is recorded according to the sequence from big to small to obtain the sequence；
(2) Respectively obtain the firstPersonal window->The degree of cross-correlation with all other windows is denoted as set +.>；
(3) If it isIn the sequence->15 percent later, then window->Is an abnormal spectrum template window; if->Not in sequence15 percent later, then window->Not an abnormal spectrum template window;
all abnormal spectrum template windows are obtained.
2. The method for quickly detecting defects of electronic components based on artificial intelligence according to claim 1, wherein the step of obtaining a gray-scale image of the electronic components comprises the following specific steps:
acquiring the electronic element graph by using an industrial camera to obtain an original electronic element graph; and carrying out graying treatment on the original electronic element image to obtain a gray image of the electronic element.
3. The method for quickly detecting defects of electronic components based on artificial intelligence according to claim 1, wherein the tile-blocking processing is performed on the gray-scale image to obtain all windows, comprising the following specific steps:
performing tile blocking processing on the gray level image according to the position and the size of the tin spot to obtain all windows with equal size, and marking the windows as a window setOnly the area with tin spots is segmented, the size of the segmented area is consistent with the size of the tin spots of the gray level image, and the window set is +.>All windows in (1) are numbered, the obtained window set +.>Expressed as:
in the method, in the process of the invention,representing a set of all window compositions; />Representing the total number of windows in the window set; />Representing Window set +.>The%>A window.
4. The method for quickly detecting defects of electronic components based on artificial intelligence according to claim 1, wherein the constructing a frequency power map of signal frequency and signal amplitude according to the spectrum image in any window comprises the following specific steps:
performing Fourier transform operation on any window to obtain a frequency spectrum image of the window, and performing sampling analysis on the frequency spectrum image of the window to obtain the frequency signal change degree of the frequency spectrum image in the window when performing Fourier transform; and establishing a frequency power diagram of signal frequency and signal amplitude according to the frequency signal change degree of the frequency spectrum image, wherein the abscissa is the signal frequency, and the ordinate is the signal amplitude.
5. The method for quickly detecting defects of electronic components based on artificial intelligence according to claim 1, wherein the step of obtaining all signal frequency intervals in each signal amplitude interval on the frequency power map comprises the following specific steps:
the preset four signal amplitude intervals are respectively:,/>,/>,/>；
recording the signal frequency between any two adjacent wave troughs as a signal segment;
acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>；
Acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>；
Acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>；
Acquiring all signal frequency intervalsSaid signal frequency interval +.>The signal amplitude corresponding to each signal frequency belongs to the section +.>In the signal frequency interval +.>The number of the internal signal fragments is marked as +.>。
6. The method for rapidly detecting defects of electronic components based on artificial intelligence according to claim 5, wherein the step of obtaining the waveform change of the window according to the number of the signal frequency intervals comprises the following specific steps:
will beMarked as->The method comprises the steps of carrying out a first treatment on the surface of the Will->Marked as->The method comprises the steps of carrying out a first treatment on the surface of the Will->Marked as->；
Further acquiring the waveform change of the signal in any window,/>And->。
7. The method for quickly detecting defects of electronic components based on artificial intelligence according to claim 6, wherein the step of performing sampling and sequencing processing on signal amplitudes of all signal frequency intervals in a window according to waveform changes to obtain a waveform change sequence of the window comprises the following specific steps:
taking the signal amplitude 0 as a reference point, assigning 1 to the signal amplitude higher than the reference point, assigning-1 to the signal amplitude lower than the reference point, and assigning 0 to the signal amplitude equal to the reference point; amplitude sampling sequencing is carried out on all signal frequency intervals of any window from large to small; i.e. to waveform variations,/>And->Corresponding signal frequency interval +.>，/>，/>And->And performing amplitude sampling sequencing to obtain a waveform change sequence of the signal in any window.
8. The method for quickly detecting defects of electronic components based on artificial intelligence according to claim 1, wherein the step of locating and identifying to obtain the positions of defects of electronic components according to all abnormal spectrum template windows comprises the following specific steps:
preset parametersWindow set for all abnormal spectrum templates +.>Performing inverse Fourier transform to obtain window set->The position information in the gray level image of the electronic element is marked as a defect area of the electronic element, so that the defect area of the electronic element is intelligently identified; then the number proportion of all abnormal spectrum template windows in all windows is marked as abnormal probabilityRate->If the abnormality probability->And judging the electronic element as a bad product.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311329968.4A CN117115142B (en) | 2023-10-16 | 2023-10-16 | Electronic element defect rapid detection method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311329968.4A CN117115142B (en) | 2023-10-16 | 2023-10-16 | Electronic element defect rapid detection method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117115142A CN117115142A (en) | 2023-11-24 |
| CN117115142B true CN117115142B (en) | 2024-01-26 |
Family
ID=88805698
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311329968.4A Active CN117115142B (en) | 2023-10-16 | 2023-10-16 | Electronic element defect rapid detection method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117115142B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120089357A1 (en) * | 2010-10-07 | 2012-04-12 | Choudur Lakshminarayan | Method and apparatus for identifying anomalies of a signal |
| CN109285140A (en) * | 2018-07-27 | 2019-01-29 | 广东工业大学 | A kind of printed circuit board image registration appraisal procedure |
| CN115311263A (en) * | 2022-10-09 | 2022-11-08 | 南通市通州区顺行纺织有限公司 | Method and system for detecting textile printing defects |
-
2023
- 2023-10-16 CN CN202311329968.4A patent/CN117115142B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120089357A1 (en) * | 2010-10-07 | 2012-04-12 | Choudur Lakshminarayan | Method and apparatus for identifying anomalies of a signal |
| CN109285140A (en) * | 2018-07-27 | 2019-01-29 | 广东工业大学 | A kind of printed circuit board image registration appraisal procedure |
| CN115311263A (en) * | 2022-10-09 | 2022-11-08 | 南通市通州区顺行纺织有限公司 | Method and system for detecting textile printing defects |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117115142A (en) | 2023-11-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114972357B (en) | Roller surface defect detection method and system based on image processing | |
| CN113989228A (en) | Method for detecting defect area of color texture fabric based on self-attention | |
| Dubey et al. | Maximally stable extremal region marking-based railway track surface defect sensing | |
| CN114170208A (en) | Paper product defect detection method based on artificial intelligence | |
| CN109239073A (en) | A kind of detection method of surface flaw for body of a motor car | |
| JP2008090529A (en) | Abnormality detection device, abnormality detection method | |
| Fekri-Ershad et al. | A robust approach for surface defect detection based on one dimensional local binary patterns | |
| NU | Automatic detection of texture defects using texture-periodicity and Gabor wavelets | |
| Jian et al. | Imbalanced defect classification for mobile phone screen glass using multifractal features and a new sampling method | |
| CN114700587B (en) | Missing welding defect real-time detection method and system based on fuzzy inference and edge calculation | |
| Sun et al. | Data-driven fault diagnosis method based on second-order time-reassigned multisynchrosqueezing transform and evenly mini-batch training | |
| CN109708877B (en) | Mechanical fault analysis method based on wavelet fuzzy recognition and image analysis theory | |
| CN114841957A (en) | Steel plate surface defect detection method based on deep learning | |
| CN117115142B (en) | Electronic element defect rapid detection method based on artificial intelligence | |
| CN117237747A (en) | Hardware defect classification and identification method based on artificial intelligence | |
| Lin et al. | Detection of surface flaws on textured LED lenses using wavelet packet transform based partial least squares techniques | |
| CN116777917A (en) | Defect detection method and system for optical cable production | |
| Mansano et al. | Inspection of metallic surfaces using local binary patterns | |
| CN110059679A (en) | The non-targeted detection method of pollutant in a kind of feedstuff based on near-infrared image | |
| Niskanen et al. | Experiments with SOM based inspection of wood | |
| CN114596296A (en) | High-sensitivity hot-rolled steel coil end surface defect identification system and method | |
| CN105115987B (en) | Taper roller upside-down mounting defect inspection method based on digital filtering | |
| Aydin et al. | A vision based inspection system using gaussian mixture model based interactive segmentation | |
| Konovalenko et al. | Using fuzzy sets to estimate the geometric parameters of surface damage | |
| Hu et al. | Research on rice grain shape detection method based on machine vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |