CN117152157A - Electronic element identification method based on artificial intelligence - Google Patents
Electronic element identification method based on artificial intelligence Download PDFInfo
- Publication number
- CN117152157A CN117152157A CN202311432425.5A CN202311432425A CN117152157A CN 117152157 A CN117152157 A CN 117152157A CN 202311432425 A CN202311432425 A CN 202311432425A CN 117152157 A CN117152157 A CN 117152157A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- feature
- characteristic
- standard
- electronic
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 22
- 239000013598 vector Substances 0.000 claims abstract description 304
- 238000001514 detection method Methods 0.000 claims abstract description 18
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000010586 diagram Methods 0.000 claims description 11
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 abstract description 2
- 239000003990 capacitor Substances 0.000 description 8
- 238000004519 manufacturing process Methods 0.000 description 8
- 239000003086 colorant Substances 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000000463 material Substances 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30148—Semiconductor; IC; Wafer
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The application discloses an electronic component identification method based on artificial intelligence, which relates to the field of electronic component detection technology, in particular to an electronic component identification method based on artificial intelligence, comprising the steps of acquiring a circuit board image of an integrated circuit board, preprocessing the circuit board image, and acquiring a component image of an electronic component; acquiring a first feature vector of the electronic element based on the element image, and judging the element type of the electronic element; establishing a time period, acquiring the characteristic vector of each electronic element identified in the time period, and acquiring a characteristic vector group; judging whether the feature vector is qualified or not, if not, rejecting to obtain a corrected feature vector group; calculating the feature difference value between every two feature vectors in the correction feature vector group based on the correction feature vector group to obtain a feature difference value sequence; and judging the characteristic change rule of the characteristic vector, and correcting the standard characteristic vector. The application has the effect of improving the self-correcting capability in the detection process of the electronic element.
Description
Technical Field
The application relates to the field of electronic element detection technology, in particular to an electronic element identification method based on artificial intelligence.
Background
Electronic components are the basis of electronic products, which are filled everywhere in modern life, and the basis for supporting these electronic products for operation is the electronic components. The common electronic components include resistors, capacitors, inductors, potentiometers, transformers and the like, and detection of the electronic components plays an important role in production of the electronic components, so that defective products are reduced to enter the market, and problems occur in the use process of the electronic products.
In the prior art, when electronic components on an integrated circuit board are identified, a machine identification mode is generally adopted to identify the electronic components on the integrated circuit board, information of standard electronic components is generally input into a machine as a judgment standard, but the same electronic components can be found to be different to a certain extent due to production batch, material characteristics and external environment factors in an actual detection process, but the same electronic components are identified as other types of electronic components in the detection process due to immobilization of the detection standard in the detection process, so that errors occur in identification.
Therefore, how to improve the self-correction capability in the electronic device inspection process is a urgent problem to be solved.
Disclosure of Invention
In order to improve self-correction capability in the electronic element detection process, the application provides an electronic element identification method based on artificial intelligence.
The application provides an electronic element identification method based on artificial intelligence, which adopts the following technical scheme:
acquiring circuit board information of an integrated circuit board in advance, acquiring standard feature vectors of electronic elements on the integrated circuit board based on the circuit board information, collecting all the standard feature vectors, and establishing a standard set;
acquiring a circuit board image of the integrated circuit board, preprocessing the circuit board image, and acquiring an element image of the electronic element;
based on the element image, acquiring a first feature vector of the electronic element, comparing the first feature vector with the standard feature vector in the standard set, and judging the element type of the electronic element;
establishing a time period, acquiring the characteristic vector of each identified electronic element in the time period, and arranging the characteristic vectors according to a time relation to acquire a characteristic vector group;
based on the feature vector group, sequentially comparing the feature vector in the feature vector group with the standard feature vector, judging whether the feature vector is qualified or not, and eliminating if the feature vector is unqualified, so as to obtain a corrected feature vector group;
calculating feature difference values between every two feature vectors in the correction feature vector group based on the correction feature vector group, and arranging the feature difference values to obtain a feature difference value sequence;
and analyzing the characteristic difference value sequence, judging the characteristic change rule of the characteristic vector, and correcting the standard characteristic vector based on the characteristic change rule.
By adopting the technical scheme, the integrated circuit board is identified, the image of the electronic element is obtained, the characteristic vector of the electronic element is extracted, the type of the electronic element is judged according to the characteristic vector of the electronic element, meanwhile, the characteristic vectors of different electronic elements in a time period are obtained, a characteristic vector group is established, the standard characteristic vector is corrected in real time according to the change condition of the characteristic vector in the characteristic vector group, and the self-correction capability in the detection process of the electronic element is improved.
Preferably, the step of acquiring a circuit board image of the integrated circuit board and preprocessing the circuit board image to acquire a component image of the electronic component includes:
acquiring a circuit board image of the integrated circuit board, identifying the background content of the circuit board image, and removing the background content to acquire an element base diagram of the electronic element;
acquiring shape characteristics of each electronic element based on the element base map, and arranging the electronic elements based on the shape characteristics to acquire an element arrangement map of the electronic elements;
and amplifying and rotating each electronic element based on the element arrangement diagram to acquire an element image of the electronic element.
By adopting the technical scheme, the image of the integrated circuit board is preprocessed, the background image of the integrated circuit board is removed, the base images of a plurality of electronic elements are obtained, the element arrangement diagram of the electronic elements is generated according to the shape characteristics of the electronic elements, the element arrangement diagram of the electronic elements is amplified and rotated, the original image of the electronic elements is generated, and the accuracy of obtaining the image of the electronic elements is improved.
Preferably, the step of acquiring a first feature vector of the electronic component based on the component image, comparing the first feature vector with the standard feature vector in the standard set, and determining a component type of the electronic component includes:
based on the element image, extracting the characteristics of the electronic element to obtain a first characteristic vector of the electronic element;
based on the standard set, the first feature vector is matched with the standard feature vector in the standard set one by one, and the matching degree is obtained;
sorting the obtained matching degrees, selecting a standard feature vector with the highest matching degree as a target feature vector, and calculating a target feature difference value between the first feature vector and the target feature vector;
comparing the target characteristic difference value with a preset threshold value, and judging whether the target characteristic difference value is larger than the threshold value or not;
if the electronic component is larger than the first threshold value, the electronic component is incorrect;
if the electronic component is smaller than the standard feature vector, the electronic component is correct, and the type of the electronic component is confirmed to be the same as the type of the electronic component corresponding to the standard feature vector.
By adopting the technical scheme, the first feature vector of the electronic element is obtained and matched with the standard feature vector, the generated matching degree is sequenced, the most target feature vector with the highest matching degree is selected, the difference value between the first feature vector and the target feature vector is obtained according to the target feature vector, if the difference value is too large, the electronic element is incorrect, the recognition range of the electronic element is improved, and the recognition accuracy of the electronic element is improved.
Preferably, the step of extracting features of the electronic component based on the component image to obtain a first feature vector of the electronic component includes:
the first feature vector includes at least one of a color feature, a shape feature, and a texture feature;
wherein the color features include at least one of a color distribution category, a color distribution location, and a color distribution duty cycle.
By adopting the technical scheme, the first feature vector is composed of at least one of color features, shape features and texture features, and meanwhile, the color features are composed of at least one of distribution types, distribution positions and distribution duty ratios of colors, so that the replaceability of the feature vector is reduced.
Preferably, the step of sequentially comparing the feature vector in the feature vector group with the standard feature vector based on the feature vector group, judging whether the feature vector is qualified, and rejecting if not, to obtain a corrected feature vector group includes:
performing traversal matching on the feature vectors in the feature vector group based on the standard feature vector, and acquiring matching values;
based on the matching values, sequencing all the obtained matching values to obtain a matching value sequence, and taking the highest matching value in the matching value sequence as a target matching value;
and comparing the target matching value with a preset threshold value, if the target matching value is smaller than the threshold value, judging that the feature vector is unqualified, and eliminating the unqualified feature vector to obtain a corrected feature vector group.
By adopting the technical scheme, the standard feature vectors are used as the reference, the feature vectors in the feature vector group are traversed and matched at the same time, the acquired matching values are ordered, the sound field matching value sequence is obtained, meanwhile, unqualified matching values in the matching value sequence are removed, the feature vector group is corrected, and the possibility that the feature vector group is interfered by the interference information possibly existing is reduced.
Preferably, the step of comparing the target matching value with a preset threshold, if the target matching value is smaller than the threshold, judging that the feature vector is not qualified, and rejecting the unqualified feature vector to obtain a corrected feature vector set further includes:
collecting the unqualified feature vectors, and obtaining matching values of the unqualified feature vectors and the standard vectors;
analyzing a difference point between the unqualified feature vector and the standard feature vector based on the matching value;
and judging possible reasons for generating the unqualified feature vectors based on the difference points, generating correction comments based on the possible reasons, and issuing the correction comments.
By adopting the technical scheme, the rejected unqualified feature vectors are collected, the matching value between the rejected unqualified feature vectors and the standard feature vectors is obtained, the difference point between the unqualified feature vectors and the standard vectors is judged according to the matching value, the difference point is analyzed, the possible cause of damage or production problems to the electronic element in the production and transportation process is judged, and the possible cause is output, so that the improvement of the production and transportation mode is facilitated, and the utilization rate of the unqualified electronic element is improved.
Preferably, the step of analyzing the feature difference sequence, judging a feature change rule of the feature vector, and correcting the standard feature vector based on the feature change rule includes:
performing traversal comparison on the characteristic difference value sequences, and combining the characteristic difference values smaller than a preset value to obtain new characteristic difference value sequences;
analyzing the difference value in the new characteristic difference value sequence to obtain a change curve of the characteristic difference value sequence;
carrying out regression processing on the change curve to obtain a regression change curve, and obtaining a characteristic change rule of the characteristic vector based on the regression change curve;
and based on the characteristic change rule, evolution is carried out on the standard characteristic vector, wherein the evolution process is later than two unit times of the current detection time, and the unit time is the time required for detecting one integrated circuit board.
By adopting the technical scheme, the characteristic difference values with smaller differences in the characteristic difference value sequences are combined to obtain a new characteristic difference value sequence, data in the new characteristic difference value sequence are subjected to contouring to obtain a change curve, the change curve is regressed, the change rule of the characteristic vector is judged, the standard characteristic vector is evolved according to the change rule, and meanwhile, the evolution process is later than the current detection time, so that the adaptability of the standard value to the product in the detection process of the electronic element is improved.
Preferably, after the step of performing regression processing on the change curve to obtain a regression change curve and obtaining the feature change rule of the feature vector based on the regression change curve, the method further includes:
correcting the characteristic difference value which is not on the regression variation curve based on the regression variation curve, and obtaining a corrected value;
judging a correction relation between the two feature vectors corresponding to the feature difference value based on the correction value;
and acquiring correction rules among the feature vectors based on the correction relation, and integrating the correction rules into the feature change rules.
By adopting the technical scheme, the characteristic difference value on the regression change curve is corrected, the correction value is obtained, the correction relation between the two characteristic vectors is judged according to the correction value, the correction rule between the characteristic vectors is judged according to the correction relation, and the correction rule is integrated into the characteristic change rule, so that the accuracy of the characteristic change rule in describing the characteristic vector change is improved.
Preferably, after the step of analyzing the feature difference sequence and judging the feature change rule of the feature vector and correcting the standard feature vector based on the feature change rule, the method further includes:
based on the characteristic change rule, carrying out characteristic prediction on the electronic elements of the same type, and obtaining a characteristic prediction result;
judging a predicted feature vector of the next electronic element based on the predicted result, and generating an advance based on the standard feature vector;
and extracting the characteristics of the next electronic element, obtaining an actual characteristic vector, comparing the actual characteristic vector with the standard characteristic vector based on the advance, and judging the type of the electronic element.
By adopting the technical scheme, the electronic element to be identified is predicted according to the characteristic change rule, and the lead is generated according to the prediction result, when the electronic element is identified, the actual characteristic vector is compared with the standard characteristic vector, if a larger error occurs, the lead can be corrected, and the accuracy of identifying the electronic element is improved.
In summary, the present application includes at least one of the following beneficial technical effects:
1. the method comprises the steps of identifying an integrated circuit board, acquiring an image of an electronic element, extracting a characteristic vector of the electronic element, judging the type of the electronic element according to the characteristic vector of the electronic element, acquiring characteristic vectors of different electronic elements in a time period, establishing a characteristic vector group, correcting a standard characteristic vector in real time through changing conditions of the characteristic vector in the characteristic vector group, and improving self-correction capability in the detection process of the electronic element.
2. Collecting the rejected unqualified feature vectors, acquiring a matching value between the rejected unqualified feature vectors and the standard feature vectors, judging a difference point between the unqualified feature vectors and the standard vectors according to the matching value, analyzing the difference point, judging possible reasons for damage or production problems of the electronic element in the production and transportation process of the electronic element, and outputting the possible reasons, so that the improvement of the production and transportation mode is facilitated, and the utilization rate of the unqualified electronic element is improved.
3. And predicting the electronic element to be identified according to the characteristic change rule, generating an advance according to the prediction result, and when the electronic element is identified, comparing the actual characteristic vector with the standard characteristic vector, if a larger error occurs, correcting the electronic element by the advance, thereby improving the accuracy of electronic element identification.
Drawings
FIG. 1 is a flowchart showing steps of an electronic component recognition method based on artificial intelligence according to the embodiment;
FIG. 2 is a schematic flow chart of the substeps of steps S200 and S300 in the electronic component recognition method based on artificial intelligence according to the present embodiment;
FIG. 3 is a flow chart showing the sub-steps after steps S301, S500 and S500 in the electronic component recognition method based on artificial intelligence according to the present embodiment;
FIG. 4 is a flow chart showing the substeps of steps S700 and S703 in the electronic component recognition method based on artificial intelligence according to the present embodiment;
fig. 5 is a schematic flow chart of the substeps after step S700 in the electronic component recognition method based on artificial intelligence according to the present embodiment.
Detailed Description
The application is described in further detail below with reference to fig. 1-5.
The embodiment of the application discloses an electronic element identification method based on artificial intelligence.
In this embodiment, referring to fig. 1, the electronic component recognition method based on artificial intelligence includes the following steps:
s100: the method comprises the steps of acquiring circuit board information of an integrated circuit board in advance, acquiring standard feature vectors of electronic elements on the integrated circuit board based on the circuit board information, collecting all the standard feature vectors, and establishing a standard set;
s200: acquiring a circuit board image of an integrated circuit board, preprocessing the circuit board image, and acquiring an element image of an electronic element;
s300: based on the element image, acquiring a first feature vector of the electronic element, comparing the first feature vector with a standard feature vector in a standard set, and judging the element type of the electronic element;
s400: establishing a time period, acquiring the characteristic vector of each electronic element identified in the time period, and arranging the characteristic vectors according to the time relation to acquire a characteristic vector group;
s500: based on the feature vector group, sequentially comparing the feature vector in the feature vector group with the standard feature vector, judging whether the feature vector is qualified or not, and if not, removing to obtain a corrected feature vector group;
s600: calculating characteristic difference values between every two characteristic vectors in the corrected characteristic vector group based on the corrected characteristic vector group, and arranging the characteristic difference values to obtain a characteristic difference value sequence;
s700: and analyzing the characteristic difference value sequence, judging the characteristic change rule of the characteristic vector, and correcting the standard characteristic vector based on the characteristic change rule.
It should be noted that the above steps are only preferred embodiments, and in the specific implementation process, part of the steps may be exchanged without affecting the overall implementation effect.
Referring to fig. 2, in step S200, a circuit board image of an integrated circuit board is acquired, and the circuit board image is preprocessed to acquire a component image of an electronic component, including:
s201: acquiring a circuit board image of the integrated circuit board, identifying the background content of the circuit board image, and removing the background content to acquire an element bottom diagram of the electronic element;
s202: acquiring shape characteristics of each electronic element based on the element base diagram, and arranging the electronic elements based on the shape characteristics to acquire an element arrangement diagram of the electronic elements;
s203: and amplifying and rotating each electronic element based on the element arrangement diagram to acquire an element image of the electronic element.
In operation, the image of the integrated circuit board is processed to obtain the image of the electronic component. For example, the background color of an integrated circuit board is green, a circuit board image of the integrated circuit board is obtained, the background of the circuit board is eliminated according to the background color, the base pattern of the electronic elements is reserved, the electronic elements are arranged according to the shapes of different electronic elements, an element arrangement diagram is obtained, one resistor element is arranged, and if the resistor element is smaller, the element is amplified and rotated, so that an element image of the resistor element is obtained.
Referring to fig. 2, in step S300, a first feature vector of an electronic component is acquired based on a component image, and the first feature vector is compared with a standard feature vector in a standard set to determine a component type of the electronic component, including:
s301: based on the element image, extracting the characteristics of the electronic element to obtain a first characteristic vector of the electronic element;
s302: based on a standard set, carrying out one-to-one matching on the first feature vector and the standard feature vector in the standard set to obtain matching degree;
s303: sorting the obtained matching degrees, selecting a standard feature vector with the highest matching degree as a target feature vector, and calculating a target feature difference value between the first feature vector and the target feature vector;
s304: comparing the target characteristic difference value with a preset threshold value, and judging whether the target characteristic difference value is larger than the threshold value or not;
s305: if the value is larger than the preset value, the electronic element is incorrect;
s306: if the electronic component is smaller than the standard feature vector, the electronic component is correct, and the type of the electronic component is confirmed to be the same as the type of the electronic component corresponding to the standard feature vector.
In use, a first feature vector of the electronic element is obtained according to the original image of the electronic element, and is compared with a standard feature vector to judge the type of the electronic element. For example, when a certain electronic element is identified, a feature vector of the electronic element is obtained and compared with a standard feature vector, the matching degree of the feature vector of the electronic element and two standard feature vectors is found to be very high, wherein the electronic elements corresponding to the two standard feature vectors are resistors and capacitors, a target feature difference value is calculated according to the two matching degrees, the target feature difference value of the feature vector of the capacitor is found to be larger than a preset threshold value, and the target feature difference value of the feature vector of the resistor is smaller than the preset threshold value, so that the electronic element can be judged to be a resistor element.
Referring to fig. 3, in step S301, a step of extracting features of an electronic component based on a component image to obtain a first feature vector of the electronic component includes:
s3011: the first feature vector includes at least one of a color feature, a shape feature, and a texture feature;
s3012: wherein the color features include at least one of a color distribution category, a color distribution location, and a color distribution duty cycle.
In use, the first feature vector is composed of a plurality of features. For example, if a columnar electronic element has a colored annular mark, an image of the electronic element is acquired and a first feature vector thereof is extracted, wherein the first feature vector comprises shape features and color features, and meanwhile, the color types of the annular mark on the electronic element, the positions of different colors and the ratio of different colors can be distinguished according to the color features.
Referring to fig. 3, in step S500, based on the feature vector set, feature vectors in the feature vector set are sequentially compared with standard feature vectors, whether the feature vectors are qualified is determined, and if the feature vectors are not qualified, the step of eliminating is performed to obtain a corrected feature vector set, including:
s501: based on the feature vector group, traversing and matching the feature vectors in the feature vector group by taking the standard feature vector as a reference, and obtaining a matching value;
s502: based on the matching values, sequencing all the obtained matching values to obtain a matching value sequence, and taking the highest matching value in the matching value sequence as a target matching value;
s503: and comparing the target matching value with a preset threshold value, if the target matching value is smaller than the threshold value, judging that the feature vector is unqualified, and eliminating the unqualified feature vector to obtain a corrected feature vector group.
In application, feature vectors acquired within a period of time are organized into feature vector groups, the feature vector groups are sequentially matched with standard feature vectors, matching values are ordered, unqualified matching values are removed, and the feature vector groups are corrected. For example, within ten minutes, 20 integrated circuit boards are detected, the feature vectors of the unified electronic components are grouped and matched with the standard feature vectors, the matching values of the 20 electronic components are ordered, meanwhile, if 3 disqualified matching values exist in the matching values, the feature vectors corresponding to the 3 matching values are removed, and the 17 remaining feature vectors form a corrected feature vector group.
Referring to fig. 3, after step S500, comparing the target matching value with a preset threshold, if the target matching value is smaller than the threshold, determining that the feature vector is not qualified, and rejecting the unqualified feature vector to obtain a corrected feature vector set, the step further includes:
s511: collecting unqualified feature vectors, and obtaining matching values of the unqualified feature vectors and standard vectors;
s512: based on the matching value, analyzing the difference points between the unqualified feature vector and the standard feature vector;
s513: and judging possible reasons for generating unqualified feature vectors based on the difference points, generating correction comments based on the possible reasons, and issuing the correction comments.
In the application, unqualified feature vectors and matching values thereof are collected, difference points between the feature vectors and standard feature vectors are judged according to the matching values, and correction comments are generated according to the difference points. For example, if a capacitor of a certain signal is knocked during transportation, the appearance is damaged, but the function is normal, and the capacitor is identified as unqualified during the subsequent detection process, the characteristic vector of the capacitor can be judged to be compared with the standard characteristic vector of the standard capacitor according to the condition, and the difference point is the position where the capacitor is knocked, so that a transportation correction suggestion can be generated according to the knocked condition.
Referring to fig. 4, in step S700, the step of analyzing the feature difference sequence, determining a feature change rule of the feature vector, and correcting the standard feature vector based on the feature change rule includes:
s701: performing traversal comparison on the characteristic difference value sequences, and combining the characteristic difference values smaller than a preset value to obtain new characteristic difference value sequences;
s702: analyzing the difference value in the new characteristic difference value sequence to obtain a change curve of the characteristic difference value sequence;
s703: carrying out regression processing on the change curve to obtain a regression change curve, and obtaining a characteristic change rule of the characteristic vector based on the regression change curve;
s704: based on the characteristic change rule, the standard characteristic vector is evolved, and the evolution process is later than two unit time of the current detection time, wherein the unit time is the time required by detecting one integrated circuit board.
In the application, the characteristic difference values close to the data in the characteristic difference value sequence are combined, the data processing amount is reduced, regression processing is carried out according to the sound field change curve in the new characteristic difference value sequence, the change rule of the characteristic vector is judged, and the standard characteristic vector is evolved. For example, in a certain period of time, a set of characteristic difference value sequences are obtained, 100 sets of data are included in the sequences, then the data with similar characteristic difference values are combined, 30 sets of data are included in the obtained new characteristic difference value sequences, a change curve is obtained according to the change condition of the 30 sets of characteristic difference values, regression processing is performed, the change rule of the 30 sets of characteristic difference values is judged, and the standard characteristic vector is evolved according to the change rule, so that the standard characteristic vector can meet the subsequent detection task.
Referring to fig. 4, after step S703, the step of performing regression processing on the change curve to obtain a regression change curve, and obtaining the feature change rule of the feature vector based on the regression change curve further includes:
s7031: correcting the characteristic difference value which is not on the regression change curve based on the regression change curve, and obtaining a corrected value;
s7032: judging a correction relation between two feature vectors corresponding to the feature difference value based on the correction value;
s7033: based on the correction relation, a correction rule among the feature vectors is acquired, and the correction rule is integrated into a feature change rule.
In the application, the characteristic difference value which is not on the regression change curve is corrected, a correction value is obtained, and according to the correction value, a correction rule is analyzed and integrated into the characteristic change rule, so that the characteristic change rule is more complete. For example, if 10 feature differences exist on the regression curve graph and are not on the curve, calculating a correction value between the 10 feature differences and the curve, judging a correction relation between 20 feature vectors corresponding to the 10 feature differences according to the correction value, analyzing and summarizing correction rules between the 20 feature vectors based on the correction relation, and integrating the correction rules into the feature change rules.
Referring to fig. 5, after step S700, the step of analyzing the feature difference sequence, judging the feature change rule of the feature vector, and correcting the standard feature vector based on the feature change rule further includes:
s701: based on the characteristic change rule, carrying out characteristic prediction on the electronic elements of the same type, and obtaining a characteristic prediction result;
s702: judging a predicted feature vector of the next electronic element based on the predicted result, and generating an advance based on the standard feature vector;
s703: and extracting the characteristics of the next electronic element, obtaining an actual characteristic vector, comparing the actual characteristic vector with a standard characteristic vector based on the advance, and judging the type of the electronic element.
In application, according to the characteristic change rule, the electronic element which is possibly identified is subjected to characteristic prediction, and the lead is generated by combining the prediction result and the standard characteristic vector. For example, a characteristic change rule is obtained in the process of identifying the resistor element of the current round, then the characteristic vector of the resistor element to be identified next is predicted according to the rule, meanwhile, the standard characteristic vector is combined to generate an advance, when the resistor element of the next round is identified next, the identification efficiency is low possibly due to the standard characteristic vector or the characteristic vector of the resistor element, and at the moment, the standard characteristic vector or the characteristic vector of the resistor element is corrected according to the advance, so that the identification rate is improved.
The above embodiments are not intended to limit the scope of the present application, so: all equivalent changes in structure, shape and principle of the application should be covered in the scope of protection of the application.
Claims (9)
1. The electronic element identification method based on artificial intelligence is characterized by comprising the following steps of:
acquiring circuit board information of an integrated circuit board in advance, acquiring standard feature vectors of electronic elements on the integrated circuit board based on the circuit board information, collecting all the standard feature vectors, and establishing a standard set;
acquiring a circuit board image of the integrated circuit board, preprocessing the circuit board image, and acquiring an element image of the electronic element;
based on the element image, acquiring a first feature vector of the electronic element, comparing the first feature vector with the standard feature vector in the standard set, and judging the element type of the electronic element;
establishing a time period, acquiring the characteristic vector of each identified electronic element in the time period, and arranging the characteristic vectors according to a time relation to acquire a characteristic vector group;
based on the feature vector group, sequentially comparing the feature vector in the feature vector group with the standard feature vector, judging whether the feature vector is qualified or not, and eliminating if the feature vector is unqualified, so as to obtain a corrected feature vector group;
calculating feature difference values between every two feature vectors in the correction feature vector group based on the correction feature vector group, and arranging the feature difference values to obtain a feature difference value sequence;
and analyzing the characteristic difference value sequence, judging the characteristic change rule of the characteristic vector, and correcting the standard characteristic vector based on the characteristic change rule.
2. The electronic component recognition method based on artificial intelligence according to claim 1, wherein the steps of acquiring a circuit board image of the integrated circuit board and preprocessing the circuit board image to acquire a component image of the electronic component, comprise:
acquiring a circuit board image of the integrated circuit board, identifying the background content of the circuit board image, and removing the background content to acquire an element base diagram of the electronic element;
acquiring shape characteristics of each electronic element based on the element base map, and arranging the electronic elements based on the shape characteristics to acquire an element arrangement map of the electronic elements;
and amplifying and rotating each electronic element based on the element arrangement diagram to acquire an element image of the electronic element.
3. The electronic component recognition method based on artificial intelligence according to claim 1, wherein the step of acquiring a first feature vector of the electronic component based on the component image, comparing the first feature vector with the standard feature vector in the standard set, and judging a component type of the electronic component comprises:
based on the element image, extracting the characteristics of the electronic element to obtain a first characteristic vector of the electronic element;
based on the standard set, the first feature vector is matched with the standard feature vector in the standard set one by one, and the matching degree is obtained;
sorting the obtained matching degrees, selecting a standard feature vector with the highest matching degree as a target feature vector, and calculating a target feature difference value between the first feature vector and the target feature vector;
comparing the target characteristic difference value with a preset threshold value, and judging whether the target characteristic difference value is larger than the threshold value or not;
if the electronic component is larger than the first threshold value, the electronic component is incorrect;
if the electronic component is smaller than the standard feature vector, the electronic component is correct, and the type of the electronic component is confirmed to be the same as the type of the electronic component corresponding to the standard feature vector.
4. The electronic component recognition method based on artificial intelligence according to claim 3, wherein the step of extracting features of the electronic component based on the component image to obtain a first feature vector of the electronic component comprises:
the first feature vector includes at least one of a color feature, a shape feature, and a texture feature;
wherein the color features include at least one of a color distribution category, a color distribution location, and a color distribution duty cycle.
5. The electronic component recognition method based on artificial intelligence according to claim 1, wherein the step of comparing the feature vector in the feature vector group with the standard feature vector in order based on the feature vector group, judging whether the feature vector is qualified, and if not, removing to obtain a corrected feature vector group comprises:
performing traversal matching on the feature vectors in the feature vector group based on the standard feature vector, and acquiring matching values;
based on the matching values, sequencing all the obtained matching values to obtain a matching value sequence, and taking the highest matching value in the matching value sequence as a target matching value;
and comparing the target matching value with a preset threshold value, if the target matching value is smaller than the threshold value, judging that the feature vector is unqualified, and eliminating the unqualified feature vector to obtain a corrected feature vector group.
6. The electronic component recognition method based on artificial intelligence according to claim 5, wherein the step of comparing the target matching value with a preset threshold value, if the target matching value is smaller than the threshold value, determining that the feature vector is not qualified, and rejecting the feature vector that is not qualified, to obtain a corrected feature vector group, further comprises:
collecting the unqualified feature vectors, and obtaining matching values of the unqualified feature vectors and the standard vectors;
analyzing a difference point between the unqualified feature vector and the standard feature vector based on the matching value;
and judging possible reasons for generating the unqualified feature vectors based on the difference points, generating correction comments based on the possible reasons, and issuing the correction comments.
7. The electronic component recognition method based on artificial intelligence according to claim 1, wherein the step of analyzing the sequence of feature differences, judging a feature change rule of the feature vector, and correcting the standard feature vector based on the feature change rule comprises:
performing traversal comparison on the characteristic difference value sequences, and combining the characteristic difference values smaller than a preset value to obtain new characteristic difference value sequences;
analyzing the difference value in the new characteristic difference value sequence to obtain a change curve of the characteristic difference value sequence;
carrying out regression processing on the change curve to obtain a regression change curve, and obtaining a characteristic change rule of the characteristic vector based on the regression change curve;
and based on the characteristic change rule, evolution is carried out on the standard characteristic vector, wherein the evolution process is later than two unit times of the current detection time, and the unit time is the time required for detecting one integrated circuit board.
8. The method for identifying electronic components based on artificial intelligence according to claim 7, wherein after the steps of performing regression processing on the change curve to obtain a regression change curve and obtaining the feature change rule of the feature vector based on the regression change curve, further comprising:
correcting the characteristic difference value which is not on the regression variation curve based on the regression variation curve, and obtaining a corrected value;
judging a correction relation between the two feature vectors corresponding to the feature difference value based on the correction value;
and acquiring correction rules among the feature vectors based on the correction relation, and integrating the correction rules into the feature change rules.
9. The electronic component recognition method according to claim 1, wherein the step of analyzing the feature difference sequence to determine a feature change rule of the feature vector, and correcting the standard feature vector based on the feature change rule further comprises:
based on the characteristic change rule, carrying out characteristic prediction on the electronic elements of the same type, and obtaining a characteristic prediction result;
judging a predicted feature vector of the next electronic element based on the predicted result, and generating an advance based on the standard feature vector;
and extracting the characteristics of the next electronic element, obtaining an actual characteristic vector, comparing the actual characteristic vector with the standard characteristic vector based on the advance, and judging the type of the electronic element.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311432425.5A CN117152157B (en) | 2023-10-31 | 2023-10-31 | Electronic element identification method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311432425.5A CN117152157B (en) | 2023-10-31 | 2023-10-31 | Electronic element identification method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117152157A true CN117152157A (en) | 2023-12-01 |
| CN117152157B CN117152157B (en) | 2023-12-29 |
Family
ID=88912481
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311432425.5A Active CN117152157B (en) | 2023-10-31 | 2023-10-31 | Electronic element identification method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117152157B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6282440B1 (en) * | 1999-12-31 | 2001-08-28 | Ge Marquette Medical Systems, Inc. | Method to identify electrode placement |
| CN105631458A (en) * | 2015-12-22 | 2016-06-01 | 广州视源电子科技股份有限公司 | Electronic component sample labeling method and device |
| CN112862770A (en) * | 2021-01-29 | 2021-05-28 | 珠海迪沃航空工程有限公司 | Defect analysis and diagnosis system, method and device based on artificial intelligence |
| CN112990080A (en) * | 2020-10-16 | 2021-06-18 | 陈晶晶 | Rule determination method based on big data and artificial intelligence |
| US20210224479A1 (en) * | 2020-01-19 | 2021-07-22 | Beijing Xiaomi Pinecone Electronics Co., Ltd. | Method for processing information, and storage medium |
| CN114596294A (en) * | 2022-03-14 | 2022-06-07 | 国网黑龙江省电力有限公司供电服务中心 | Circuit element missing detection method based on difference image, electronic device and storage medium |
| CN114693554A (en) * | 2022-03-28 | 2022-07-01 | 唐山学院 | Big data image processing method and system |
| CN115661115A (en) * | 2022-11-09 | 2023-01-31 | 上海闻泰信息技术有限公司 | Component detection method, device, electronic equipment and storage medium |
-
2023
- 2023-10-31 CN CN202311432425.5A patent/CN117152157B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6282440B1 (en) * | 1999-12-31 | 2001-08-28 | Ge Marquette Medical Systems, Inc. | Method to identify electrode placement |
| CN105631458A (en) * | 2015-12-22 | 2016-06-01 | 广州视源电子科技股份有限公司 | Electronic component sample labeling method and device |
| US20210224479A1 (en) * | 2020-01-19 | 2021-07-22 | Beijing Xiaomi Pinecone Electronics Co., Ltd. | Method for processing information, and storage medium |
| CN112990080A (en) * | 2020-10-16 | 2021-06-18 | 陈晶晶 | Rule determination method based on big data and artificial intelligence |
| CN112862770A (en) * | 2021-01-29 | 2021-05-28 | 珠海迪沃航空工程有限公司 | Defect analysis and diagnosis system, method and device based on artificial intelligence |
| CN114596294A (en) * | 2022-03-14 | 2022-06-07 | 国网黑龙江省电力有限公司供电服务中心 | Circuit element missing detection method based on difference image, electronic device and storage medium |
| CN114693554A (en) * | 2022-03-28 | 2022-07-01 | 唐山学院 | Big data image processing method and system |
| CN115661115A (en) * | 2022-11-09 | 2023-01-31 | 上海闻泰信息技术有限公司 | Component detection method, device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117152157B (en) | 2023-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108257121B (en) | Method, apparatus, storage medium and the terminal device that product defects detection model updates | |
| CN110992349A (en) | Underground pipeline abnormity automatic positioning and identification method based on deep learning | |
| CN109060817B (en) | Artificial intelligence reinspection system and method thereof | |
| CN116188475B (en) | Intelligent control method, system and medium for automatic optical detection of appearance defects | |
| CN113838054B (en) | Mechanical part surface damage detection method based on artificial intelligence | |
| CN114266944B (en) | Rapid model training result checking system | |
| CN111311576A (en) | Point cloud information-based defect detection method | |
| CN114066848A (en) | FPCA appearance defect visual inspection system | |
| CN110935646A (en) | Full-automatic crab grading system based on image recognition | |
| CN117152119A (en) | Profile flaw visual detection method based on image processing | |
| CN117152157B (en) | Electronic element identification method based on artificial intelligence | |
| CN117372424B (en) | Defect detection method, device, equipment and storage medium | |
| WO2019176988A1 (en) | Inspection system, identification system, and device for evaluating identification apparatus | |
| CN114581415A (en) | Method and device for detecting defects of PCB, computer equipment and storage medium | |
| CN114578015B (en) | Intelligent quality detection method for soft magnetic ferrite | |
| CN116618878A (en) | Pre-welding process parameter determination method, welding quality online prediction method, device and storage medium | |
| TWI801973B (en) | Device and method for estimating error factors | |
| CN115200513A (en) | Coaxiality jumping detection, analysis and control system for rotating body clamp | |
| CN114819768A (en) | Condenser lead-out wire welding control system | |
| CN113313704B (en) | Integrated circuit pin detection method, device, terminal and storage medium | |
| CN112070847A (en) | Wood floor color sorting method and device | |
| CN113808129A (en) | Effective detection method for confusable defects | |
| JPH1164235A (en) | Automatic packaging parts inspection system and device and method for setting packaging parts inspecting standard | |
| CN117299596B (en) | Material screening system and method for automatic detection | |
| CN117523324B (en) | Image processing method and image sample classification method, device and storage medium |
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 |