Product detection method, system and computer medium based on intelligent recognition of design drawings

Abstract

The present invention discloses a product detection method, system and computer medium based on intelligent identification of design drawings, the method comprising: establishing a silk screen data feature library, the silk screen data feature library comprising silk screen features of product design drawings; judging whether there is a silk screen design error in the product design drawings of the product to be detected according to the silk screen data feature library; obtaining an image of the product to be detected, comparing the image of the product to be detected with the same type of silk screen features in the product design drawings of the product to be detected without silk screen design errors, if the comparison is consistent, the product to be detected passes the test, if the comparison is inconsistent, the product to be detected fails the test. The present invention can improve the efficiency of R&D design and production of ordinary manufacturing enterprises by a more comprehensive, lower-cost and more efficient inspection method, bringing greater development space for enterprises.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence recognition and detection technology, and more specifically, to a product detection method, system and computer medium based on intelligent recognition of design drawings. Background Art

[0002] At present, the existing image recognition technology has been widely used in various industries, such as automobile license plate recognition, commodity shopping recognition, face recognition, object detection, etc. However, in manufacturing enterprises, image recognition technology is rarely used in various stages of product R&D design, product manufacturing, and product shipment inspection. Therefore, it cannot bring high-efficiency R&D and production products to enterprises, and thus cannot create greater benefits for enterprises.

[0003] In the R&D and manufacturing of enterprises, most of them still rely on manual inspection and judgment of product design drawings and sample verification, which often leads to human negligence or misjudgment. Once the errors in the R&D stage flow to the production and manufacturing end or even more seriously appear on the customer end, it will cause immeasurable losses to the enterprise. Therefore, some enterprises will try to find machines to replace manual inspection to prevent such errors from happening. However, some current inspection tools have some problems, such as uneven inspection categories, expensive inspection tools, and inspection methods that cannot meet requirements. Summary of the invention

[0004] One of the purposes of the present invention is to provide a product inspection method based on intelligent recognition of design drawings, so as to solve the technical problems in the prior art of manual inspection and judgment of product design drawings and sample verification, such as human negligence or misjudgment; the second purpose of the present invention is to provide a product inspection system based on intelligent recognition of design drawings; the third purpose of the present invention is to provide a computer medium.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] A first aspect of the present invention provides a product detection method based on intelligent recognition of design drawings, comprising the following steps:

[0007] Establishing a silk screen data feature library, wherein the silk screen data feature library includes silk screen features of product design drawings;

[0008] According to the silk screen data feature library, determining whether there is a silk screen design error in the product design drawing of the product to be tested;

[0009] Obtain an image of the product to be inspected, and compare the image of the product to be inspected with the same type of silk screen features in the product design drawing of the product to be inspected without silk screen design errors. If the comparison is consistent, the product to be inspected passes the inspection; if the comparison is inconsistent, the product to be inspected fails the inspection.

[0010] In the above-mentioned technical means, a silk screen feature database of the product is first established. When it is necessary to compare and judge the product design drawings with the actual products, the design drawings are first uploaded, the silk screen features of the drawings that need to be inspected are selected, and the silk screen features of the drawings are judged according to the silk screen feature database; then images of the products that need to be compared and tested are collected, and then the images are uploaded. The silk screen features of the data in the images are quickly extracted according to the intelligent image recognition method, and then these silk screen features are compared and judged with the silk screen features extracted from the original drawings. If any inconsistencies are found, they are marked to remind the designers. The designers confirm the defects of the actual product based on these marks, and then optimize the design of the actual product, so as to solve the technical problems in the prior art of manual inspection and judgment of product design drawings and sample verification, such as human negligence or misjudgment.

[0011] Furthermore, the establishment of the silk screen data feature library includes the following steps:

[0012] Obtain product design drawings;

[0013] Extracting silk screen features of the product design drawing;

[0014] The extracted silk-screen features form the silk-screen data feature library.

[0015] Furthermore, after the extracted silk screen features are used to form the silk screen data feature library, the following steps are also included:

[0016] Performing data cleaning on the silk screen data feature library;

[0017] The silk screen printing features are extracted and classified from the silk screen printing data feature library after data cleaning.

[0018] Furthermore, the data cleaning includes missing data processing, duplicate data processing, abnormal data processing and inconsistent data sorting.

[0019] Furthermore, the missing data processing, duplicate data processing, abnormal data processing and inconsistent data sorting are repeated several times.

[0020] Further, judging whether there is a silk-screen design error in the product design drawing of the product to be tested according to the silk-screen data feature library comprises the following steps:

[0021] Extracting the silk screen features of the product design drawing of the product to be tested to obtain the silk screen features to be compared;

[0022] Select a silk screen feature of the same type as the silk screen feature to be compared from the silk screen data feature library, and compare it with the silk screen feature to be compared. If the comparison is consistent, it is judged that there is no silk screen design error in the product design drawing of the product to be tested; if the comparison is inconsistent, it is judged that there is a silk screen design error in the product design drawing of the product to be tested.

[0023] Furthermore, if it is determined that there are silk-screen design errors in the product design drawings of the product to be tested, a mark is made and a reminder is issued.

[0024] Furthermore, after extracting the silk screen features of the product design drawings of the product to be tested, silk screen feature screening is also performed, including:

[0025] Question (1):

[0026]

[0027] In the formula, ψ(.) is the penalty term, and the one used is L 2 norm, ψ(w) = ‖w‖ 2 ; w represents the n-dimensional vector of silk screen features; F represents the set of silk screen features; p λ (.) represents a training set of independent and identically distributed silk screen features; x i is an input vector of silk screen features; l(.) is the loss function, using the hinge loss function, l(x i ; w) = max{0,1-y i f(x i )},y i represents the class corresponding to the input vector of each silk-screen feature, f(.) represents the classification function of binary imbalanced data; λ>0 is the regularization parameter;

[0028] Solve the dual problem of problem (1):

[0029]

[0030] st0≤α i ≤λ,1≤i≤n

[0031] Where α represents an n-dimensional vector of the Lagrange multiplier method; D λ (.) represents a set of independent and identically distributed training sets; Construct a primitive space for the kernel function defined by the silkscreen feature map Φ The original space includes the optimal solution w * , calculate the lower interval of the interval:

[0032]

[0033] Upper range:

[0034]

[0035] Contains the optimal solution w * The spherical region of is defined as follows:

[0036] For any have:

[0037]

[0038] in, domP λ and domD λ Respectively represent P λ and D λ The domain of definition is given by the feasible solution and The duality gap is defined, and for a given spherical region, the lower interval of the region is defined as follows:

[0039]

[0040] The upper interval is defined as follows:

[0041]

[0042] Solve to get the optimal solution w * .

[0043] A second aspect of the present invention provides a product detection system based on intelligent recognition of design drawings, comprising:

[0044] A database establishment module, wherein the database establishment module establishes a silk screen data feature library, wherein the silk screen data feature library includes silk screen features of product design drawings;

[0045] A drawing judgment module, wherein the drawing judgment module judges whether there is a silk-screen design error in the product design drawing of the product to be tested according to the silk-screen data feature library;

[0046] A product judgment module obtains an image of the product to be tested, and compares the image of the product to be tested with the same type of silk screen features in the product design drawings of the product to be tested without silk screen design errors. If the comparison is consistent, the product to be tested passes the inspection; if the comparison is inconsistent, the product to be tested fails the inspection.

[0047] A third aspect of the present invention provides a computer medium having a computer program stored thereon, and when the computer program is executed by a processor, the product detection method based on intelligent identification of design drawings is implemented.

[0048] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0049] The present invention establishes a silk screen feature database for products, processes the database to obtain silk screen feature classification, collects images of products that need to be compared and tested, uploads the images, and quickly extracts silk screen feature data in the images based on an intelligent image recognition method. These silk screen features are then compared and judged with the silk screen features extracted from the original drawings. If any inconsistencies are found, they are marked to remind designers. This method can improve the R&D, design and production efficiency of ordinary manufacturing companies through a more comprehensive, lower-cost and more efficient inspection method, thereby bringing greater development space for the companies. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A schematic flow chart of a product detection method based on intelligent identification of design drawings provided in an embodiment of the present invention.

[0051] Figure 2 A schematic diagram of the silk screen feature comparison process of a product image to be inspected provided in an embodiment of the present invention.

[0052] Figure 3 A schematic flow chart of a product inspection system based on intelligent recognition of design drawings provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0054] In order to better illustrate the present embodiment, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product;

[0055] It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0056] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0057] Example 1

[0058] This embodiment provides a product detection method based on intelligent recognition of design drawings, such as Figure 1 and Figure 2 As shown, the following steps are included:

[0059] Establishing a silk screen data feature library, wherein the silk screen data feature library includes silk screen features of product design drawings;

[0060] According to the silk screen data feature library, determining whether there is a silk screen design error in the product design drawing of the product to be tested;

[0061] Obtain an image of the product to be inspected, and compare the image of the product to be inspected with the same type of silk screen features in the product design drawing of the product to be inspected without silk screen design errors. If the comparison is consistent, the product to be inspected passes the inspection; if the comparison is inconsistent, the product to be inspected fails the inspection.

[0062] In this embodiment, a silk screen data feature library is first established, which includes silk screen features of product design drawings that have been verified to be correct. The silk screen data feature library is used to determine whether the product design drawings of the product to be tested have silk screen design errors, thereby ensuring that when comparing the products to be tested, there will be no misjudgment due to silk screen design errors in the product design drawings; after ensuring that there are no silk screen design errors in the product design drawings of the product to be tested, the silk screen features in the product design drawings of the product to be tested and the silk screen features of the image of the product to be tested are respectively extracted, and the two are compared and judged. If they are consistent, the product to be tested meets the requirements of the product design drawings and is a qualified product; if they are inconsistent, the product to be tested does not meet the requirements of the product design drawings and is an unqualified product, thereby achieving a more comprehensive, lower-cost, and more efficient inspection method to improve the R&D design and production efficiency of ordinary manufacturing enterprises, thereby bringing greater development space for enterprises.

[0063] In a further embodiment, the establishment of the silk screen data feature library comprises the following steps:

[0064] Obtain product design drawings;

[0065] Extracting silk screen features of the product design drawing;

[0066] The extracted silk-screen features form the silk-screen data feature library.

[0067] In this embodiment, the silk screen feature database is collected by designers from a large number of product design drawings. These product design drawings have been verified in the early stage to ensure the correctness of the source data, thereby ensuring the correctness of the silk screen feature database and classifying the extracted silk screen features.

[0068] In this embodiment, silk screen printing is a printing technology, and by identifying silk screen features in an image, such as a specific pattern, text, or logo, it can be determined whether silk screen printing exists on a product.

[0069] In a further embodiment, after the extracted silk screen features are formed into the silk screen data feature library, the following steps are also included:

[0070] Performing data cleaning on the silk screen data feature library;

[0071] The silk screen printing features are extracted and classified from the silk screen printing data feature library after data cleaning.

[0072] In this embodiment, the main tasks of data cleaning are format standardization, abnormal data removal, error correction and duplicate data removal, so as to provide well-structured and more suitable data for the next step of data mining.

[0073] In a further embodiment, the data cleaning includes missing data processing, duplicate data processing, abnormal data processing and inconsistent data sorting.

[0074] In a further embodiment, the missing data processing, duplicate data processing, abnormal data processing and inconsistent data sorting are repeated several times.

[0075] In this embodiment, by cleaning the data, it is possible to effectively prevent data entry errors from affecting subsequent data processing, and by reciprocating processing, it is possible to prevent data residue from affecting data processing, thereby further improving the effect of subsequent data processing.

[0076] In a further embodiment, judging whether there is a silk-screen design error in the product design drawing of the product to be inspected according to the silk-screen data feature library comprises the following steps:

[0077] Extracting the silk screen features of the product design drawing of the product to be tested to obtain the silk screen features to be compared;

[0078] Select a silk screen feature of the same type as the silk screen feature to be compared from the silk screen data feature library, and compare it with the silk screen feature to be compared. If the comparison is consistent, it is judged that there is no silk screen design error in the product design drawing of the product to be tested; if the comparison is inconsistent, it is judged that there is a silk screen design error in the product design drawing of the product to be tested.

[0079] In a further embodiment, if it is determined that there are silk-screen design errors in the product design drawings of the product to be inspected, a mark is made and a reminder is issued.

[0080] In this embodiment, when the product design drawings of the product to be tested are compared and a silk screen design error is found, the erroneous silk screen design needs to be marked and a reminder is issued to the staff; at the same time, if inconsistent silk screen features are found in the product to be tested during the comparison, the inconsistent silk screen features are also marked and a reminder is issued to the staff, thereby improving the effect of product defect detection.

[0081] In a further embodiment, after extracting the silk screen features of the product design drawing of the product to be tested, silk screen feature screening is also performed, including:

[0082] Question (1):

[0083]

[0084] In the formula, ψ(.) is the penalty term, and the one used is L 2 norm, ψ(w) = ‖w‖ 2 ; w represents the n-dimensional vector of silk screen features; F represents the set of silk screen features; p λ (.) represents a training set of independent and identically distributed silk screen features; x i is an input vector of silk screen features; l(.) is the loss function, using the hinge loss function, l(x i ; w) = max{0,1-y i f(x i )},y i represents the class corresponding to the input vector of each silk-screen feature, f(.) represents the classification function of binary imbalanced data; λ>0 is the regularization parameter;

[0085] Solve the dual problem of problem (1):

[0086]

[0087] st0≤α i ≤λ,1≤i≤n

[0088] Where α represents an n-dimensional vector of the Lagrange multiplier method; D λ (.) represents a set of independent and identically distributed training sets. The value on the left is the maximum optimal set of sample data. α i is the Lagrange multiplier; Construct a primitive space for the kernel function defined by the silkscreen feature map Φ The silk screen feature map Φ refers to the process of collecting defects by comparing the input design drawings (the feature extraction algorithm and the feature extraction goal are to extract useful features from the image to facilitate subsequent image processing and analysis); the original space includes the optimal solution w* , calculate the lower interval of the interval:

[0089]

[0090] Upper range:

[0091]

[0092] Contains the optimal solution w * The spherical region of is defined as follows:

[0093] For any have:

[0094]

[0095] in, domP λ and domD λ Respectively represent P λ and D λ The domain of definition is given by the feasible solution and The duality gap is defined, and for a given spherical region, the lower interval of the region is defined as follows:

[0096]

[0097] The upper interval is defined as follows:

[0098]

[0099] Solve to get the optimal solution w * .

[0100] In this embodiment, the safety sample screening rules are obtained through the upper and lower intervals, and the data in the data set is screened according to the algorithm screening model, and the worthless and unimportant data are removed from the data set, thereby realizing data set screening and then realizing accurate extraction of the data set, avoiding invalid data from affecting the effect of subsequent detection.

[0101] Example 2

[0102] This embodiment provides a product detection system based on intelligent recognition of design drawings on the basis of embodiment 1. Figure 3 As shown, including:

[0103] A database establishment module, wherein the database establishment module establishes a silk screen data feature library, wherein the silk screen data feature library includes silk screen features of product design drawings;

[0104] A drawing judgment module, wherein the drawing judgment module judges whether there is a silk-screen design error in the product design drawing of the product to be tested according to the silk-screen data feature library;

[0105] A product judgment module obtains an image of the product to be tested, and compares the image of the product to be tested with the same type of silk screen features in the product design drawings of the product to be tested without silk screen design errors. If the comparison is consistent, the product to be tested passes the inspection; if the comparison is inconsistent, the product to be tested fails the inspection.

[0106] In a further embodiment, a warning module is also included. When the product design drawings of the product to be tested are compared and a silk screen design error is found, the erroneous silk screen design needs to be marked and a reminder is issued to the staff; at the same time, if inconsistent silk screen features are found in the product to be tested during the comparison, the inconsistent silk screen features are also marked and a reminder is issued to the staff.

[0107] Example 3

[0108] This embodiment provides a computer medium having a computer program stored thereon. When the computer program is executed by a processor, the product detection method based on intelligent identification of design drawings as described in Embodiment 1 is implemented.

[0109] The same or similar reference numerals correspond to the same or similar components;

[0110] The terms used in the drawings to describe positional relationships are only used for illustrative purposes and should not be construed as limiting this patent;

[0111] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A product detection method based on intelligent recognition of design drawings, characterized in that: The following steps are involved: Establishing a silk screen data feature library, wherein the silk screen data feature library includes silk screen features of product design drawings; According to the silk screen data feature library, determining whether there is a silk screen design error in the product design drawing of the product to be tested; Obtain an image of the product to be tested, and compare the image of the product to be tested with the same type of silk screen features in the product design drawing of the product to be tested without silk screen design errors. If the comparison is consistent, the product to be tested passes the test; if the comparison is inconsistent, the product to be tested fails the test; The step of judging whether there is a silk screen design error in the product design drawing of the product to be tested according to the silk screen data feature library comprises the following steps: Extracting the silk screen features of the product design drawing of the product to be tested to obtain the silk screen features to be compared; After extracting the silk screen features of the product design drawings of the product to be tested, silk screen feature screening is also performed, including: Question (1): In the formula, ψ(.) is the penalty term, using the L2 norm, ψ(w) = ‖w‖ 2 ; w represents the n-dimensional vector of silk screen features; F represents the set of silk screen features; P λ (.) represents a training set of independent and identically distributed silk screen features; x i is an input vector of silk-screen features; As the loss function, the hinge loss function is used. y i represents the class corresponding to the input vector of each silk-screen feature, f(.) represents the classification function of binary imbalanced data; λ>0 is the regularization parameter; Solve the dual problem of problem (1): s.t.0≤α i ≤λ,1≤i≤n Where α represents an n-dimensional vector of the Lagrange multiplier method; D λ (.) represents a set of independent and identically distributed training sets; Construct a primitive space for the kernel function defined by the silkscreen feature map Φ The original space includes the optimal solution w * , calculate the lower interval of the interval: Upper range: Contains the optimal solution w * The spherical region of is defined as follows: For any have: in, domP λ and domD λ Respectively represent P λ and D λ The domain of definition is given by the feasible solution and The duality gap is defined, and for a given spherical region, the lower interval of the region is defined as follows: The upper interval is defined as follows: Solve to get the optimal solution w * .

2. The product detection method based on intelligent identification of design drawings according to claim 1 is characterized in that: The establishment of the silk screen data feature library comprises the following steps: Obtain product design drawings; Extracting silk screen features of the product design drawing; The extracted silk-screen features form the silk-screen data feature library.

3. The product detection method based on intelligent identification of design drawings according to claim 2 is characterized in that: After the extracted silk screen features are used to form the silk screen data feature library, the following steps are also included: Performing data cleaning on the silk screen data feature library; The silk screen printing features are extracted and classified from the silk screen printing data feature library after data cleaning.

4. The product detection method based on intelligent identification of design drawings according to claim 3 is characterized in that: The data cleaning includes missing data processing, duplicate data processing, abnormal data processing and inconsistent data sorting.

5. The product detection method based on intelligent identification of design drawings according to claim 4 is characterized in that: The missing data processing, duplicate data processing, abnormal data processing and inconsistent data sorting are repeated several times.

6. The product detection method based on intelligent identification of design drawings according to claim 1 is characterized in that: The step of judging whether there is a silk screen design error in the product design drawing of the product to be tested according to the silk screen data feature library comprises the following steps: Extracting the silk screen features of the product design drawing of the product to be tested to obtain the silk screen features to be compared; Select a silk screen feature of the same type as the silk screen feature to be compared from the silk screen data feature library, and compare it with the silk screen feature to be compared. If the comparison is consistent, it is judged that there is no silk screen design error in the product design drawing of the product to be tested; if the comparison is inconsistent, it is judged that there is a silk screen design error in the product design drawing of the product to be tested.

7. The product detection method based on intelligent identification of design drawings according to claim 1 is characterized in that: If it is determined that there are silk-screen design errors in the product design drawings of the product to be tested, a mark is made and a reminder is issued.

8. A product detection system based on intelligent recognition of design drawings, characterized in that: The system applies the product detection method for intelligent identification of design drawings according to any one of claims 1 to 7, comprising: A database establishment module, wherein the database establishment module establishes a silk screen data feature library, wherein the silk screen data feature library includes silk screen features of product design drawings; A drawing judgment module, wherein the drawing judgment module judges whether there is a silk-screen design error in the product design drawing of the product to be tested according to the silk-screen data feature library; A product judgment module obtains an image of the product to be tested, and compares the image of the product to be tested with the same type of silk screen features in the product design drawings of the product to be tested without silk screen design errors. If the comparison is consistent, the product to be tested passes the inspection; if the comparison is inconsistent, the product to be tested fails the inspection.

9. A computer medium, characterized in that The computer medium stores a computer program, and when the computer program is executed by the processor, the product detection method based on intelligent identification of design drawings as described in any one of claims 1 to 7 is implemented.

Images