JP2019056591A - Visual inspection device and visual inspection method - Google Patents

Abstract

To provide a visual inspection device capable of easily, accurately, and visually inspecting an inspection object.SOLUTION: A visual inspection device 1 includes: a reference data acquisition unit 10 for acquiring reference data related to visual inspection of an inspection object M from a plurality of features in a plurality of reference images P related to the inspection object M; an image acquisition unit 20 for acquiring the image of the inspection object M; and a determination unit 30 that compares the reference data acquired by the reference data acquisition unit 10 with the image of the inspection object M acquired by the image acquisition unit 20, and determines non-defectives and defectives of the inspection object M on the basis of the matching rate acquired from the comparison result.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an appearance inspection apparatus and an appearance inspection method.

  2. Description of the Related Art An appearance inspection apparatus that performs an appearance inspection of an inspection object is known. As such an appearance inspection device, for example, Patent Document 1 discloses a mounting component inspection device that selects component inspection data that matches the type of component to be inspected and performs an appearance inspection based on the component inspection data. .

  The mounting component inspection apparatus stores in advance a component inspection reference data corresponding to the type of the component in the storage unit when performing the appearance inspection of a plurality of types of components with different appearances mounted on the same position on the substrate. When performing the component inspection, the type of the inspection target component is determined, and the component inspection reference data corresponding to the component is selected from the data stored in the storage unit to perform the inspection.

JP 2002-299900 A

  The appearance inspection apparatus disclosed in the above-mentioned Patent Document 1 can perform component inspection according to the determination result of the type of component to be inspected. However, the appearance inspection apparatus of Patent Document 1 needs to create reference data for component inspection corresponding to the type of component to be inspected, and store the reference data for component inspection in a storage unit in advance.

  As described above, when the part inspection reference data is set in advance, the inspection person needs to determine the reference for each part, and the inspection person needs to input the changed reference even when changing the reference. . In addition, when inspecting the inspection object, if the position of the inspection object is different from the reference position at the time of inspection, data for correcting the component inspection reference data is required.

  Therefore, in the inspection using the appearance inspection apparatus as described above, the work burden on the inspection person is heavy.

  An object of the present invention is to provide an appearance inspection apparatus capable of easily and accurately inspecting an inspection object.

  An appearance inspection apparatus according to an embodiment of the present invention is an appearance inspection apparatus that performs an appearance inspection of an inspection object using an image of the inspection object. The appearance inspection apparatus includes a reference data acquisition unit that acquires reference data related to an appearance inspection of the inspection object from a plurality of features in a plurality of reference images related to the inspection object, and an image acquisition that acquires an image of the inspection object And the reference data obtained by the reference data acquisition unit and the image of the inspection object obtained by the image acquisition unit, and the inspection object based on the matching rate obtained from the comparison result A determination unit that determines whether the product is non-defective or defective.

  According to the appearance inspection apparatus according to an embodiment of the present invention, it is possible to easily and accurately inspect the inspection object.

FIG. 1 is a functional block diagram illustrating a schematic configuration of the appearance inspection apparatus according to the embodiment. FIG. 2 is a diagram illustrating an example of a reference image of a non-defective product and a defective product as an inspection target. FIG. 3 is a flowchart showing an operation of generating reference data of the appearance inspection apparatus. FIG. 4 is a flowchart showing operations of type determination and appearance inspection by the determination unit.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, the same code | symbol is attached | subjected about the same or an equivalent part in a figure, and the description is not repeated. Moreover, the dimension of the structural member in each figure does not faithfully represent the actual dimension of the structural member, the dimensional ratio of each structural member, or the like.

(overall structure)
FIG. 1 is a functional block diagram showing a schematic configuration of an appearance inspection apparatus 1 according to an embodiment of the present invention. The appearance inspection apparatus 1 performs an appearance inspection of the inspection object M using an image of the inspection object M. That is, the appearance inspection apparatus 1 compares the reference data obtained from the plurality of reference images P of the inspection object M with the image of the inspection object M, and the comparison result is the inspection result of the appearance inspection of the inspection object M. Output as.

  Specifically, the appearance inspection apparatus 1 includes a reference data acquisition unit 10, an image acquisition unit 20, a determination unit 30, a storage unit 40, and an output unit 50.

  In the appearance inspection apparatus 1, the reference data acquisition unit 10, the determination unit 30, and the storage unit 40 are configured by a computer device. The reference data acquisition unit 10, the determination unit 30, and the storage unit 40 may be configured as a server that can communicate with the image acquisition unit 20 and the output unit 50 via a communication line. The storage unit 40 may be configured by a storage device different from the computer device or the server. The reference data acquisition unit 10 and the determination unit 30 are preferably constructed by artificial intelligence (AI).

  The reference data acquisition unit 10 acquires reference data used for appearance inspection of the inspection object M using a plurality of reference images P of the inspection object M. The reference image P is an image of the appearance of the inspection object M, and includes images of non-defective products and defective products in a plurality of types of inspection objects M. Further, the reference image P includes an image obtained in advance and an image of the inspection object M that has been visually inspected by the determination unit 30 described later. That is, as for the reference image P, when the appearance inspection apparatus 1 performs the appearance inspection of the inspection object M, the number of target images increases. Note that the reference image P may include only an image obtained in advance.

  In the present embodiment, the defective product of the inspection object M is an inspection object having damage such as scratches or dents on the outer surface. On the other hand, a non-defective product of the inspection object M is an inspection object in which damage such as a scratch or a dent is not detected on the outer surface by the appearance inspection apparatus 1.

  Specifically, the reference data acquisition unit 10 includes a reference image acquisition unit 11, a feature extraction unit 12, and a reference data generation unit 13.

  The reference image acquisition unit 11 acquires a plurality of reference images P of the inspection object M stored in the storage unit 40. That is, the reference image acquisition unit 11 acquires a plurality of non-defective reference images for a plurality of types of inspection objects M and a plurality of defective reference images for a plurality of types of inspection objects M. As will be described later, the reference image P stored in the storage unit 40 includes an image of reference image data and an image of an appearance inspection result by the determination unit 30 described later. An example of the reference images P1 to P8 of the non-defective product and the defective product of the inspection object M acquired by the reference image acquisition unit 11 is shown in FIG.

  In the example of FIG. 2, the inspection object M is a metal part of the same type having a circular shape in plan view. In FIG. 2, reference signs P1 to P8 are reference images, reference signs M1 to M8 are inspection objects, and reference signs X1 to X4 are scratches on the outer surfaces of the inspection objects M5 to M8. The inspection object M may have a shape other than a circle, or may be other than a metal part such as a resin product.

  The feature extraction unit 12 extracts features from the plurality of reference images P acquired by the reference image acquisition unit 11. Specifically, the feature extraction unit 12 extracts features to be used for determining the type of the inspection object M and determining non-defective and defective products of the inspection object M from the plurality of reference images P. The feature extraction unit 12 extracts, for example, the shape of the inspection object M, the color (hue, brightness, saturation), the color arrangement (pattern), and the like as the characteristics.

  The reference data generation unit 13 generates reference data including feature amounts obtained by weighting the features extracted from the plurality of reference images P by the feature extraction unit 12. The reference data includes type discrimination reference data for discriminating the type of the inspection object M, and determination reference data for determining the non-defective product and the defective product of the inspection object M. That is, the reference data generation unit 13 generates the type determination reference data and the determination reference data from the features extracted by the feature extraction unit 12.

  Specifically, the reference data generation unit 13, for example, among the features extracted by the feature extraction unit 12, the shape of the inspection object M, the color (hue, brightness, saturation), and color of the inspection object M The feature amount is generated from the arrangement (pattern) or the like. The reference data generation unit 13 increases the weight of the shape of the inspection object M in the feature amount when generating the type determination reference data.

  In addition, when generating the determination reference data, the reference data generation unit 13 increases the weight of the feature relating to, for example, a color that frequently appears as a defective product and its arrangement in the feature amount, while The weight of the feature related to the overall color of the object M is reduced.

  That is, the reference data generation unit 13 analyzes the tendency of the features obtained from the plurality of reference images P for the non-defective product and the defective product, respectively, and reflects the tendency to be determined as the good product and the defective product in the feature amount. Then, the determination reference data matching the tendency is generated. The determination reference data includes feature amounts obtained based on features extracted from the plurality of reference images P of non-defective products and defective products. For example, when there are many abnormalities on the outer peripheral side of the inspection object M in defective products, the reference data generating unit 13 generates reference data for determination with a high priority of the features on the outer peripheral side. The reference data generation unit 13 generates determination reference data for each type of inspection object M.

  As a result, it is possible to inspect the appearance of the inspection object M more accurately and easily using the plurality of reference images P.

  Note that the feature amount obtained based on the features extracted from the plurality of reference images has a predetermined distribution. Therefore, the type determination reference data and the determination reference data are also data having distributions.

  As described above, the reference data includes an image obtained in advance and an image of the inspection object M that has been visually inspected by the appearance inspection apparatus 1 as described later. Therefore, the reference data generation unit 13 generates appropriate reference data according to the result of the appearance inspection of the inspection object M. Therefore, the accuracy of the reference data can be improved according to the result of the appearance inspection of the inspection object M. That is, the appearance inspection apparatus 1 has a learning function for learning the reference data.

  The image acquisition unit 20 acquires an image of the inspection object M. The image acquisition unit 20 is, for example, an imaging device that can capture an image such as a camera and acquire it as data. The image acquisition unit 20 outputs the acquired image data of the inspection object M to the storage unit 40. The storage unit 40 stores the image data.

  The determination unit 30 reads the image data of the inspection object M from the storage unit 40 and compares the image data with the reference data generated by the reference data generation unit 13. The determination unit 30 determines the type of the inspection object M from the comparison result between the image data and the reference data, and performs an appearance inspection according to the type.

  Specifically, the determination unit 30 includes a type determination unit 31 and an appearance inspection unit 32.

  The type discriminating unit 31 extracts features from the image data, compares the features of the image data with the type discriminating reference data, and obtains their matching rate. Specifically, the type determination unit 31 determines how much the feature of the image data deviates from the feature amount in each type of inspection object. The type discriminating unit 31 discriminates the type of the inspection object M according to the degree of deviation from the feature amount. That is, the type discriminating unit 31 determines that the inspection object M is the predetermined type when the degree of deviation from the feature amount is within the range of the type discrimination reference data distribution in the predetermined type inspection object. Is determined.

  The type discrimination unit 31 outputs the discrimination result of the type of the inspection object M to the appearance inspection unit 32.

  The appearance inspection unit 32 selects the determination reference data according to the determination result of the type of the inspection object M by the type determination unit 31 and extracts features from the image data. The appearance inspection unit 32 compares the selected reference data for determination with the features extracted from the image data, and obtains the matching rate thereof. Specifically, the appearance inspection unit 32 determines how much the feature of the image data deviates from the feature amount of the selected determination reference data. The appearance inspection unit 32 determines whether the inspection object M is a non-defective product or a defective product depending on whether the deviation from the feature amount is within a predetermined range. That is, the appearance inspection unit 32 determines that the inspection object M is a non-defective product if the deviation from the feature amount is within a predetermined range, and determines the inspection object M if the deviation is outside the predetermined range. Judged as defective. The appearance inspection unit 32 outputs the result of the appearance inspection of the inspection object M to the output unit 50. The predetermined range is set to a range in which the inspection object M is determined as a non-defective product.

  Thereby, the non-defective product and the defective product of the inspection object M can be determined more accurately. In addition, when the characteristics of the inspection object are within a predetermined range with respect to the feature amount, the non-defective product and the defective product of the inspection object M are determined, thereby changing the conditions during the appearance inspection, and Even when the inspection object M varies within the range of non-defective products, the appearance inspection of the inspection object M can be performed with high accuracy.

  As described above, the appearance inspection unit 32 determines a non-defective product and a defective product of the inspection object M based on the matching rate between the determination reference data and the features extracted from the image data. As a result, the whole image of the inspection object M can be compared with the reference data for determination, and a good product and a defective product can be determined. Therefore, it is possible to inspect the appearance of the non-defective product and the defective product of the inspection object M with high accuracy.

  The storage unit 40 stores reference image data that is a part of a plurality of reference images P, the reference data generated by the reference data acquisition unit 10, and the inspection object M acquired by the image acquisition unit 20. The image data and the appearance inspection result by the determination unit 30 are stored. The storage unit 40 is a storage medium such as a memory or a hard disk.

  The reference data acquisition unit 10 and the determination unit 30 are connected to the storage unit 40 so as to be able to exchange data. The image acquisition unit 20 is connected to the storage unit 40 so that data can be transmitted. The reference image data and the appearance inspection result stored in the storage unit 40 are used when the reference data acquisition unit 10 generates reference data as a plurality of reference images P. The reference data and the image data stored in the storage unit 40 are used when the determination unit 30 performs an appearance inspection of the inspection object M.

  The output unit 50 outputs the appearance inspection result of the inspection object M in the determination unit 30. The output unit 50 outputs the appearance inspection result as an image and sound to notify the operator. The output unit 50 includes, for example, a display device that displays the appearance inspection result, an output device that outputs the appearance inspection result with sound, and the like. The output unit 50 may output the appearance inspection result to another device.

(Visual inspection)
Next, a method of appearance inspection by the appearance inspection apparatus 1 having the above-described configuration, that is, an operation of the appearance inspection apparatus 1 will be described with reference to FIGS. The appearance inspection apparatus 1 generates reference data for appearance inspection, and performs an appearance inspection of the inspection object M using the reference data. FIG. 3 is a flowchart showing the operation of generating reference data by the reference data generating unit 10. FIG. 4 is a flowchart showing operations of type determination and appearance inspection by the determination unit 30.

  The operation of generating reference data by the appearance inspection apparatus 1 will be described with reference to FIG.

  When the flow shown in FIG. 3 starts (START), first, the reference image acquisition unit 11 of the reference data generation unit 10 captures the reference image (step SA1). In step SA1, the reference image acquisition unit 11 acquires a plurality of reference images of good and defective products for each type of inspection object M. The reference image includes an image obtained in advance and an image of the inspection object M that has been visually inspected by the appearance inspection apparatus 1 as will be described later.

  Next, the feature extraction unit 12 uses features for use in determining the type of the inspection object M and determining the non-defective product and the defective product of the inspection object M from the plurality of reference images acquired by the reference image acquisition unit 11. Extract (step SA2). The features include, for example, the shape of the inspection object M, the color (hue, brightness, saturation) of the inspection object M, and the color arrangement (pattern).

  Then, the reference data generation unit 13 generates reference data including feature amounts obtained by weighting the features extracted by the feature extraction unit 12 (step SA3). The reference data includes type determination reference data for determining the type of the inspection object M, and determination reference data for determining non-defective products and defective products of the inspection object M.

  The reference data generation unit 13, for example, among the features extracted by the feature extraction unit 12, the shape of the inspection object M, the color (hue, brightness, saturation) of the inspection object M, the color arrangement (pattern), etc. Then, the feature amount is generated. In addition, the reference data generation unit 13 analyzes the tendency of the feature extracted by the feature extraction unit 12 for each of the non-defective product and the defective product, and reflects the tendency determined as the good product and the defective product in the feature amount. The reference data for determination matching the tendency is generated. The determination reference data includes feature amounts obtained based on features extracted from the plurality of reference images P of non-defective products and defective products. The reference data generation unit 13 generates determination reference data for each type of inspection object M.

  Note that the feature amount obtained based on the features extracted from the plurality of reference images has a predetermined distribution. Therefore, the type determination reference data and the determination reference data are also data having distributions.

  The reference data generated by the reference data generation unit 13 is stored in the storage unit 40. Thereafter, this flow ends (END).

  Next, operations of type discrimination and appearance inspection by the appearance inspection apparatus 1 will be described with reference to FIG.

  When the flow shown in FIG. 4 starts (START), first, the image acquisition unit 30 acquires image data of the inspection object M (step SB1). The image data acquired by the image acquisition unit 30 is stored in the storage unit 40.

  Next, the type determination unit 31 of the determination unit 30 reads the type determination reference data and the image data stored in the storage unit 40, extracts the characteristics of the image data, and extracts the characteristics of the image data and the types of the image data. Compare with the reference data for discrimination, and find their matching rate. That is, the type determination unit 31 determines how much the feature of the image data deviates from the feature amount of the type determination reference data for each type of inspection object M. The type discrimination unit 31 outputs the discrimination result of the type of the inspection object M to the appearance inspection unit 32.

  The appearance inspection unit 32 selects determination reference data corresponding to the type of the inspection object M from the determination reference data stored in the storage unit 40 using the type determination result by the type determination unit 31. . The appearance inspection unit 32 reads the image data stored in the storage unit 40, extracts the features of the image data, compares the features of the image data with the selected reference data for determination, The matching rate is obtained. That is, the appearance inspection unit 32 determines how much the feature of the image data deviates from the feature amount of the determination reference data (step SB3). The appearance inspection unit 32 determines that the inspection object M is a non-defective product if the degree of deviation is within a predetermined range (YES in Step SB3) (Step SB4), and if the degree of deviation is outside the predetermined range (Step SB4). In the case of NO at step SB3), the inspection object M is determined to be defective (step SB5).

  The appearance inspection unit 32 outputs the results of the appearance inspection of the inspection object M to the output unit 50 and the storage unit 40, respectively.

  The output unit 50 outputs the inspection result of the appearance inspection unit 32 (step SB6). Thereafter, this flow is ended (END).

  Here, steps SA1 to SA3 correspond to the reference data acquisition process, step SB1 corresponds to the image acquisition process, and steps SB3 to SB5 correspond to the determination process. Step SB2 corresponds to a type determination step.

  As described above, the appearance inspection apparatus 1 obtains reference data used for appearance inspection by weighting features extracted from a plurality of reference images P in a plurality of types of inspection objects M. Specifically, the appearance inspection apparatus 1 acquires data including type determination reference data for determining the type and determination reference data for determining non-defective products and defective products as the reference data. Then, the appearance inspection apparatus 1 determines the type of the inspection object M using the reference data, and performs an appearance inspection of the inspection object M according to the type.

  This eliminates the need to set the reference data in accordance with the type of the inspection object M, and eliminates the need to reset the reference data when changing the inspection standard. Moreover, when the inspection object M is inspected, even if the position of the inspection object M is different from the reference position at the time of inspection, it is not necessary to correct the reference data. Therefore, the labor for setting the reference data can be omitted.

  In addition, by using the reference data obtained by weighting the features extracted from the plurality of reference images P during the appearance inspection of the inspection object M, the appearance inspection of the inspection object M can be performed with high accuracy. Can be done quickly.

  Therefore, according to the configuration of the present embodiment, the appearance inspection of the inspection object M can be performed easily and accurately.

  In addition, in the appearance inspection apparatus 1 of the present embodiment, the reference image P includes an image of the inspection object M that has undergone appearance inspection. That is, the appearance inspection apparatus 1 also uses the result of the appearance inspection when generating the reference data. Thereby, the accuracy of the reference data can be improved according to the result of the appearance inspection. Thereby, the inspection accuracy of the appearance inspection of the inspection object M can be improved without correcting the reference data.

(Other embodiments)
While the embodiments of the present invention have been described above, the above-described embodiments are merely examples for carrying out the present invention. Therefore, the present invention is not limited to the above-described embodiment, and the above-described embodiment can be appropriately modified and implemented without departing from the spirit of the invention.

  In the embodiment, the appearance inspection apparatus 1 determines the type of the inspection object M before the appearance inspection. However, the appearance inspection apparatus may perform the determination of the type of the inspection object and the appearance inspection at the same time. In addition, the appearance inspection apparatus may not determine the type of the inspection object.

  The present invention can be applied to an appearance inspection apparatus that performs an appearance inspection of the inspection object using an image of the inspection object.

DESCRIPTION OF SYMBOLS 1 Appearance inspection apparatus 10 Reference | standard data acquisition part 11 Reference | standard image acquisition part 12 Feature extraction part 13 Reference | standard data generation part 20 Image acquisition part 30 Determination part 31 Type discrimination | determination part 32 Appearance inspection part 40 Memory | storage part 50 Output part M Inspection object

Claims (8)

An appearance inspection apparatus that performs an appearance inspection of the inspection object using an image of the inspection object,
Reference data acquisition unit for acquiring reference data related to appearance inspection of the inspection object from a plurality of features in a plurality of reference images related to the inspection object;
An image acquisition unit for acquiring an image of the inspection object;
The reference data obtained by the reference data acquisition unit is compared with the image of the inspection object obtained by the image acquisition unit, and the non-defective product of the inspection object is based on the matching rate obtained from the comparison result. And a determination unit for determining defective products,
An appearance inspection apparatus comprising: The appearance inspection apparatus according to claim 1,
The visual inspection apparatus, wherein the reference data acquisition unit acquires a feature amount as the reference data from the plurality of features in the plurality of reference images. The appearance inspection apparatus according to claim 2,
The feature amount is a value obtained by weighting the plurality of features in the plurality of reference images,
The determination unit determines a non-defective product and a defective product of the inspection object based on whether the image feature of the inspection object obtained by the image acquisition unit is within a predetermined range with respect to the feature amount. Appearance inspection device. The appearance inspection apparatus according to any one of claims 1 to 3,
The visual inspection apparatus, wherein the plurality of reference images include an image of the inspection object determined by the determination unit. The appearance inspection apparatus according to any one of claims 1 to 4,
An appearance inspection apparatus further comprising a type discriminating unit that discriminates the type of the inspection object using the reference data. An appearance inspection method for performing an appearance inspection of the inspection object using an image of the inspection object,
Reference data acquisition step for acquiring reference data related to appearance inspection of the inspection object from a plurality of features in a plurality of reference images related to the inspection object;
An image acquisition step of acquiring an image of the inspection object;
The reference data obtained by the reference data acquisition unit is compared with the image of the inspection object obtained by the image acquisition unit, and the non-defective product of the inspection object is obtained based on the matching rate obtained from the comparison result. And a determination process for determining defective products,
A visual inspection method. The appearance inspection method according to claim 6,
The visual inspection method, wherein the plurality of reference images include an image of the inspection object determined by the determination unit. In the visual inspection method according to claim 6 or 7,
An appearance inspection method further comprising a type determination step of determining the type of the inspection object using the reference data.

Images