KR101782366B1 - Vision inspection method based on learning data using normalization of sample scale - Google Patents

Abstract

According to the present invention, a learning-based vision inspection method performing normalization by scale correction comprises: a preliminary learning step of including the same number of real samples and virtual samples to form a first learning group, collecting a first learning image group consisting of real images and virtual images from the first learning group, dividing the real images and the virtual images included in the first learning image group into grids of an equal size to derive digitized real data and virtual data from the divided spaces, allowing the same number of the real data and the virtual data to form a first data group, and displaying the first data group in a feature space to derive a classification reference which is a boundary between a real area and a virtual area; a defect determination step of deriving digitized inspected data from an inspected sample acquired from an inspection target to substitute the inspected data into the feature space and determine an area where the inspected data are located in the feature space to classify the inspection target; and an additional learning step of applying the inspected data in the preliminary learning step to correct the classification reference in the feature space. The real samples, the virtual samples, and the inspected sample are converted into the real images, the virtual images, and an inspected image by undergoing an expanded or reduced normalization step. The defect determination step and the additional learning step are repeatedly performed.

Description

[0001] VISION INSPECTION METHOD BASED ON LEARNING DATA USING NORMALIZATION OF SAMPLE SCALE [0002]

The present invention relates to a vision inspection method for inspecting defects of a polarizing layer adhered to a display panel, and more particularly, to a vision inspection method for processing a sample of an image to be an object to be examined, Vision inspection method.

Generally, a display panel is formed by stacking a sheet or panel having characteristics of a reflective layer, a light guiding layer, a diffusion layer, a polarizing layer, etc., and defects such as sticking or lifting may occur in the process of stacking the respective sheets or panels have.

Such a defective display panel is inspected through a process of determining defects, and a defective portion is subjected to a repair or discard process.

1 is an image of intrinsic defects and false defects commonly found in the inspection process of a display panel, wherein intrinsic defects are defects in the performance of a product, and false defects are detected as defects in the inspection process Quot; good "

The samples of the intrinsic defect and the false defect shown in Fig. 1 exist in various sizes and shapes on the surface of the display panel.

In most cases, they exist in very small dots, and it is easy to distinguish them by defective or non-defective products through automated equipment rather than skilled manpower through specimens of intrinsic defects and false defects existing in various sizes and shapes Did not do it.

Therefore, it is not possible to accurately distinguish and distinguish various types of true defectiveness and false defectiveness in various vision inspection apparatuses that acquire image information and read image information obtained through automated apparatuses to discriminate defects, These disadvantages lead to low discriminative power of bad discrimination, resulting in a large number of false defects classified as defective products even though they are good products.

The system for observing the flat surface to be inspected of the display panel by the human eye and performing the classification of the good and bad is required to have a skilled workforce and the defect detection rate is different according to the skill level of the inspector There has been a disadvantage in that there is a limit to the speed of inspecting defects. To overcome such disadvantages, methods and apparatuses for detecting defects through an automated system have been proposed. However, the above- .

Therefore, a method for solving such problems is required.

SUMMARY OF THE INVENTION The present invention has been devised in order to solve the problems of the prior art described above and aims at enhancing discrimination between true defects and false defects by combining identification and learning discrimination techniques with a vision inspection apparatus.

Also, by including a normalization process of processing bad samples on the display panel, which is a standard for defect classification, to be collected with optimized data, it is possible to reduce the recognition error and computation amount of the sample and reduce the detection amount of false defect. This is to provide a method.

The problems of the present invention are not limited to the above-mentioned problems, and other problems not mentioned can be clearly understood by those skilled in the art from the following description.

According to another aspect of the present invention, there is provided a vision-based vision inspection method for performing sample normalization through scale correction according to the present invention, comprising the steps of: And collects a first learning image group composed of a true image and a false image, respectively, from the first learning group, and each intrinsic image and the false image included in the first learning image group are partitioned into a grid of the same size The intrinsic data and the pseudo-data are derived from the respective partitioned spaces, the intrinsic data and the pseudo-data form the first data group, and the first data group is displayed on the feature space, A preliminary learning step of deriving a classification criterion which is a boundary of a pseudo range where the area and the pseudo data are located, A test sample is obtained from the attached display panel, test image data extracted from the test sample is inputted into the feature space, and the region where the test data is located is set in the feature space with the classification reference as a boundary, And an additional learning step of correcting the classification criterion in the feature space by applying the test data to the failure discrimination step and the pre-learning step in which the false test or the false test is judged, and the intrinsic specimen, the false specimen and the test specimen are enlarged or reduced The false image, the false image, and the image to be inspected, and the failure determination step and the additional learning step are repeatedly performed.

Then, the normalization step enlarges or reduces the intrinsic specimen, the false specimen, and the specimen to a predetermined size, and corrects the intrinsic specimen, the false specimen, and the specimen to minimize the margin of the specimen.

Alternatively, the intrinsic image, the vital image, and the image to be inspected are each partitioned into a grid of 16 divisions and 16 intrinsic data, pseudo-data, and test data are derived from each intrinsic image, a false image, and a test image.

The intrinsic specimen, the false specimen, and the specimen are image information of a point where the specimen surface of the inspected object is moved according to a certain pattern and the modulated signal of the photographed image is abruptly changed.

Alternatively, the pre-learning step may include an image collecting step in which intrinsic images and a false image are collected as image information from each of the intrinsic specimens and the false specimens included in the first learning group, The intrinsic data and the pseudo-data are obtained from the n minutiae points derived from the respective divided spaces, and the intrinsic data and the pseudo-data form the first data group, and the first data group is schematized, A dictionary creation step of creating a dictionary composed of k average points by obtaining a minutiae point average for each group, a step of creating a diagram in the feature space by substituting the intrinsic data and the pseudo data in advance, Classification that determines the classification criteria that is the boundary of the pseudo-zone where the zone and pseudo-data are located .

And, in the pre-creation step, K average clustering is performed to determine a dictionary having k average points as reference words.

Alternatively, in the failure determination step, the test data, which is n feature points obtained from the image to be inspected, are compared with the dictionary, and the test data is classified based on k average points and displayed on the feature space.

In order to solve the above-described problems, the learning-based vision inspection method in which the sample normalization is performed through the scale correction of the present invention is performed by obtaining data on the genuine defectiveness and false defectiveness of the display panel as image information, By deriving the boundaries, it is possible to obtain an effect of increasing the accuracy of the faulty discrimination and decreasing the detection of the false defect by discriminating the true defect and the false defect.

Further, there is an advantage in that the image information obtained from the object to be inspected is corrected to a predetermined size, and the normalized information is extracted by extracting the normalized information, The accuracy of the image can be expected even if the characteristics of the image are simple.

This normalization of the image information has an effect of reducing the error of recognition when examining a specimen suspected to be defective and making it easy to acquire data meeting predetermined criteria, thereby obtaining the inspection result with improved reliability .

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description of the claims.

FIG. 1 is an image of the intrinsic defect and the false defect formed on the surface of the display panel.
FIG. 2 is a flowchart illustrating a learning-based vision inspection method in which sample normalization is performed through scale correction according to an exemplary embodiment of the present invention.
FIG. 3 is a flowchart illustrating data defining a classification criterion in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention.
4 is a state diagram illustrating a procedure of performing a sample normalization step of a learning-based vision inspection method in which sample normalization is performed through scale correction according to an exemplary embodiment of the present invention.
FIG. 5 is a state diagram showing an intrinsic image corrected through a normalization step in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an exemplary embodiment of the present invention.
6 is a state diagram illustrating a pseudo-image corrected through a normalization step in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention.
FIG. 7 is a schematic diagram illustrating a process of deriving a dictionary in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention.
FIG. 8 is a schematic diagram showing a process of assigning a first learning image group in advance in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention.
FIG. 9 is a schematic diagram showing a learning step and a defect determination step in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention.
FIG. 10 is a graph illustrating a three-dimensional state in which an intrinsic region and a false region are derived on a feature space in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an exemplary embodiment of the present invention.
11 is a state diagram showing an apparatus for implementing a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. In describing the present embodiment, the same designations and the same reference numerals are used for the same components, and further description thereof will be omitted.

The learning-based vision inspection method in which the sample normalization is performed through the scale correction according to the present invention can be performed as follows.

FIG. 2 is a flowchart illustrating a learning-based vision inspection method in which sample normalization is performed through scale correction according to an exemplary embodiment of the present invention. FIG. 3 is a flowchart illustrating a method of performing a normalization using a scale correction according to an exemplary embodiment of the present invention. Based vision inspection method according to an embodiment of the present invention.

Based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention is a learning-based vision inspection method for inspecting defects of the polarizing layer 304 adhered to the display panel 302 .

Referring to Figs. 2 to 3, a first learning group formed by including the intrinsic specimen 100, which is a sample of intrinsic defects, and the tentative specimen 200, which is a sample of a false defective, are provided in the same number.

A first learning image group consisting of the intrinsic image 110 and the false image 210 is collected from the first learning group.

Each of the intrinsic image 110 and the tentative image 210 included in the first learning image group is partitioned into a lattice of the same size and the intrinsic data 120 and the tentative data 220 formulated from each of the divided spaces Lt; / RTI >

The first data group in which the formulated intrinsic data 120 and the false data 220 are provided in the same number is extracted.

The first data group is displayed on the feature space 400 to derive the classification reference 410 that is the boundary between the intrinsic region 130 where the intrinsic data 120 is located and the false region 230 where the false data 220 is located Learning step S100 is performed.

The test data 320 extracted and extracted from the display panel 302 on which the polarizing layer 304 which is the inspection object 300 is adhered and formulated is substituted into the feature space 400 so that the classification reference 410 is bounded A defect determination step S200 of determining a region where the test data 320 is located on the feature space 400 and determining the test object 300 as an intrinsic defect or a false defect is performed.

The intrinsic specimen 100, the false specimen 200 and the inspected specimen are converted into the intrinsic image 110, the false image 210 and the inspected image 310 through the normalization step of being enlarged or reduced, And the additional learning step S300 are repeatedly performed.

Each of the above steps will be described in detail below.

The first learning group includes the same number as the intrinsic specimen 100, which is a sample of the intrinsic defective product, and the simulated specimen 200, which is a sample of the false defective product.

The first learning image group includes the same number of the intrinsic image 110 derived from the intrinsic specimen 100 and the intrinsic image 210 derived from the pseudo-specimen 200. [

The first data group includes the same number of the intrinsic data 120 in which the intrinsic image 110 is formulated and the false data 220 in which the virgin image 210 is formulated.

Therefore, it is assumed in the embodiment of the present invention that the intrinsic specimen 100 and the intrinsic image 110 are the same A, and the false specimen 200 and the false image 210 are also the same A.

The feature space 400 is a multi-dimensional space having a plurality of variables, and refers to a space in which specific data according to an embodiment of the present invention is depicted.

The pre-learning step S100 includes an image collecting step S110 for obtaining an intrinsic image 110 and a false image 210, which are image images of suspected bad points from the intrinsic specimen 100 and the false specimen 200, Creating step S120 and class classification step S130.

In the image collection step S110, an image is collected through a photographing unit 520 for photographing the moving target planes of the intrinsic specimen 100 and the tentative specimen 200 in a predetermined pattern, The virtual image 110 or the false image 210 is acquired in the form of the image shown in FIG. 1 by photographing the image of the position at the time of deriving the result.

In an embodiment of the present invention, when a change in brightness is rapidly increased or decreased in an image captured through the photographing unit 520, the point may be photographed as an image shown in FIG. 1 and acquired as a video image.

FIG. 4 is a state diagram illustrating a procedure of performing a sample normalization step of a learning-based vision inspection method in which sample normalization is performed through scale correction according to an exemplary embodiment of the present invention. FIG. FIG. 6 is a diagram illustrating a state in which an original image is corrected through a normalization step in a learning-based vision inspection method in which sample normalization is performed through scale correction. FIG. 6 is a diagram illustrating a state in which sample normalization is performed through scale correction according to an exemplary embodiment of the present invention Based vision inspection method in which a normalized image is corrected.

The sample normalization step S10 is a step of generating the intrinsic image 110 and the tentative image 210 from the intrinsic image 100 and the tentative image 200, A background included in the false image 210 is removed, and a correction for enlarging or reducing a portion having a singular point is performed to normalize the image to a preset image size.

Then, the intrinsic image 110 and the pseudo image 210 normalized to a preset image size are divided into n divided grid shapes, and the feature points are respectively derived from the divided n regions as vector quantities to generate intrinsic data 120 and As the falsification data 220.

In this sample normalization step S10, the magnitude of the defective point, which appears as a phenomenon in the polarizing layer of the display panel or appears as a floating phenomenon, is not constant, and the pattern of the characteristic is simple, so that one intrinsic image 110 or the false image 210, It is not easy to derive n feature points from a plurality of feature points and it is not effective to derive n feature points by scanning the intrinsic image 110 or the false image 210. Accordingly, In deriving the data 120 or the false data 220, it is possible to normalize the sample to reduce the computational process, as well as to obtain the feature points under relatively similar conditions, thereby improving the reliability of the intrinsic data 120 and the false data 220 There is an effect that can be increased.

FIG. 7 is a schematic diagram illustrating a process of deriving a dictionary in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention. FIG. FIG. 2 is a schematic view showing a process of assigning a first learning image group in advance in a learning-based vision inspection method in which sample normalization is performed through a normalization process.

The preprocessing step S120 derives the intrinsic data 120 and the pseudo-data 220 derived from the intrinsic image 110 and the voiced image 210, respectively, into n feature points.

The intrinsic data 120 is obtained A x n, and the mantissa data 220 is also obtained A x n.

At this time, the sample normalization step S10 may be performed upon acquiring the intrinsic data 120 and the false data 220 from the intrinsic image 110 and the false image 210.

7 through 8, intrinsic data 120 and tentative data 220 derived from intrinsic image 110 and tentative image 210 are used to generate color, Such as brightness, saturation, shape, curvature, etc., can be extracted from n points.

The intrinsic data 120 and the voiceless data 220 are vector quantities having characteristics such as color, lightness, saturation, shape, curvature, etc. as variables and are plotted on a multidimensional space having such vector quantities as variables.

FIG. 7 schematically shows a process in which n feature points are derived as vector quantities from the intrinsic image 110 and the tentative image 210 and are expressed in a multidimensional space.

This is an illustrative example, and the intrinsic data 120 and the false data 220 can be extracted based on various factors according to an embodiment to which the present invention is applied.

As described above, A x n number of false data 220 are collected from A number of intrinsic images 110, A x n number of intrinsic data 120, A number of false images 210, and the intrinsic data 120 ) And the false data 220 are schematized on the multidimensional space as described above, and are grouped into k groups by grouping the adjacent intrinsic data 120 and the false data 220, and the intrinsic data 120 belonging to each group 120 and the false data 220 are calculated to calculate k average points.

N feature points derived from the intrinsic image 110 and the tentative image 210 are enlarged or reduced as shown in Figs. 4 to 6 in an embodiment of the present invention to generate an intrinsic image 110 with a minimized margin, The image 210 may be represented by the number of spaces derived by dividing the image 210 by a grid of equal intervals.

Specifically, the intrinsic image 110 and the false image 210 are divided into quadrants and quadrangled vertically, so that the sixteen lattice-shaped spaces generated are n feature points, respectively, where n = 16.

The intrinsic image 110 and the tentative image 210 are divided into sixteen equally spaced spaces and a vector amount having characteristics such as color, brightness, saturation, shape, curvature, etc. as variables is extracted from each space Thereby generating the intrinsic data 120 and the false data 220. [

In the dictionary creation step S120, each group is defined as a word, a dictionary having k words is defined by defining k average points as dictionaries, and a k average clustering algorithm, which is an algorithm for grouping the given data into k clusters, . ≪ / RTI >

The class classification step S130 is a step of classifying the intrinsic data 120 into dictionaries created through the preparation step S120 and histologizing the intrinsic data 120 belonging to each of the k words as shown in Fig. The histogramming step is performed first.

FIG. 9 is a schematic diagram showing a learning step and a failure determination step in a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention. FIG. 3 is a graph illustrating a state in which an intrinsic region and a false region are derived on a feature space in a learning-based vision inspection method in which sample normalization is performed through scale correction.

10 shows the words and the intrinsic data 120 belonging to the respective words on the basis of the histogram obtained through the histogramming step on the feature space 400 as variables, The intrinsic region 130 is derived.

In addition, the histogramming step of substituting the pseudo-data 220 into the dictionary created through the preprocessing step S120 and histogramizing the pseudo-data 220 belonging to each of the k words, as shown in Fig. 8 .

The words and the pseudo data 130 belonging to each word are displayed on the feature space 400 on the basis of the histogram obtained through the pseudo data 220 and the pseudo data 130 on the feature space 400, Area 230 is derived.

The reference variable shown on the feature space 400 is the number of each word and the pseudo data 130 belonging to each word.

The classification reference 410 may be expressed as a line, a plane, or a hyperplane which is a boundary between the intrinsic region 130 and the pseudo region 230 on the feature space 400, and the intrinsic region 130 and the pseudo- A support vector machine (SVM) technique may be used to determine the boundary between the feature space 400 and the feature space 400 to maximize the space between them.

In the defect determination step S200, the surface on which the polarizing layer 304 is adhered is continuously photographed on the display panel 302, which is the inspection object 300, along a predetermined path through the photographing unit 520, (S210) of acquiring an image 310 of an image of a position indicating the position of the subject image 310 and the intrinsic image 110 and the false image 210 of the image 310, And the visible image 310 are previously substituted and displayed on the feature space 400 using the same method as the method of transforming the image into the visible image 310 and the false data 220, Judging whether the genuine region 130 or the false region 230 belongs to the classification reference 410 obtained through the classification step 410 and determining whether the genuine region 130 or the false region 230 is genuine or false S220).

In the image sensing step S210, the sample normalization step S10 of the preprocessing step S120 described above may be performed even when the inspection data 320 is acquired from the image 310 to be inspected.

The sample normalization step S10 is a step of removing the background included in the image to be inspected 310 as shown in FIG. 4 when extracting the image 310 from the object to be inspected 100, And performs normalization to a preset image size.

Then, the image 310 to be normalized with a predetermined image size is divided into n divided grid shapes, and the feature points are derived from the divided n regions as the vector quantities, and are derived as the test data 220.

In the sample normalization step S10, the size of the defective point appearing as a phenomenon to be picked up or floating in the polarizing layer of the display panel is not fixed, and the pattern of the characteristic is simple, Since it is not easy to derive n feature points from the image 310 and derive the n feature points from the image 310, it is difficult to derive n feature points to derive the test data 320, So that the feature points can be obtained under relatively similar conditions, thereby improving the reliability of the test data 320. [

Similar to the normalization method of the intrinsic image 110 and the voiced image 210 described above, the image to be inspected 310 is divided into four by four and divided into four vertically, Are n feature points, respectively, and n = 16 at this time.

The test image 310 is divided into 16 equally spaced spaces and extracted from each space to extract a vector amount having parameters such as color, lightness, saturation, shape, curvature, .

In an embodiment of the present invention, the intrinsic image 110, the false image 210, and the imaged image 310 may be configured to partition into a lattice form evenly dividing into 16 parts, as shown in Figs. 4 to 6 And extracting the vector quantities in the respective spaces by dividing them into 16 equal parts can provide sufficient data to discriminate between the true defectiveness and false defectiveness in the vision inspection apparatus for inspecting the defect caused by lifting or dropping on the display panel However, it should be understood that the present invention is not limited to the above-described embodiments, but may be variously embodied according to an embodiment to which the present invention is applied.

In the additional learning step S300, a step of correcting the dictionary through the process of adding the image 310 to be inspected obtained in the failure determination step S200 in advance and re-reproducing k average points may be added.

Alternatively, a classification criterion correction step S320 for adding the test data 320 obtained through the failure judgment step S200 on the feature space and correcting the classification criterion 410 by reflecting the added test data 320 is performed .

In the step of correcting the dictionary, the classification criterion correction step (S320) may be performed by performing defect determination on the learned defect criterion learned through the pre-learning step (SlOO), adding new information to be acquired, Can be obtained.

It is possible to increase the accuracy of performing the failure discrimination by increasing the number of learning samples and to reduce the phenomenon in which the false defect is detected by making the boundary between the true defect and the false defect more apparent in the feature space 400. [

11 is a state diagram showing an apparatus for implementing a learning-based vision inspection method in which sample normalization is performed through scale correction according to an embodiment of the present invention.

11, the transfer unit 510 to which the inspection object 300 is transferred, the imaging unit 520 for continuously moving the inspection surface of the inspection object 300 along a predetermined path, A reading unit 530 for reading the image obtained through the reading unit 530 and a sensing unit 530 for reading the image obtained through the reading unit 520, An operation unit 540 for rendering the data on the display unit 400 and a storage unit 550 for storing the test data 320 obtained through the operation unit 540.

It will be apparent to those skilled in the art that the present invention can be embodied in other specific forms without departing from the spirit or scope of the invention as defined in the appended claims. It is obvious to them. Therefore, the above-described embodiments are to be considered as illustrative rather than restrictive, and the present invention is not limited to the above description, but may be modified within the scope of the appended claims and equivalents thereof.

100: intrinsic specimen 110: intrinsic image
120: intrinsic data 130: intrinsic region
200: False sample 210: False image
220: falsifying data 230:
300: object to be inspected 302: display panel
304: polarization layer 310: image to be imaged
320: Test data
400: Feature Space 410: Classification Criteria
510: transfer means 520: photographing unit
530: Reading section 540:
550:

Claims (7)

A learning-based vision inspection method for inspecting defects of a polarizing layer adhered to a display panel,
A first learning group is formed by including the same number of false samples that are samples of intrinsic defective products and a false sample that is a sample of false defects and a first learning image group composed of a true image and a false image are collected from the first learning group, Wherein each of the intrinsic image and the pseudo image included in the first learning image group is partitioned into a lattice of the same size so that intrinsic data and pseudo data formulated from each of the partitioned spaces are derived, Wherein the first data group is formed with the same number of the pseudo-data, the first data group is displayed on the feature space, and a dictionary for deriving a classification criterion that is a boundary between the intrinsic region where the intrinsic data is located and the pseudo- Learning stage;
A method of analyzing an object to be inspected, comprising the steps of: acquiring a specimen from a display panel to which a polarizing layer as an object to be inspected is adhered; extracting an image to be inspected from the specimen to be inspected; Determining a region in which the data is located to determine that the object to be inspected is an intrinsic defect or a false defect; And
And an additional learning step of applying the test data to the pre-learning step and correcting the classification criteria in the feature space,
The false sample, the false sample, and the test sample are converted into the true image, the false image, and the detected image through a normalization step of being enlarged or reduced, and the failure determination step and the additional learning step are repeatedly performed,
Wherein the normalizing step includes a step of enlarging or reducing the intrinsic specimen, the pseudo specimen, and the specimen to a preset size, and performing a scale correction to scale the intrinsic specimen, the false specimen, A learning-based vision inspection method in which normalization is performed.
delete The method according to claim 1,
Wherein the intrinsic image, the false image, and the image to be examined are each partitioned into a grid shape of 16 divisions, and 16 intrinsic data, the false data, and the test data from each of the intrinsic image, the false image, Based vision inspection method in which sample normalization is performed through scale correction.
The method according to claim 1,
Wherein the intrinsic specimen, the false specimen,
Wherein the sample normalization is performed through scale correction which is image information of a point at which a scanned surface of the object to be inspected is moved according to a predetermined pattern and a modified signal of the captured image is abruptly changed.
The method according to claim 1,
Wherein the pre-
An image collection step in which the intrinsic image and the false image are collected as image information from each of the intrinsic specimen and the pseudo specimen included in the first learning group;
Each of the intrinsic image and the pseudo image is divided into n divided spaces and the intrinsic data and the pseudo data are obtained from n minutiae points derived from the respective divided spaces, 1) grouping the first data group, grouping adjacent feature points into k groups, obtaining a feature point average for each group, and creating a dictionary composed of k average points; And
A classifying unit for classifying the intrinsic data and the pseudo data into the dictionary and drawing it on the feature space and determining a classification reference that is a boundary between the intrinsic region where the intrinsic data is located and the pseudo- step;
Based vision inspection method in which sample normalization is performed through scale correction that includes scale correction.
6. The method of claim 5,
The pre-
K-means clustering is performed to perform sample normalization through scale correction to determine the dictionary with k average points as a reference word.
6. The method of claim 5,
In the failure determination step,
A learning-based vision in which sample normalization is performed through scale correction in which the test data, which is n feature points obtained from the image to be inspected, is compared with the dictionary and the test data is classified based on the k average points, method of inspection.

Images