KR101609007B1 - Image data processing apparatus and method for defect inspection, defect inspecting apparatus and method using the image data processing apparatus and method, board-like body manufacturing method using the defect inspecting apparatus and method, and recording medium - Google Patents

Abstract

Among the photographed images of the plate-like body, the defect candidates are extracted using the signal threshold value. By searching for a defect candidate having the same position in the width direction among the plurality of extracted defect candidates and obtaining the interval between the position in the carrying direction of the searched defect candidate and the position in the carrying direction of the adjacent defect candidate, Find the occurrence frequency distribution. When the occurrence frequency of the attention interval exceeds the frequency threshold value, it is determined that the plate-shaped body has a periodic defect in the conveying direction. At this time, the frequency threshold value is set so as to become substantially larger as the notable interval becomes shorter.

Description

TECHNICAL FIELD [0001] The present invention relates to an apparatus and method for processing image data for defect inspection, a defect inspection apparatus and method using the same, a method of manufacturing a plate-shaped body using the same, and a recording medium. USING THE IMAGE DATA PROCESSING APPARATUS AND METHOD, BOARD-LIKE BODY MANUFACTURING METHOD USING THE DEFECT INSPECTING APPARATUS AND METHOD, AND RECORDING MEDIUM}

The present invention relates to an apparatus and method for processing image data for defect inspection for detecting defects present in a transparent plate-like body such as a glass plate, a defect inspection apparatus and a defect inspection method each using the same, And a computer-readable recording medium on which a program for executing a method of processing image data for defect inspection is recorded.

In recent years, glass plates are used in electronic devices such as flat panel displays and thin-film solar cells. Therefore, there is a strong demand for a glass plate having a small plate thickness and few or no defects such as bubbles and scratches.

Examples of defects present on the glass plate include scratches formed on the surface of the glass plate. For example, in the float method, a long plate-shaped body having a certain thickness is taken out from the melting furnace, and is conveyed on a driving roller. At this time, foreign matter adhered to the drive roller, minute projections on the drive roller, and the like cause scratches on the surface of the glass plate, thereby causing scratches. Since a large number of drive rollers are used for transporting the glass plate, there is a very large chance that minute scratches are generated on the surface of the glass plate. Since such flaws occur with periodicity, various methods have been proposed for inspecting defects having periodicity in the in-process inspection of a product having a long intermediate shape instead of the glass plate.

In order to measure the cycle of defects present in a running inspection object, the defect data and the normal data are binarized to obtain the distance between the defect data, and the frequency is calculated for each of the obtained distances, A periodic component is calculated, and a fundamental period of the defect is extracted from the periodic component.

The following Patent Document 2 discloses a method of inspecting periodic defects by classifying defects detected in image data obtained by picking up an inspection object into periodic defects and non-periodic defects.

Japanese Patent Publication No. 7-86474 Japanese Patent Application Laid-Open No. 2006-308473

However, in Patent Document 1, when calculating the period component of the distance between defect data, it is difficult to calculate the periodic component effectively because defect data and noise data can not be distinguished. When the defect is small, since the threshold value to be binarized must be lowered, the number of handling of the noise data as defect data becomes extremely large, and the accuracy of calculation of the periodic component gradually decreases.

On the other hand, in Patent Document 2, when image data obtained by picking up an image of a subject is classified into a periodic defect and an aperiodic defect, the image data is binarized and the area of a portion regarded as a defect obtained by binarization Is larger than a predetermined value is classified as an acyclic periodic defect. As a result, a portion regarded as a defect due to noise data having an extremely small area is misidentified as a periodic defect and identified.

That is, since the defects of the small scratches caused by the driving roller or the like generated on the glass plate are small, it is difficult to distinguish the noise data randomly generated even when the threshold value is lowered it's difficult. Therefore, it is extremely difficult to discriminate the periodicity of a defect of small scratches caused by the driving roller or the like.

SUMMARY OF THE INVENTION Accordingly, the present invention has been made keeping in mind the above problems occurring in the prior art, and an object of the present invention is to provide a defect inspection method and a defect inspection method capable of detecting the presence of a periodic defect even when a defect existing in a plate- And a defect inspection method using the defect inspection apparatus and a defect inspection apparatus, a method of manufacturing a plate-shaped body using the inspection method or a defect inspection apparatus, and a method of processing image data for defect inspection And a computer readable recording medium storing a program to be executed.

In order to achieve the above object, a first aspect of the present invention is a method for inspecting a defect existing in a plate-like body by using an image obtained by photographing a plate- And extracting a plurality of defect candidates from the image using the first signal threshold value and extracting a plurality of defect candidates from among the plurality of extracted defect candidates by using a defect candidate having the same position in the width direction orthogonal to the moving direction in the predetermined direction Of the defective candidate detected by the search and the position of the defective candidate in the moving direction of the adjacent defective candidate in the moving direction with respect to the position of the detected defective candidate in the moving direction in the plate- A plurality of intervals are obtained by repeating obtaining the interval, and the occurrence frequency in the plurality of intervals is obtained, The frequency threshold value used in the processing section is determined according to the interval of interest, and the frequency threshold value used in the processing section is determined according to the interval of interest , And when the two frequency thresholds are set as different values, the frequency interval threshold value defining the larger frequency threshold value is set so that the frequency threshold value is smaller than the frequency interval of interest, which defines the smaller frequency threshold value And the processing time is set to a predetermined value.

In a second aspect of the present invention, in the first aspect, the processing section has a reference table indicating a relationship between the density of occurrence of the defect candidate and the first signal threshold value, and the density of occurrence of the defect candidate is And the frequency threshold value is a value that varies in accordance with the value of the target generation density in addition to the noted interval.

According to a third aspect of the present invention, in the first or second aspect of the present invention, the frequency threshold value used in the processing section is a value obtained by assuming that noise components are randomly distributed in an area, The noise component is determined as the defect candidate and the frequency of occurrence of the noise component with respect to the interval is analytically determined or the image of the noise component in the simulation image formed with the noise component is determined as the defect candidate, Wherein the frequency of occurrence of the noise component is determined based on the frequency of occurrence of the noise component.

A fourth aspect of the present invention provides the processing apparatus according to the third aspect, wherein the generation of the noise component changes the generation density of the noise component in the image in accordance with the area of the image.

In a fifth aspect of the present invention, in any one of the first to fourth aspects, the processing unit is configured to divide an image of a search target for searching for a defect candidate into a plurality of unit areas in the width direction and the moving direction, When a plurality of unit regions including a plurality of defect candidates are located at the same position in the width direction, the defect candidates are assumed to have the same position in the width direction, and the intervals and the occurrence frequency And provides a processing device for obtaining the processing result.

In a sixth aspect of the present invention, in any one of the first to fifth aspects, the processing section obtains a width direction occurrence frequency distribution indicating a distribution of the occurrence frequency at the position in the width direction, The defect occurrence pattern is classified using the deviation of the occurrence frequency along the width direction.

In a seventh aspect of the present invention, in any one of the first to sixth aspects, the plate-shaped body is a long shape continuous in the moving direction, and the processing unit is configured such that the plate- And an image of the region is subjected to inspection of one unit and the determination is made for a plurality of units.

In a seventh aspect of the present invention, in the seventh aspect, the processing section records the occurrence frequency distribution of the interval determined by the interval and the position in the width direction for the plurality of time series units, There is provided a processing apparatus for displaying occurrence information of defects on the screen by calculating the occurrence frequency by determining the interval to be noticed from the distribution and the position in the width direction and displaying the occurrence frequency as time series data.

In a ninth aspect of the present invention, in the form 8, time series data of a plurality of occurrence frequencies in which at least one of the noted interval and the position in the width direction is changed with respect to the time series data of the occurrence frequency is overwritten on the same graph And provides a processing device for displaying on a screen.

According to a 10th aspect of the present invention, in any one of the first to ninth aspects, when the defect candidate having the same position in the width direction is searched for and detected in the movement direction, A plurality of intervals are obtained by repeating obtaining the interval in the moving direction between a plurality of defective candidates and a plurality of defective candidates in addition to obtaining the interval in the moving direction as the previous defect candidates, , And the plate-like body has a periodic defect in the moving direction when the occurrence frequency of the noticed interval exceeds a set frequency threshold value.

According to a eleventh aspect of the present invention, in any one of the first to tenth aspects, when the interval determined to have periodic defects in the moving direction is a pitch interval, the processing unit further includes: The attention area including the position in the width direction where the defect candidate having the gap is located is determined and the detail defect candidate is extracted from the end of the image using the second signal threshold value among the images of the attention area, Determining a search area centered on a position spaced apart by the pitch interval in the moving direction from a position of the obtained detailed defect candidate and searching for a detailed defect candidate using the second signal threshold value for the search area, Evaluating the attributes of the detailed defect candidates detected by searching and the detailed defect candidates extracted and obtained, A zone provides a processing apparatus to determine whether or not including a periodic defect candidate in detail in the movement direction.

According to a twelfth aspect of the present invention, in any one of the first to eleventh aspects, the processing section searches for a defect candidate having the same position in the width direction in the moving direction and detects the gap, And evaluates the degree of similarity between the attribute of the candidate or the defect candidate and a predetermined defect candidate and obtains the interval when at least one of the attribute and the degree of similarity satisfies the set condition.

A defect inspection apparatus according to a thirteenth aspect of the present invention is a defect inspection apparatus for inspecting a defect existing in a plate-like body, comprising: a light source for projecting light onto a surface of the plate-like body; , A camera for photographing an image of a plate-like body projected onto the light source, and a processing device according to any one of the above-mentioned forms 1 to 12, wherein the processing section of the processing device And extracting the plurality of defect candidates using the first signal threshold value and extracting a plurality of defect candidates from among the plurality of defect candidates extracted in the width direction position orthogonal to the moving direction that is the direction of relative movement between the camera and the plate- The defective defect candidate is searched in the movement direction for the same defect candidate.

Form 14 of the present invention is a method of processing image data for defect inspection for inspecting a defect existing in the plate-like body using an image obtained by photographing the plate-shaped body while relatively moving the plate-shaped body in a predetermined direction, Extracting a plurality of defect candidates from the obtained image using the first signal threshold value and extracting a defect candidate having the same position in the width direction orthogonal to the moving direction as the predetermined direction among the extracted plurality of defect candidates, Obtaining a plurality of intervals by repeating the position of the defect candidate detected by the search in the moving direction and the interval between the position of the defect candidate and the moving direction of the adjacent defect candidate in the moving direction, When the frequency of occurrence in the plurality of intervals is found and the frequency of occurrence of the interval of interest exceeds the set frequency threshold value, The upper body is determined to have a periodic defect in the moving direction, the frequency threshold is determined according to the noted interval, and when the two frequency thresholds are different, the higher frequency threshold value Wherein the frequency threshold value is set such that the frequency threshold value is smaller than the noticed interval that defines the smaller frequency threshold value.

In a fifteenth aspect of the present invention, in the method according to the fourteenth aspect of the present invention, before the determination, the inspection condition is set, and in the step of setting the inspection condition, the reference table is used so that the density of occurrence of the defect candidate is the set target density Wherein the first signal threshold value is set and the frequency threshold value is a value that changes in accordance with the value of the target generation density in addition to the noted interval.

The 16th aspect of the present invention is the image processing method according to the 14th or 15th aspect, wherein the frequency threshold value is obtained by using an image of the noise component in a simulation image formed by a noise component as a defect candidate, And a defect inspection method is provided which is determined on the basis of the occurrence frequency.

A seventeenth aspect of the present invention provides the defect inspection method according to the sixteenth aspect, wherein the simulation image is generated so that the density of generation of noise components in the image is different depending on the area of the image.

According to a twelfth aspect of the present invention, in any one of the fourteenth to seventeenth aspects, when the interval determined to have periodic defect candidates in the moving direction is a pitch interval, Thereafter, further, an attention area including the position in the width direction in which the defect candidate having the pitch interval is located is determined, and from among the images of this attention area, the detail defect candidate is determined from the end of the image using the second signal threshold value Extracting the defect candidate from the position of the detailed defect candidate obtained by the extraction, determining a search region centered on a position spaced apart by the pitch interval in the movement direction from the position of the detailed defect candidate obtained by the extraction, Defect candidates are searched and searched to evaluate the attributes of the detected detailed defect candidates and the extracted detailed defect candidates, respectively, Depending on the result, the attention area provides a method for determining whether or not including a periodic defect candidate in the movement direction.

A nineteenth aspect of the present invention provides the image processing method according to the eighteenth aspect, wherein the image used for discriminating the periodic defect candidates of the attention area is an image of a plate after cutting the plate-like body to a predetermined size.

According to a twentieth aspect of the present invention, in any one of the fourteenth to twenty-first aspects, when the defective candidate having the same position in the width direction is searched for and detected in the moving direction and the gap is found, Or defect candidates of a defect candidate with respect to a predetermined defect candidate, and when at least one of the attribute and the similarity satisfies a set condition, the process is provided.

A twenty-first aspect of the present invention is a defect inspection method for inspecting a defect existing in a plate-like body, the method comprising the steps of: projecting light onto a surface of the plate-like body and moving the plate- There is provided a defect inspection method, characterized in that the processing method according to any one of the above-mentioned form 14 to 20 is carried out by using the image obtained by photographing and photographing.

A 22nd aspect of the present invention is a method of manufacturing a plate-shaped member, which is a belt-shaped continuous body conveyed by a conveying roller, using the defect inspection apparatus of form 13 or the defect inspection method of form 21, Characterized in that a conveying roller for inspecting the upper body during movement and causing defects on the path of movement of the plate-shaped body according to the result of inspection is specified, and the specified conveying roller is removed or repaired. ≪ / RTI >

A 23rd aspect of the present invention is a method for manufacturing a plate-shaped member, which is a belt-shaped continuous body conveyed by a conveying roller, using the defect inspection apparatus of the 13th aspect or the defect inspection method of the 21st aspect, A method for manufacturing a plate-shaped body, characterized by inspecting an upper body during movement and cutting out the plate-like body from the widthwise position of a defect determined to have the periodic defect, and taking out the plate-shaped body.

Further, a twenty-fourth aspect of the present invention provides a computer-executable program for executing a method of processing image data for defect inspection according to any one of the fourteenth to twentieth aspects, and a computer readable recording medium storing the program.

Further, according to a twenty-fifth aspect of the present invention, in any one of the first to tenth aspects, when the interval determined to have periodic defects in the moving direction is a pitch interval, Determining an attention area including the position in the width direction in which the defect candidate having the pitch interval is located and extracting the detail defect candidate from the end of the image using the second signal threshold value from the image of the attention area, Wherein a search area centered on a position spaced apart by a distance corresponding to a circumferential length of a conveying roller for conveying the plate-like object in the moving direction is determined from the position of the detailed defect candidate obtained by extraction, A detailed defect candidate is searched for using the second signal threshold value, a search is made to find the detailed defect candidate, and a detailed defect candidate And evaluating the properties of the beam and, based on the evaluation result, determining whether the attention area includes the periodic detailed defect candidate in the movement direction.

Further, in a twenty-sixth aspect of the present invention, in any one of the fourteenth to seventeenth aspects, when the interval determined to have periodic defect candidates in the moving direction is a pitch interval, Further comprising the step of determining an attention area including the position in the width direction in which the defect candidate having the pitch interval is located and from the end of the image using the second signal threshold, Determining a search area centered on a position spaced apart by a distance corresponding to the circumferential length of the conveying roller that conveys the plate-shaped body in the moving direction from the position of the detailed defect candidate obtained by the extraction; Searching for a detailed defect candidate with respect to the search area using the second signal threshold value, searching for a detailed defect candidate detected, Output to each of the evaluation of the properties detailed defect candidate obtained, according to the evaluation result, the attention area provides a method for determining whether or not including a periodic defect candidate in the movement direction.

In the apparatus for processing image data for defect inspection of form 1 of the present invention, the defect inspection apparatus of form 13, the processing method of form 14, the defect inspection method of form 21, and the program and recording medium of form 24, And frequency thresholds are used to determine the presence or absence of periodic defects. Further, in these, the frequency threshold value is determined according to the interval of interest, and when the two frequency threshold values are set as different values, the noticed interval that defines the larger frequency threshold value is set to a smaller frequency threshold value The frequency threshold value is set to be a smaller value than the noted interval.

This makes it possible to discriminate the presence of periodic defects even when the noise component is included in the photographed image.

First, in the second and the fifteenth aspects, it is assumed that the density of occurrence of the defect candidate is equal to or greater than the set generation density of the image data for the defect inspection of the present invention By setting the first signal threshold value as high as possible and also by changing the frequency threshold value in accordance with the value of the target generation density in addition to the interval to be noted, the presence of more efficient and periodic defects can be determined. Even if the density of occurrence of the defect candidates in the plate-shaped body fluctuates due to various conditions of production of the plate-shaped body, the presence of periodic defects can be optimally discriminated by adapting to the fluctuation on a periodic basis.

In the form 3 and 16, it is possible to easily detect the periodic defect without overcoming the randomly occurring noise component, that is, without being affected by the noise component even in the presence of the noise component. In addition, the frequency threshold value can be easily determined by the occurrence frequency or the simulation image obtained analytically.

In Forms 4 and 17, the image is divided and processed, whereby periodically occurring defects locally occurring in the plate-like body can be easily detected. It is also possible to carry out inspections except for locally generated noise components.

In the fifth aspect, the unit area of the same size is formed in consideration of the positional deviation of the generated defect candidate in the image to be inspected, so that the existence of the periodic defect can be determined more efficiently in a short time.

In the sixth aspect, the defect occurrence pattern can be classified by obtaining the frequency occurrence frequency distribution of the defect candidates and using the deviation of the occurrence frequency along the width direction, so that the defect occurrence pattern can be effectively used for estimating the cause of the defect In addition to this, it is helpful to identify the situation of whether or not the plate-shaped body can be continuously produced stably, which is very helpful for the management of the production process.

In the seventh aspect, the plate-shaped body is divided into regions of a plate-shaped body having a predetermined length, and the images of the regions are subjected to inspection in a time series unit and are discriminated for a plurality of time series units, The frequency information can be represented as time series data. This makes it possible to more reliably discriminate the presence of periodic defects and to grasp the situation of defects that change in a time-wise manner, and it is possible to determine the defectiveness of a plate-like body (glass substrate) This is very helpful for processing in the post-process.

In the form 8, since the occurrence frequency distribution is recorded for a plurality of time series units, it is possible to obtain information on how long periodic defects continue to occur for a period of time, and it is helpful for real-time management in the production process.

In the ninth aspect, time series data of a plurality of occurrence frequencies are overwritten on the same graph, so that a cause of occurrence of a defect that can change complicatedly can be detected easily in a short time.

In the form 10, since the occurrence frequency is determined by obtaining the interval between a plurality of defective candidates and a gap between adjacent defective candidates one before the other, it is determined whether there is a periodical defect, It is possible to easily detect the periodic defects even when the generation positions of the defects are partially overlapped in the plane.

In form 11, form 18, form 25 and form 26, the attention area and the search area are determined to search for the detail defect candidate, and it is determined whether or not the defect candidate includes the periodic detail defect candidate. Can be detected within a short time. On the other hand, if the time required for detection is relatively long, the accuracy of defect detection can be increased.

In the twelfth and twentieth aspects, since the interval between defective candidates is determined when the attribute of the defect candidate or the degree of similarity of the defect candidates satisfies the set condition, it is possible to reliably detect the periodical defect without overlooking it, It is possible to improve the reliability.

Further, in the method for manufacturing a plate-shaped member according to a twenty-second aspect of the present invention, it is possible to shorten the time required for the restoration work in the process in which the conveying roller is provided, and to perform the management when the process is continuously operated So that the yield can be stabilized. In addition, it is possible to facilitate the management of the conveying roller in use in the production process. In addition, the future time at which the conveying roller is to be replaced can be confirmed in advance.

In the method of manufacturing a plate-shaped body according to a twenty-third aspect of the present invention, even if there is a defect in the conveying roller or the like, the plate-shaped body as a continuous strip can be efficiently cut and taken out, .

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1A is a schematic diagram showing a configuration of a defect inspection apparatus according to an embodiment of the defect inspection apparatus of the present invention. FIG.
1B is a view for explaining a glass plate to be inspected in a defect inspection apparatus according to an embodiment of the defect inspection apparatus of the present invention.
2 is a flowchart showing an example of a flow of an embodiment of a defect inspection method of the present invention.
3 is a diagram for explaining a part of the processing of the defect inspection method of the present invention.
4 is a diagram for explaining a part of processing of a defect inspection method of the present invention.
FIG. 5A is a diagram illustrating an example of a frequency threshold value used in the defect inspection method of the present invention. FIG.
FIG. 5B is a view illustrating an example of a frequency threshold value used in the defect inspection method of the present invention. FIG.
FIG. 6A is another example of a frequency threshold value used in the defect inspection method of the present invention; FIG.
6B is another example of a frequency threshold value used in the defect inspection method of the present invention.
FIG. 6C is another example of a frequency threshold value used in the defect inspection method of the present invention; FIG.
7 is a flowchart showing an example of a flow of another embodiment of the defect inspection method of the present invention.
8A is a diagram showing an example of a frequency distribution in a width direction obtained by the defect inspection method of the present invention.
8B is a diagram showing an example of a frequency distribution in the width direction obtained by the defect inspection method of the present invention.
FIG. 8C is a diagram showing an example of the occurrence frequency distribution in the width direction obtained by the defect inspection method of the present invention; FIG.
9 is a diagram showing an example of a time series distribution obtained by the defect inspection method of the present invention.
10 is a flowchart showing an example of a flow of another embodiment of the defect inspection method of the present invention.
11 is a flowchart showing an example of a flow of another embodiment of the defect inspection method of the present invention.
12 is a flowchart showing an example of a flow of another embodiment of the defect inspection method of the present invention.
13A is a view for explaining a defect inspection method shown in Fig. 12; Fig.
13B is a view for explaining a defect inspection method shown in Fig.
14A is a two-dimensional density image showing another example of a time series distribution of occurrence frequencies obtained by the defect inspection method of the present invention.
14B is a three-dimensional graph showing another example of the time series distribution of the occurrence frequency obtained by the defect inspection method of the present invention.
15A is a two-dimensional density image showing another example of a time-series distribution of occurrence frequencies obtained by the defect inspection method of the present invention.
15B is a three-dimensional graph showing another example of the time series distribution of the occurrence frequency obtained by the defect inspection method of the present invention.
16A is a three-dimensional graph showing another example of occurrence frequency obtained in the defect inspection method of the present invention in a time-wise manner;
FIG. 16B is a three-dimensional graph showing another example of occurrence frequency obtained in the defect inspection method of the present invention in a time-wise manner; FIG.
16C is a three-dimensional graph showing another example of occurrence frequency obtained in the defect inspection method of the present invention in a time-wise manner.
17A is a two-dimensional density image that shows another example of the frequency of occurrence obtained in the defect inspection method of the present invention in a time-series manner.
FIG. 17B is a two-dimensional density image that shows another example of the frequency of occurrence obtained in the defect inspection method of the present invention in a time-series manner. FIG.
FIG. 17C is a two-dimensional density image that shows another example of the occurrence frequency obtained in the defect inspection method of the present invention in a time-series manner. FIG.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an apparatus and a method for processing image data for defect inspection of the present invention, a defect inspection apparatus and a defect inspection method using the same, a method of manufacturing a plate- Readable recording medium will be described in detail on the basis of a preferred embodiment shown in the accompanying drawings.

The defect inspection apparatus 1 of FIG. 1A is a defect inspection apparatus of the present invention which implements the defect inspection method of the present invention, and determines whether there is a periodicity of the defect. The defect inspection apparatus 1 mainly has a defect inspection unit 10, a processing unit 16, and a defect inspection unit 26. [

As a plate-like body in the present invention, a glass plate (G) having transparency is exemplified in the following description, but the plate-like body of the present invention is not limited to this. For example, Film or paper.

The glass plate G described below is a continuous strip-like continuous body before being cut into a predetermined size, and mainly the one in the carrying state is described. As the type of the mother glass substrate to be produced, for example, there are G6, G8, G10, and furthermore G12 and the tendency of becoming larger size is shown.

In the present invention, a glass plate cut to a predetermined size may be used. However, in order to determine the periodicity of a defect candidate, it is preferable to use a long glass plate, which is a belt-shaped continuous body, and a glass plate in a carrying state.

In the following embodiments, the glass plate G to be conveyed is photographed using a stationary light source and a camera. In the present invention, the glass plate G may be photographed while the light source and the camera are moving. In the present invention, it is sufficient that there is a relative movement between the light source and the camera and the glass plate (G). The carrying direction described in the following embodiments corresponds to the first moving direction in the present invention, and the width direction corresponds to the second moving direction in the present invention.

The defect inspection unit 10 shown in Fig. 1A is provided in a conveyance path formed by conveyance rollers 11 having a plurality of different radii. The glass plate G constitutes a continuous strip-shaped body taken out to a constant thickness from the melting furnace, and is continuously conveyed on the conveying path and moved.

Here, as the conveying roller 11 for conveying the glass plate G, a driving roller is used, but at least one driven roller may be used between the driving roller. As the conveying roller 11, various types of conveying rollers can be used. For example, it may be a normal conveying roller which contacts the glass plate G in its entire width, such as a pure roller, a sleeve roller, a bag roller, a coating roller, a roller with a film or the like. It is also possible to use a conveying roller that comes into contact with the glass plate G with a gap therebetween in the width direction such as a roller with a jaw or a stepped roller.

The defect inspection unit 10 includes a light source 22 for projecting light onto the surface of the glass plate G, a camera 14 for photographing an image of the plate-like object projected by the light source 22, And a processing unit 16 for discriminating the user. An output system 18 such as a display 18a for displaying an inspection result or the like as a soft copy image on the screen or a printer 18b for outputting a check result or the like as a hard copy image, May be connected to the defect inspection unit 10. [ The defect inspection unit 10 is an apparatus for inspecting a defect candidate in a transmission image by transmission light transmitted through a glass plate G. A light source 22 and a camera 24 are disposed on one side of the glass plate G and the image of defects on the glass plate G is taken as a reflected image by the camera 22, And a defect inspection unit 26 for a reflection image for extracting the position of the defect candidate.

When the glass plate G is taken out from the melting furnace to a constant thickness and is continuously conveyed by the conveying roller 11, as shown in Fig. 1B, (D) occurs periodically. In addition, a point-like defect (X) is generated on the surface of the glass plate (G). Further, a contaminated area Y spreading to a certain area also occurs. In addition, in the image read by the defect inspection unit 10, the dot-like noise randomly generated by the processing at the time of reading the image in addition to the above defects is also visually recognized as the defect (X).

The defect inspection unit 10 periodically generates in the image of this glass plate G at substantially the same position in the width direction of the glass plate G orthogonal to the carrying direction that occurs on the surface of the glass plate G being transported The pitch interval P of the defects D to be distinguished from the defects X is discriminated reliably.

The light source 12 is a light source that emits substantially parallel light and emits substantially parallel light having a substantially uniform light intensity in the width direction of the glass plate G (the direction perpendicular to the plane of FIG. 1A). A halogen light source, an LED light source, or the like is used as the light source, and the type of light is not particularly limited, and white is preferably used.

The camera 14 is a line sensor type camera which is installed at a position opposite to the light source 12 with the glass plate G therebetween and reads the transmitted light transmitted through the glass plate G directly from the light receiving surface. The type of the line sensor of the camera 14 may be a CCD type or a CMOS type, and is not particularly limited. A plurality of cameras 14 are installed in a direction perpendicular to the paper surface in FIG. 1A to photograph the same position in the transport direction. Further, a plurality of cameras are set so that the viewing ranges in the width direction of the glass plate G partially overlap each other .

The image signal photographed by the camera 14 is sent to the processing unit 16. [

The processing unit 16 is a part constituting the processing apparatus of the present invention which performs a method of processing image data for defect inspection of the present invention. The processing section 16 generates image data to be inspected of the glass plate G from the sent image signal and performs defect inspection using the image data. The image signal sent from the camera 14 averages the partially overlapping areas as described above to generate image data constituting one image. The presence or absence of the periodicity of the detected defect candidate is determined by using such image data. This determination will be described later.

The light source 22 of the defect inspection unit 26 is a light source that emits light in a strip shape for irradiating illumination light on the surface of the glass plate G, that is, a substantially parallel light, . The light source 22 is a linear light source extending in a direction perpendicular to the paper plane of Fig. 1A like the light source 12 of the defect inspection unit 10 and is a linear light source that is perpendicular to the width direction of the glass plate G It is preferable to emit substantially parallel light having substantially uniform light intensity. In the present invention, two light sources are provided. As the light source 22, for example, an LED light source or a halogen light source is used. The type of light is not particularly limited and may be red, blue, white, Do.

The camera 24 is a line sensor type camera for condensing the reflected light emitted from the surface of the glass plate G and capturing a reflected image, and a camera of the same type as the camera 14 can be used. The camera 24 is provided on the same side of the glass plate G as the light source 22. The camera 24 and the light source 22 are provided so as to be positioned on the upstream side and the downstream side in the carrying direction and the light output direction of the light source 22 and the view direction of the camera 24 And the reflected light is adjusted to be incident on the camera 24.

The image photographed by the camera 24 is an image that is illuminated by the light source 22 and is reflected on the front and back surfaces of the glass plate G. An image of a defect existing in the glass plate G as an dark portion to be. This image is incident on the surface of the glass plate G from an inclined direction with respect to the surface of the glass plate G and is reflected by the back surface of the glass plate G and then passes through the defective area Actual images are included first.

The incident light incident on the surface of the glass plate G from the inclined direction with respect to the surface of the glass plate G is reflected by the back surface of the glass plate G after passing through the region of the defect in the optical path in the glass plate G Includes a mirror image of the resulting defect.

The image data obtained by the camera 24 in this manner is sent to the sequential processing section 16 every time it is read in a line form. In the processing unit 16, as in the case of the defect inspection unit 10, defect inspection is performed using the sent image data.

Inspection of the defect using the defect inspection unit 26 is performed by checking the surface of the glass plate G (the surface on the upper side of the glass plate G in Fig. 1A) by the displacement of the above- (The lower surface of the glass sheet G in FIG. 1A and the surface on the side of the conveying roller 11), and the attribute as a defect can be determined. In other words, it can be seen that there is a defect on the back surface when there is no positional displacement amount and there is only one image, and it is found that it exists on the surface of the glass plate G or on the surface of the glass plate G when there is a positional displacement amount and two images are seen.

The defect inspection apparatus 1 shown in Fig. 1A includes two defect inspection units 10 and 26, but the present invention is not limited to this and may be provided with only one of them.

In the processing section 16, the image data for defect inspection is processed as follows from the obtained image of the inspection object. 2 is a flowchart showing a flow of a defect inspection method, in particular, a method of processing image data for defect inspection.

First, a defect inspection condition to be detected is set (step S100). Specifically, the allowable amount of fluctuation of the position in the width direction when a defect candidate having periodicity is detected, the allowable amount of fluctuation from a predetermined pitch interval of the conveying direction position when a defect candidate with periodicity is detected, Information on how long a defect candidate is generated in succession, information on how often a defect candidate of a group having a periodicity occurs continuously, information on the occurrence frequency of a defect candidate having a periodicity, (See Fig. 1A) by the operator's input.

Subsequently, the unit size is determined based on the set inspection condition for the image to be inspected in the defect inspection (step S110).

As shown in Fig. 3, the unit size refers to a size in which a plurality of unit areas of the same size are formed in the width direction and in the conveying direction (moving direction) Direction of the unit area. The unit size is determined based on the permissible amount of fluctuation of the position in the width direction and the carrying direction set as the inspection condition. For example, the length in the width direction x the length in the transport direction is determined as 10 mm x 10 mm.

The reason for forming the unit region in this manner is to obtain the interval between the defect candidate in the unit region and a defect candidate in another region by the interval (distance) between the unit regions. Therefore, as long as the defect candidate is within one unit area, the position of the defect candidate is handled as being unchanged at any position. By setting the unit size even if the position of the defect candidate having the periodicity fluctuates within the allowable range in the width direction and the carrying direction, the unit region can absorb, reduce, or eliminate the variation of the position of the defect candidate, Can be extracted.

Then, the inspection unit length is determined (step S120). The inspection unit length is a length in the carrying direction of an image to be subjected to one defect inspection. It is possible to obtain the result of the time-series defect inspection and to estimate the cause of the defect or the like by performing the repeated defect inspection with the inspection unit length being constant.

The inspection unit length is determined on the basis of the information of how long the defect has been set as the inspection condition, and how long the defect has occurred in succession with the periodicity. For example, the length of the glass sheet G conveyed for one hour or one day, or the length of 100 m or 1000 m is determined.

Then, the value of the target generation density of the defect candidate is determined (step S130). The defect candidate refers to an area of the dark area that is divided into dark areas in the image when the photographed image of the photographed glass plate G is binarized. In the present embodiment, the presence or absence of periodicity of fine scratches caused by the conveying rollers 11 of the glass sheet G is to be inspected. Therefore, when the density of defect candidates is high, the dark- Are formed.

Therefore, it is difficult to accurately determine the presence or absence of the periodicity of the fine scratches to be originally inspected. On the other hand, when the density of the defect candidates is reduced by lowering the first signal threshold value, which will be described later, the fine scratches to be inspected may not be distinguished as the dark portions of the defect candidates.

As a result, the target generation density of the dark region in the image is determined using the information of the occurrence frequency of the defect candidate having periodicity set as the inspection condition.

Then, a first signal threshold value is determined based on the determined target occurrence frequency (step S140). The first signal threshold value is a threshold value of the image data when the photographed image of the glass plate G is binarized. (Defect candidate) when the value of the image data is lower than the first signal threshold value. The processing unit 16 has a reference table showing the relationship between the first signal threshold value and the density of occurrence of the dark region, and the first signal threshold value is determined from the determined target generation density by referring to the reference table.

Generally, the smaller the target generation density, the lower the first signal threshold value. In the reference table, in the defect inspection unit 10, the relationship between the first signal threshold value and the density of occurrence of the dark region is obtained in advance for a photographed image of a predetermined glass plate G and stored in the memory.

Then, the frequency threshold value is determined according to the value of the target generation density of the determined defect candidate (step S150). The frequency threshold value is a threshold value used for determining whether or not the glass plate G has periodic defects in the defect inspection (step S160) described later. The frequency threshold value is a frequency threshold value used to determine that there is a periodicity in the histogram obtained by employing the interval of defect candidates on the horizontal axis and the frequency of the intervals on the vertical axis.

For example, when a histogram as shown in Fig. 4 is created in the processing section 16, it is determined whether or not the occurrence frequency of the interval B is large with respect to the frequency threshold value A determined for the interval B, do. When the occurrence frequency of the interval B is high with respect to the frequency threshold value A, it is determined that the frequency has the periodicity, and the interval B becomes the pitch interval.

The presence or absence of the periodicity is determined for each interval on the horizontal axis of the histogram. The frequency threshold value changes according to the interval of interest, and as the interval decreases, the frequency threshold value increases. It is also preferable to set the frequency threshold value to change in accordance with the value of the target generation density of the determined defect candidate other than the interval. Specifically, it is preferable to set the frequency threshold value to become smaller as the target generation density becomes smaller.

Setting the frequency threshold value as described above is intended to prevent the erroneous determination of the periodicity caused by the defect candidate due to the noise component since the defect candidate includes the defect candidate attributable to the noise component as described above .

5A and 5B show graphs showing the relationship between the intervals of defect candidates in a simulation image made only of a noise component and the occurrence frequency with respect to the intervals. Assuming that the noise component occurs randomly, it is generated in the region of 2500 mm in length and unit size of 10 mm x 10 mm on the assumption of the glass plate (G).

The vertical axes of FIGS. 5A and 5B all indicate the number of occurrences per 1 m ² (occurrence frequency). In Fig. 5A, the generation density of the noise component is changed into three kinds (50 / m ² , 100 / m ² , 200 / m ² ). According to Fig. 5A, the occurrence frequency increases as the interval becomes smaller at any generation density, and the occurrence frequency becomes higher as the density of generation of the noise component increases. Therefore, in order to prevent the misjudgment of the periodicity of the defect candidate due to the noise component, it is preferable to set the frequency threshold value high (with margin) with respect to the occurrence frequency of the vertical axis shown in Fig. 5A. For example, a value of 1.1 to 2 times the occurrence frequency of each interval may be set as the frequency threshold value.

Of course, the above value is set to a large value depending on the inspection condition. Therefore, this frequency threshold value is set to be roughly lower as the interval becomes larger and the generation density of the noise component becomes smaller. Here, the fact that it is set to be roughly low means that the frequency threshold value does not change (equivalent) even if the intervals to be noted are different.

For example, a frequency threshold of 200 mm spacing is greater than a frequency threshold of 500 mm, 1000 mm, but equal to a frequency threshold of 500 mm spacing and a frequency threshold of 1000 mm. In the present invention, when the two frequency threshold values are different, the frequency threshold value is set such that the notable interval that defines the larger frequency threshold value is smaller than the notable interval that defines the smaller frequency threshold value.

FIG. 5B is a graph showing the occurrence frequencies of the setting intervals of 1000 mm, 750 mm, and 500 mm for the density of noise components. As the density of generated noise components decreases from FIG. 5B, the frequency of occurrence in each of the setting intervals of 1000 mm, 750 mm, and 500 mm becomes smaller.

Further, as can be seen from FIG. 5B, when the density of generation of the noise component reaches a certain level or more, the occurrence frequency decreases inversely. That is, the occurrence frequency has a peak with respect to the generation density of the noise component. This is because the density of noise components is increased, so that noise components enter between the respective setting intervals of 1000 mm, 750 mm, and 500 mm, and as a result, the occurrence frequency of each setting interval of 1000 mm, 750 mm, and 500 mm is lowered.

Therefore, the target generation density of the defect candidate is determined so as to be smaller than the value of the occurrence density of the peak position in the notable interval to discriminate the presence or absence of the periodicity.

In the examples shown in Figs. 5A and 5B, the frequency threshold value is determined using the intervals of the defect candidates in the simulation image formed by the noise components and the frequency of occurrences in the intervals. However, in the present invention, The present invention is not limited to the use of the formed simulation image.

For example, assuming that noise components are randomly distributed in the region, a noise component whose position in the width direction is the same among the regions is considered as a defect candidate, the frequency of occurrence of the noise component is analytically determined, The frequency threshold may also be determined based on the frequency of occurrence.

To obtain the frequency of occurrence by analytical calculation, the frequency of occurrence is calculated using the formula. For example, in the one-dimensional region, the probability pp at which the interval of n sets is obtained is expressed by the following equation. At this time, the probability pp is calculated by changing n from 1 to N / P (P is the pitch length indicated in unit units), and the occurrence frequency with respect to the pitch P can be obtained by adding the expected value.

pp = {p ² (1-p) ^(P-1) } ⁿ _(NnP) C _n

Here, p is the probability that a noise component occurs in one unit, N is the total number of units, and P is a pitch length expressed in unit units. _{(Nn · P)} C in C _n means combination by combination.

As another mode, the defect inspection unit 10 is used for the glass plate G which knows in advance that there is no defect of the periodicity due to the conveying roller, the frequency of occurrence due to the noise component is previously determined by actual measurement, The frequency threshold value may be determined using the measurement result. The occurrence frequency of the defect candidate due to the noise component actually varies in each region of the image and there are cases where the noise component does not occur at the same occurrence density in the entire region.

For this reason, the frequency threshold value in consideration of the deviation can be determined by actual measurement using the glass plate G. From this point of view, it is preferable to determine in advance the relationship between the interval of the measurement results and the occurrence frequency of the defect candidate, and determine the frequency threshold value using this relationship. Also in this case, the frequency threshold value is set so as to increase substantially as the interval becomes smaller.

FIG. 6A shows a graph of the occurrence frequency with respect to intervals when defect inspection is performed (actual measurement) in the defect inspection unit 10. FIG. Symbols {circle over (2)} in the graph indicate the results of actual measurement using the defect inspection unit 10. [ On the other hand, the result obtained from the simulation image in which the noise component is generated by matching the generation density (average generation density) of the defect candidates as described above is indicated by the symbol ◇.

As can be seen from Fig. 6A, when the interval between defect candidates is 500 mm or less, there is a gap between the frequency of occurrence by actual measurement and the frequency of occurrence by simulation. This is considered to be due to the fact that the density of occurrence of defect candidates varies depending on the region in the image to be inspected obtained in actual measurement, even though the average occurrence density of noise components in the simulation image is equal to the average occurrence density in actual measurement. When an image of the inspection object actually obtained by actual measurement is divided into small regions and the number of defective candidates is counted to find out a probability density function that generates a defect candidate, as shown in FIG. 6B, .

Thus, by providing the same probability density function in the simulation image so as to conform to the actual probability density function, the symbol 사용한 using the simulation image as shown in Fig. 6C shows the occurrence frequency close to the actual symbol. Therefore, it is also possible to create the relationship between the interval and the occurrence frequency from the simulation image obtained by distributing the probability density function used for generation of the noise component so as to fit the actual measurement, and determine the frequency threshold value using this relationship.

As described above, the relationship between the interval and the frequency of occurrence of defect candidates can be prepared in advance by actual measurement using a glass plate G having no periodic scratches, and the frequency threshold value can be determined using this relationship.

Subsequently, defect inspection is performed (step S160). In the defect inspection, first, an image to be inspected sent from the camera 14 is cut into a region of the inspection unit length, and the image is segmented into unit sizes as shown in Fig. 3 for the image of the inspection unit length region .

The segmentation of the image is performed in accordance with the unit size determined for the image of the search object to search for a defect candidate, and a plurality of unit regions having the same unit size are formed. When a plurality of unit areas including a plurality of defect candidates are located at the same position in the width direction, it is determined that the defect candidates have the same position in the width direction, and the interval and frequency of occurrence of defect candidates to be described later are obtained.

Subsequently, the image of the unit length region is binarized using the determined signal threshold value (first signal threshold value), the regions of the plurality of arm portions are extracted as defect candidates, and the direction of conveyance from the end in the conveying direction of the image Therefore, search and detection of defect candidates are repeated. If there is a defect candidate in the unit size division, the position in the width direction and the position in the carrying direction of the representative point (the central point of the division or the vertex of the rectangular section) of the division are detected by the processing unit (Not shown).

It is also searched whether there is a defect candidate along the transport direction. When a defect candidate is detected, the position in the carrying direction of the defect candidate stored in the memory is the same as the position in the width direction stored in the memory, and the difference between the position in the carrying direction of the found defect candidate and the position in the called carrying direction is obtained, This difference becomes an interval.

Then, the count value which is set for each position in the width direction set in the unillustrated memory of the processing section 16 and which indicates the occurrence frequency determined for each interval is advanced by one. The position in the carrying direction and the position in the width direction are all expressed using the values of the representative points of the compartments divided into unit sizes. In this manner, the defect candidate is searched and detected until the entire image to be inspected is searched. Finally, using the count value stored in the memory, the position in each width direction and the occurrence frequency of defect candidates for each interval are obtained.

Subsequently, on the basis of the frequency of occurrence obtained in the processing section 16, a histogram of defect candidates showing the intervals on the horizontal axis and the occurrence frequency on the vertical axis as shown in Fig. 4 is created (step S170). Concretely, the occurrence frequency stored in the memory is accumulated with the occurrence frequency of the position in the width direction for each interval, and a histogram showing the occurrence frequency for each interval is created.

It is examined whether or not the occurrence frequency of the noteworthy interval is higher than the determined frequency threshold value (the frequency threshold value A in FIG. 4) by summarizing the occurrence frequency for each notable interval in the histogram. If the occurrence frequency is higher than the frequency threshold, it is determined that the interval corresponding to the occurrence frequency is a pitch interval created by the periodic defect, and the glass plate G has periodic defects (step S180).

This defect inspection is performed on a plurality of images of a unit of the determined inspection unit length as one unit, and the occurrence frequency in time series is generated for each inspection unit length since the defect inspection is a long shape in which the glass plate G is continuous in the carrying direction. Of course, each of the created multiple unit images is subject to defect inspection.

In this manner, defect inspection of the long glass sheet G carried on the conveyance path is performed.

As a result of such a defect inspection, it is possible to guess whether any of the conveying rollers 11 is causing scratches or the like by using the information of the pitch interval since a defect occurs at the determined pitch interval.

In the present embodiment, in addition to the step S180, it is possible to estimate the cause of occurrence of defects having periodicity in accordance with the flow shown in Fig.

First, the occurrence frequency distribution in the width direction in the determined pitch interval determined in step S180 is calculated, and the feature quantity of this distribution is calculated (step S181). As the characteristic quantities of the distribution, for example, there can be mentioned the position in the width direction of the maximum occurrence frequency and the standard deviation of the occurrence frequency in the occurrence frequency distribution in the width direction.

Since the memory of the processing section 16 stores the intervals and the frequency of occurrence at each position in the width direction, the interval can be fixed to a predetermined pitch interval to obtain the occurrence frequency distribution at the position in the width direction. 8A, 8B and 8C show three examples of occurrence distribution in the width direction at a predetermined pitch interval. The occurrence distribution in the width direction is preferably displayed on the display 18a (see FIG. 1A). In the example shown in Fig. 8A, it is a generation pattern in which the occurrence frequency is projected at a position in one width direction. In the example of FIG. 8B, occurrence frequency occurs dominantly in a certain range in the width direction, and a distribution pattern is formed in this range. In the example of FIG. 8C, the occurrence pattern is a pattern in which the occurrence frequency fluctuates in a wide range in the width direction.

The occurrence frequency distribution at the position in the width direction is classified into a plurality of defect occurrence patterns by using the feature amount such as the position in the width direction of the maximum occurrence frequency in addition to the standard deviation (variation) of the distribution.

Subsequently, a distribution of time series of occurrence frequency and occurrence density is obtained for a defect candidate determined to have a periodical defect (hereinafter referred to as a candidate defect having periodicity), and the feature of the distribution is calculated (step S182). Since the defect inspection is performed sequentially for a plurality of time series units with the inspection unit length determined as described above as an object to be inspected in one time series unit, the occurrence density (/ m ² ) of the defect candidate having periodicity and the time series distribution Can be created.

9 shows an example of the time series distribution of the occurrence density (occurrence frequency) of a specific pitch interval, the density of occurrence of all defect candidates, and the deviation (standard deviation) of occurrence frequencies in the width direction in a specific pitch interval, Is shown.

In addition, as the characteristic quantity of the distribution, a set value is set for each time-series distribution, and a continuation time in which this value exceeds the generation density is obtained. Or the correlation coefficient of each time series distribution is obtained. Or the standard deviation of each time-series distribution as the characteristic quantity of the distribution. The range of values on the ordinate of the time-series distribution in Fig. 9 differs for each legend. This time-series distribution is preferably displayed on the display 18a.

In addition, a time series distribution (time series data) of a plurality of occurrence frequencies in which at least one of a pitch interval and a width direction position to be noticed is changed with respect to a time series distribution (time series data) of occurrence frequencies of defect candidates is overlaid on the same graph Can be displayed on the screen 18a.

A method of displaying a time series distribution such as a frequency of occurrence of a defect candidate, for example, a frequency of occurrence of a defect candidate, and a time series display method of occurrence frequency of a defect candidate will be described later.

Next, out of the plurality of pitch intervals judged to have periodicity, the occurrence frequency distribution in the width direction of defect candidates of different pitch intervals is obtained for the defect candidate having the periodicity noticed in step S182, and the defects in the pitch interval A time-series distribution corresponding to the candidate's Fig. 9 is created and displayed on the display 18a. At this time, the correlations between the obtained frequency distribution in the width direction and the frequency distribution in the width direction obtained in steps S181 and S182 of the time series distribution and the time series distribution are evaluated (step S183).

Specifically, a correlation coefficient is obtained for each of the occurrence frequency distribution in the width direction and the time series distribution. In addition, the characteristic quantities of the occurrence frequency distribution in the width direction and the characteristic quantities of the time series distribution in the defect candidates of different pitch intervals are calculated and compared with the characteristic quantities calculated in steps S181 and S182.

Further, a plurality of feature quantities? (Image size, shape, image data value, etc. of defect candidates) of an image of a plurality of defect candidates judged to have periodicity are calculated, and the average value and standard deviation of the feature amount? Is calculated (step S184).

The characteristic quantities? And the characteristic quantities? Thus calculated are compared with the characteristic quantities previously stored in the database as data in the past together with the characteristic quantities calculated in the steps S181 and S182, and as a result of the comparison, The cause of the defect registered in association with the feature amount is called and estimated as the cause of the occurrence of the periodical defect (step S185). It is determined in step S183 that the defect candidate of another pitch interval evaluated as having a high correlation is caused by the same cause as the estimated cause.

For example, since two foreign substances of the same particle diameter and the same material are attached to different circumferential positions at the same time on the surface of the same conveying roller 11 among the plural conveying rollers 11, when a defect occurs, Two pitch distances are possible. However, by calculating and comparing the occurrence frequency distribution in the width direction or the time series distribution of defect candidates, they have the same occurrence frequency distribution and have the same time series distribution.

Therefore, it is possible to assume that the defect candidates in the other pitch intervals evaluated as having high correlation in step S183 are caused by the same conveying roller together with the defect candidates to be subjected in step S181.

The comparison (matching or non-matching) of the characteristic quantities may be performed by setting conditional branching for each characteristic quantity, by using the Mahalanobis space and the Mahalanobis distance for the characteristic quantities, or by constructing the neural network .

Assuming that two pitch intervals are possible between the defect candidates, in the present invention, in order to determine the interval of defect candidates, in addition to finding the interval between adjacent defect candidates (defect candidates before one defect) The interval between the defect candidates adjacent to the defect candidates (two defect candidates before) is obtained, and the interval between the defect candidates adjacent to the two defect candidates before the defect candidates (defective candidates before three defect candidates) is obtained. , And (N-1) (N is an integer equal to or greater than 4) defective candidates adjacent to the defective candidates. For example, in the defect inspection at the step S160, the intervals are obtained with respect to the adjacent defect candidates. If two or more pitch intervals occur at the step S181, the adjacent defect candidates (defective candidates one before) It is also possible to use the information of the interval stored in the memory to obtain the interval between the defect candidates adjacent to the defect candidates before the plurality of defect candidates.

In this case, the upper limit of the interval to be obtained is the maximum circumferential length of the conveying rollers 11 as a limit. At this time, a frequency of occurrences of the intervals of 1 to N defective candidates is obtained, and a histogram is generated. The frequency threshold is separately set to determine whether or not the defects have periodic defects When the circumferential length of the conveying roller 11 is 1000 mm and the occurrence frequency of the defects candidates is 300 mm and 700 mm exceeds the frequency threshold value at the same time, the occurrence frequency of 1000 mm is 300 mm and 700 mm Since the sum of the frequencies of occurrence is added, an occurrence frequency exceeding the set frequency threshold value appears. Thus, it is possible to more accurately estimate that the defect candidate is generated due to the same conveying roller.

Fig. 10 is a view for explaining an example flow specifically performed in the defect inspection shown in Fig. 2, and may be performed as follows.

As described above in step S160 in Fig. 2, in the defect inspection, the image to be inspected sent from the camera 14 is divided into the inspection unit length area, and the image of the unit length area is shown in Fig. 3 The image is partitioned into unit sizes as described (step S161). Subsequently, the image of the unit length region is binarized using the determined first signal threshold value, and the defect candidate is searched along the carrying direction from the end in the carrying direction of the image, And the position in the width direction of the detected defect candidate is extracted (step S162).

At this time, if a defect candidate is detected, the attribute of the defect candidate is discriminated (step S163). Whether or not the values of all the image signals of the defect candidate are smaller than the predetermined value, the area of the image of the defect candidate, and the shape of the defect candidate satisfy the set condition, for example, can be cited. Whether or not the defect candidate is on the back surface of the glass plate G (on the side of the conveying roller 11) can be cited. The attribute of whether the defect candidate is on the back surface or on the surface is determined using the reflection image obtained by the defect inspection unit 26. [

In other words, since the reflected light is reflected by the back surface of the glass plate G and is photographed by the camera 24, as described above, there is no deviation between the actual image and the mirror image in the reflected image. On the other hand, the defects on the surface have a real image and a mirror image, and the position shift amount agrees with a constant value determined according to the thickness of the glass plate G. Using this, it is determined whether or not a defect candidate of the reflected image at the position corresponding to the defect candidate is on the back side. The attribute is determined using the result of the determination.

Other types of defects can also be mentioned as attributes. The type of defect can be identified by the shape of the image of the corresponding defect candidate obtained from the reflected image.

Then, the position in the conveying direction and the position in the width direction of the section to which the defect candidate corresponding to the attribute to be obtained belongs are extracted (step S164), and the position in the width direction and the conveying direction of the representative point of this section, Is stored in a memory (not shown) of the processing unit 16. [ In addition, it is searched whether there is a defective candidate of the attribute to be obtained along the carrying direction.

At this time, the correlation between the image area of the defect candidate detected as the attribute to be obtained and the image area of the defect candidate already stored in the memory, such as the correlation coefficient of the size and the shape of the image area, (Step S165). For example, when evaluating using a correlation coefficient, when the value of the correlation coefficient exceeds a predetermined value, it is determined that the degree of similarity is high. The detected defect candidates to be subjected to the similarity evaluation and the defect candidates stored in the memory may be classified into defect candidates generated at a predetermined interval or defective candidates detected and defective candidates detected and stored before the detection A defect candidate averaged as a population, or a detected defect candidate and a predetermined defect candidate model. The object of the correlation may be a characteristic amount of a defect candidate, image data of a defect candidate, or an image characteristic amount (characteristic amount of a shape, etc.) of a defect candidate.

The evaluation of the degree of similarity may be performed by using the Mahalanobis space and the Mahalanobis distance with respect to the characteristic quantity in addition to the correlation, or by constructing a neural network.

The difference is found between the position in the carrying direction of the defect candidate which is judged to have a high degree of similarity to be stored in the memory and the position in the carrying direction of the defect candidate at the same position in the width direction called. The count value indicating the occurrence frequency determined for each position and interval in the width direction is advanced by one (step S166).

It is judged whether or not the search and detection of such a defect candidate is made in the whole image of the object to be inspected (step S167), and if it is denied, the process returns to step S162. If the determination is affirmative, the process proceeds to step S170.

Thus, when detecting a defect candidate, the condition can be strictly set using the attribute of the defect candidate to be obtained and the similarity of the image region of the defect candidate, and the defect candidate to be detected can be limited. Of course, either the attribute of the defect candidate or the similarity of the image region of the defect candidate may be set as a condition.

Further, as a step subsequent to step S180 shown in Fig. 2, the cause of the defect may be eliminated in accordance with the flow shown in Fig.

If it is determined in step S180 that there is a candidate defect having periodicity, the occurrence frequency distribution and the time series distribution in the width direction in the pitch interval of the defect candidate are determined as described above, and the characteristics of the occurrence pattern are determined by the obtained occurrence frequency distribution and time series By extracting from the feature quantities of the distribution, defect candidates with periodicity are evaluated. Alternatively, the defect type of the defect candidate is identified (step S191). The type of defect refers to the type of defect candidate determined by using the value of the image data indicating the degree of light intensity or the shape of the defect candidate in the reflected image obtained by the defect inspection unit 26, G, or an adherence to the surface of the glass plate G, and the like.

Next, in accordance with the evaluation result or the identification result, it is estimated that a defect occurs by which conveying roller (step S192). For example, a conveying roller having a circumferential length corresponding to the pitch interval is presumed to be the cause of the occurrence. In addition, it is estimated by which type of defect the defect has occurred due to which conveying roller. These estimations are made possible by establishing a database that correlates the cause of the defect in advance, the type of defect and the pattern of occurrence.

Subsequently, on the basis of the estimation of the cause of occurrence, the cause of the occurrence is removed (step S193). For example, when the cause of the occurrence is a specific conveying roller, the control device and the driving device of the conveying path are so arranged that the specific conveying roller moves from the conveying path and automatically separates from the conveying path, . Alternatively, a specific conveying roller is automatically repaired. As an automatic repair, an example in which the protective film on the surface of the conveying roller is thickened can be cited.

The occurrence cause thus obtained is associated with the occurrence pattern or type of defect and is additionally registered in the database (step S194). Of course, if the estimated cause of the error is wrong, the correction is made by the input of the operator and is registered in the database. This database is used for estimation of the cause of occurrence in step S182.

Alternatively, in the cutting process of the glass plate G after the transfer, the cutter is instructed to avoid the above-mentioned widthwise position of the defect determined to have periodic defects, and the glass plate G is cut to a predetermined size, .

Further, in the present invention, it is possible to easily detect the presence of periodic defects by using the pitch interval obtained by the defect inspection described above, the data of the occurrence frequency distribution in the width direction of the defect candidate, and the following method.

That is, the processing unit 16 determines a region of interest including a position in the width direction where the defect candidate having the pitch interval is located. In this region of interest, detailed defect candidates are extracted from the end of the image using the second signal threshold value among the images obtained by photographing with the camera 14. [

A detailed defect candidate is searched for and a detailed defect candidate obtained by searching is searched for a search region centered on a position spaced by a pitch interval in the carrying direction from the position of the detailed defect candidate obtained by the extraction, And the degree of similarity of images between the extracted detailed defect candidates is evaluated. According to the result of the evaluation of the degree of similarity, it is determined that there is a periodic defect in the conveying direction in the region of interest. The second signal threshold value may be set to a lower value than the first signal threshold value determined in step S150 in FIG.

Fig. 12 shows an example of the flow of the inspection. First, an attention area AX including a position in the width direction determined as a periodicity of the defect candidate, which is obtained by defect inspection of the flow shown in Fig. 2, is determined as shown in Fig. 13A, A detailed defect candidate is detected from the end of the image using the second signal threshold value (step S195). In Fig. 13B, the defect candidate D ₁ is detected as the detected detailed defect candidate.

Subsequently, based on the position of the detected defect candidate D _{1 in} the carrying direction, as shown in FIG. 13B, a search range in a fixed range centered on a position spaced apart from the position by the pitch interval and in the carrying direction AY) is set (step S196).

In the set search area AY, the detail defect candidate is extracted using the second signal threshold value (step S197).

Subsequently, the attribute of the detail defect candidate and the detail defect candidate detected in step S195 are respectively evaluated (step S198). As an attribute, for example, a defect occurrence position (front side or back side) is discriminated. This discrimination is made based on the image data of the defect inspection unit 26 supplied to the processing section 16 as described above and the difference between the actual defect position of the defect candidate located on the surface and the mirror image ≪ / RTI >

Then, if the attributes match, it is determined that a periodic defect exists in the region of the glass plate G to be inspected (Step S199).

The attention area AX is determined using the pitch interval acquired in this way, and an accurate defect candidate is searched in this area to determine whether there is a defect in the periodicity.

This inspection method can also be applied to a sheet glass plate G cut to a predetermined size in addition to the long-shaped glass plate G being conveyed. Particularly, in the case of a sheet glass plate (G), it is possible to discriminate whether there is a periodic defect individually by using the pitch interval.

In this embodiment, the attention area is set with respect to the pitch interval detected by the defect inspection, but the present invention is not limited to this. In the present invention, there is a case where the circumferential length of the conveying roller can be limited, or a circumferential length of a specific conveying roller immediately after maintenance is paid attention, and therefore, it is also possible to set the note area based on the circumferential length of the conveying roller . In the present invention, in the case of a roller with a shoulder or a stepped roller, the relationship between the widthwise position and the conveying roller may be specified, and it is also possible to set the area of interest based on this.

In the present embodiment, as shown in Fig. 9, a graph in which a plurality of time series distributions such as a time series distribution of occurrence frequencies of defect candidates are overwritten on one screen is displayed on the display 18a. However, It is not limited.

In the present invention, the time series distribution (time series data) of the occurrence frequency of defect candidates may be displayed on the display 18a as a two-dimensional density image showing the occurrence frequency as density as shown in Fig. 14A. The two-dimensional density image is obtained by taking a position on the one axis, for example, the time on the ordinate, and the position on the other axis, for example, the abscissa in the widthwise direction and determining a target pitch corresponding to each inspection condition (for example, Dimensional density image showing the frequency of occurrence of defect candidates in terms of color or density (light and shade) at the peripheral edge of the conveying roller (corresponding to the circumferential length of the conveying roller). Further, instead of the two-dimensional density image, as shown in Fig. 14B, a three-dimensional graph showing the time as one axis and the height in the direction orthogonal to the two-dimensional coordinate as the other axis And display on the display 18a.

14A shows that frequency of occurrence of defect candidates is high at time and width positions indicated by the black spots of the two-dimensional density image shown in Fig. 14A. This black spot corresponds to the frequency peak of the three-dimensional graph shown in Fig. 14B .

The time series distribution of the frequency of occurrence of the defect candidates is changed to a pitch on the other axis (horizontal axis) as shown in Fig. 15A, and the display 18a is displayed as a two-dimensional density image, As shown in FIG. The two-dimensional density image is obtained by taking a time on one axis (ordinate axis) and a pitch on the other axis (abscissa axis), and calculating a position in a notable width direction corresponding to each inspection condition (for example, Dimensional density image expressing the occurrence frequency in a color or a density (light and shade) at a position where the image is easily generated. Further, instead of the two-dimensional density image, as shown in Fig. 15B, a three-dimensional graph showing the time as one axis and the height as a height orthogonal to the two-dimensional coordinate having the other axis 18a on the screen.

As described above, in the present embodiment, there are three time, width direction positions and pitches as parameters (variables) of occurrence frequency of defect candidates. Therefore, as shown in Figs. 14A to 15B, instead of displaying the time-series distribution of the frequency of occurrence of defect candidates as a two-dimensional density image or a three-dimensional graph of occurrence frequencies at a noticeable pitch or width position, A three-dimensional graph of the occurrence frequency in a predetermined unit time or a two-dimensional density image may be displayed in a time-series manner as shown in Figs. 16A to 16C and Figs. 17A to 17C.

16A to 16C are three-dimensional graphs each showing a position in the width direction on one axis and a pitch on the other axis and showing the occurrence frequency of defect candidates in a unit time corresponding to each inspection condition as height; The occurrence frequency data of each day is displayed in a time series.

Of course, instead of the three-dimensional graphs shown in Figs. 16A to 16C, as shown in Figs. 17A to 17C, the pitch is taken on one axis (vertical axis) and the position on the other axis Dimensional density image representing the frequency of the unit time corresponding to each inspection condition as a color or a density (darkness) may be displayed in a time series.

Here, Figs. 16A to 16C and Figs. 17A to 17C show the occurrence frequency data of each day in a time-series manner, but they may be continuously displayed on the display 18a and displayed as a moving image. The time interval for generating the occurrence frequency data is not particularly limited and may be longer or shorter than one day, or may be continuous as a so-called moving picture.

As described above, in this embodiment, by determining the pitch interval of the defect candidates and the position in the width direction of the defect candidate having this pitch interval by the defect inspection shown in Fig. 2, as shown in Figs. 7, 10, 11 and 12 It is possible to effectively detect a defect having periodicity by adding a flow as shown in Fig. It is also possible to estimate the cause of the defect.

The above-described defect inspection method can be suitably used for a manufacturing method of a glass plate (G) or the like. That is, the defect inspection method is used to carry out a defect inspection for carrying a plate-like object such as a glass plate G, and the cause of the plate-shaped object on the conveying path is estimated according to the inspection result. The estimation result is preferably displayed on the display 18a (see FIG. 1A).

Alternatively, in accordance with the estimation result, it is possible to take measures against the cause of the occurrence on the conveyance path. For example, when a defect candidate having periodicity is formed by a foreign substance attached to a conveying roller, the conveying roller is configured to be detached from the conveying path. Alternatively, the conveying rollers causing defects such as scratches to be generated in the glass may be repaired or replaced with other conveying rollers. Furthermore, the glass plate G may be cut to a predetermined size so as to avoid the widthwise position of the defect determined to have a periodic defect in the cutting process of the glass plate G after the transfer.

In this embodiment, feedback is performed to eliminate glass defects by the removal, maintenance, and replacement of the conveying rollers by using the defect inspection according to the present invention. However, the present invention is not limited to this, Feedback is also possible. Specifically, in the production of glass, the contamination in the gas does not have a periodicity but adheres to the front and back of the glass. The liquid or semi-liquid object such as tin (dross) adhering to the lower surface of the glass from the float bath, It may be transferred to the roller 11 (see FIG. 1A) and re-transferred to the glass plate G in some cases.

In this case, the contaminated area Y shown in FIG. 1B is periodically generated on the lower surface of the glass plate G. Therefore, it is possible to evaluate the manufacturing environment of glass such as a float bath as a refreshment of a defect point (contamination) having a periodicity detected on the lower surface of the glass plate G. [

By applying the defect inspection method and apparatus of the present invention to defect inspection resulting from a manufacturing environment such as a dross or the like in this manner, it is possible to evaluate the manufacturing environment of glass or the like, and feedback the evaluation result to the production of glass.

The image data processing method for defect inspection described above can be executed on a computer by executing a program.

For example, a program for processing image data for defect inspection of the present invention has a procedure for causing a computer, specifically, a CPU, to perform each step of the method for processing image data for defect inspection described above. A program including these procedures may be configured as one or a plurality of program modules.

The image data processing program for defect inspection including the procedures executed by these computers may be stored in a memory (storage device) of a computer or a server, or may be stored in a recording medium, (CPU) or other computer and read from the memory or recording medium and executed. Accordingly, the present invention may be a computer-readable memory or a recording medium storing a program for processing image data for defect inspection for causing a computer to execute a method of processing image data for defect inspection of the above-mentioned form 14.

As described above, the apparatus and method for processing image data for defect inspection of the present invention, the defect inspection apparatus and the defect inspection method each using the same, the method of manufacturing the plate-like body using the defect inspection apparatus, and the computer The present invention is not limited to the above-described embodiments and examples, and it goes without saying that various improvements and modifications may be made without departing from the gist of the present invention.

Although the present invention has been described in detail with reference to specific embodiments, it is apparent to those skilled in the art that various changes and modifications can be made without departing from the spirit and scope of the present invention. The present application is based on Japanese Patent Application (Japanese Patent Application No. 2008-187450) filed on July 18, 2008, the contents of which are incorporated herein by reference.

1: defect inspection device
10, 26: defect inspection unit
11: conveying roller
12, 22: Light source
14, 24: camera
16:
18: Output system
18a: Display
18b: Printer
20: input operation system

Claims (28)

An apparatus for processing image data for defect inspection for inspecting a defect existing in a plate-like body using an image obtained by photographing with the camera while relatively moving the plate-like body in a predetermined direction with the camera,
Extracting a plurality of defect candidates from the image by using a first signal threshold value and extracting a defect candidate having the same position in the width direction orthogonal to the moving direction as the predetermined direction among the extracted plurality of defect candidates, And finding the distance between the position of the defect candidate detected by the search in the moving direction of the plate-like body and the position of the detected defect candidate in the moving direction of the adjacent defect candidate in the moving direction And the frequency of occurrence in the plurality of intervals is obtained. When the frequency of occurrence of the interval of interest exceeds the frequency threshold to be set, the plate-like body is subjected to periodic defects in the moving direction And has a processing section for judging that it has,
Wherein the frequency threshold used by the processor is determined according to the interval of interest and when the two frequency thresholds are set as different values, the noticed interval that defines the larger frequency threshold is the smaller frequency threshold Wherein the frequency threshold value is set so that the frequency threshold value is smaller than the noticed interval for defining the value. 2. The apparatus according to claim 1, wherein the processing section has a reference table indicating a relationship between the density of occurrence of the defect candidates and the first signal threshold value, and uses the reference table so that the density of occurrence of defect candidates becomes a set target density of occurrence To set the first signal threshold value,
Wherein the frequency threshold value is a value varying in accordance with a value of the target generation density other than the noted interval. 2. The apparatus of claim 1, wherein the frequency threshold value used in the processing unit is a noise candidate that is assumed to be a random noise component distributed randomly in a region, The occurrence frequency of the noise component with respect to the interval is obtained analytically or the frequency component of the noise component with respect to the interval is determined as the defect candidate image of the noise component in the simulation image formed of the noise component, Wherein the predetermined frequency is determined based on the frequency of occurrence. 4. The processing apparatus according to claim 3, wherein the simulation image is generated so that the density of generation of noise components in an image varies depending on an area of an image. 2. The image processing apparatus according to claim 1, wherein the processing unit divides a search target image for searching for a defect candidate into a plurality of unit regions of the same size by dividing a plurality of images in the width direction and the movement direction, And when the plurality of unit areas are at the same position in the width direction, the defect candidates are determined to have the same position in the width direction, and the spacing and the occurrence frequency are found. The apparatus of claim 1, wherein the processing unit further comprises: a width direction occurrence frequency distribution indicating a distribution of the occurrence frequency at the position in the width direction, with respect to the noted interval of the defect candidate determined to have a periodical result And classifies the defect occurrence pattern using the deviation of the occurrence frequency along the width direction in the width direction occurrence frequency distribution. 2. The apparatus according to claim 1, wherein the plate-like body is a long shape continuous in the moving direction,
Wherein the processing section divides the plate-shaped body into regions of a plate-shaped body having a predetermined length, and subjects the images of the region to be inspected in a time series unit to perform the determination for a plurality of time series units. The recording apparatus according to claim 7, wherein the processing unit records the occurrence frequency distribution of the intervals determined by the intervals and the positions in the width direction, with respect to the plurality of time series units, A position in the width direction is determined to determine the frequency of occurrence, and the frequency of occurrence is indicated as time-series data, thereby displaying defect occurrence information on the screen. 9. The apparatus according to claim 8, wherein the processing unit overwrites time series data of a plurality of occurrence frequencies in which at least one of the noted intervals and the position in the width direction is changed, with respect to the time series data of the occurrence frequency, . 2. The apparatus according to claim 1, wherein, when the defect candidate having the same position in the width direction is searched for and detected in the movement direction, the processing section uses the adjacent defect candidate as a defect candidate of the previous defect candidate, A plurality of intervals are obtained by repeating obtaining an interval in the moving direction between a plurality of defect candidates and a plurality of defect candidates, and the frequency of occurrence in the plurality of intervals is obtained, and the frequency threshold value , The plate-like body is judged to have a periodic defect in the moving direction. The method according to claim 1, wherein, when the interval determined to have periodic defects in the moving direction is a pitch interval,
The processing unit further determines an attention area including the position in the width direction in which the defect candidate having the pitch interval is located and selects a detail defect from the end of the image using the second signal threshold, Determining a search area centered on a position spaced by the pitch interval in the movement direction from the position of the detailed defect candidate obtained by the extraction and using the second signal threshold value for the search area The detail defect candidates are searched and searched to evaluate the attributes of the detected detailed defect candidates and the extracted detailed defect candidates, respectively. Based on the evaluation results, the attention region is classified into the periodic detailed defect candidates Or not. The apparatus according to claim 1, wherein, when the interval determined as having a periodic defect in the moving direction is a pitch interval of the intervals, the processing unit further includes: And a detailed defect candidate is extracted from the end of the image by using the second signal threshold value among the images of the region of interest. Then, from the position of the detail defect candidate obtained by the extraction, A search area centered on a position spaced apart by a distance corresponding to the circumferential length of the conveying roller that carries the plate-like body is determined and a detail defect candidate is searched for the search area using the second signal threshold value , Evaluates the attributes of the detail defect candidates detected and detected and the detail defect candidates extracted and obtained, Group attention area processing apparatus to determine whether or not including a periodic defect candidate in detail in the movement direction. 2. The apparatus according to claim 1, wherein the processing unit searches for a defect candidate having the same position in the width direction in the movement direction and detects the defect candidate, and when the interval is determined, the attribute of the detected defect candidate, Evaluating the degree of similarity between the candidate and the candidate, and determining the interval when at least one of the attribute and the degree of similarity satisfies a set condition. A defect inspection apparatus for inspecting a defect existing in a plate-
A light source for projecting light onto the surface of the plate-
A camera for photographing an image of a plate-like body projected to the light source while moving relative to the plate-like body together with the light source;
An image forming apparatus comprising the processing apparatus according to claim 1,
Wherein the processing unit of the processing apparatus extracts the plurality of defect candidates using the first signal threshold value among the images obtained by photographing with the camera and extracts, from among the extracted plurality of defect candidates, And a defective candidate having the same position in the width direction orthogonal to the moving direction, which is a direction of relative movement between the upper bodies, in the moving direction. Shaped continuous body to be conveyed by a conveying roller,
A defect inspection apparatus according to claim 14, wherein the plate-shaped body is inspected while moving,
A conveying roller for generating a defect on the movement path of the plate-shaped body is specified according to the inspection result,
And the specified conveying roller is removed or repaired. Shaped continuous body to be conveyed by a conveying roller,
A defect inspection apparatus according to claim 14, wherein the plate-shaped body is inspected while moving,
Wherein said plate-shaped body is cut off and taken out from said widthwise position of said defect determined to have said periodic defect. A method of processing image data for defect inspection for inspecting a defect existing in the plate-like body using an image obtained by photographing with the camera while relatively moving the plate-like body in a predetermined direction with the camera,
Extracting a plurality of defect candidates from the image obtained by photographing using the first signal threshold value,
A defective candidate having the same position in the width direction orthogonal to the moving direction as the predetermined direction is searched in the moving direction among the plurality of extracted defective candidates and the position in the moving direction of the defective candidate detected by the search, Obtaining a plurality of intervals by repeating obtaining the distance between the candidates and positions in the moving direction of adjacent defect candidates in the moving direction,
The occurrence frequency in the plurality of intervals is obtained,
When the frequency of occurrence of the noticed interval exceeds a set frequency threshold value, the plate-like body is determined to have a periodic defect in the moving direction,
Wherein the frequency threshold value is determined according to the noted interval and wherein when the two frequency threshold values are different, the noticed interval that sets the larger frequency threshold value is set to the notable interval that defines the smaller frequency threshold value Wherein the frequency threshold value is set so as to be smaller than the frequency threshold value. 18. The method according to claim 17, wherein, prior to the determination,
In the step of setting the inspection condition, the first signal threshold value is set using the reference table so that the density of occurrence of the defect candidate becomes the set target density,
Wherein the frequency threshold value is a value that varies in accordance with the value of the target generation density in addition to the noted interval. The method according to claim 17, wherein the frequency threshold value is obtained by using an image of the noise component in a simulation image formed by a noise component as a defect candidate, determining a frequency of occurrence of the noise component with respect to the interval, Based on the number of the processing steps. The processing method according to claim 19, wherein the simulation image is generated so that the density of generation of noise components in the image is different depending on the region of the image. The method as claimed in claim 17, wherein, when the interval determined to have periodic defect candidates in the moving direction is a pitch interval, the defect candidate having the pitch interval Determining a region of interest which includes the position in the width direction in which the position is located,
Among the images of the region of interest, a detailed defect candidate is extracted from the end portion of the image by using the second signal threshold value,
Determining a search area centered on a position spaced apart from the position of the detailed defect candidate by the pitch interval in the movement direction,
Searching the detailed defect candidate for the search area using the second signal threshold value,
Evaluates the attributes of the detailed defect candidates detected by the search and the detailed defect candidates obtained by the extraction,
And judging whether or not the attention area includes a periodic defect candidate in the movement direction according to the evaluation result. The method according to claim 21, wherein the image used for discriminating the periodic defect candidates of the attention area is an image of a plate after cutting the plate-shaped body to a predetermined size. The method as claimed in claim 17, wherein, when the interval determined to have periodic defect candidates in the moving direction is a pitch interval, the defect candidate having the pitch interval Determining a region of interest which includes the position in the width direction in which the position is located,
Among the images of the region of interest, a detailed defect candidate is extracted from the end portion of the image by using the second signal threshold value,
Form a search area centered on a position spaced apart by a distance corresponding to a circumferential length of a conveying roller which conveys the plate-shaped body in the moving direction from the position of the extracted detailed defect candidate,
Searching the detailed defect candidate for the search area using the second signal threshold value,
Evaluates the attributes of the detailed defect candidates detected by the search and the detailed defect candidates obtained by the extraction,
And judging whether or not the attention area includes a periodic defect candidate in the movement direction according to the evaluation result. The method according to claim 17, further comprising: searching for and searching for a defect candidate having the same position in the width direction in the movement direction, and calculating the distance between the attribute of the detected defect candidate or the predetermined defect candidate of the defect candidate , And when the condition of at least one of the attribute and the degree of similarity satisfies a set condition, the process determines the interval. A defect inspection method for inspecting a defect existing in a plate-
The light is projected onto the surface of the plate-shaped body, the image of the projected plate-shaped body is photographed while the plate-shaped body is relatively moved,
A defect inspection method according to claim 17, wherein the processing method according to claim 17 is performed using the image obtained by photographing. Shaped continuous body to be conveyed by a conveying roller,
26. A defect inspection method according to claim 25, wherein the plate-
A conveying roller for generating a defect on the movement path of the plate-shaped body is specified according to the inspection result,
And the specified conveying roller is removed or repaired. Shaped continuous body to be conveyed by a conveying roller,
26. A defect inspection method according to claim 25, wherein the plate-
Wherein said plate-shaped body is cut off and taken out from said widthwise position of said defect determined to have said periodic defect. A computer-readable recording medium storing a computer-executable program for executing the method of processing image data for defect inspection according to claim 17.

Images