JP2004226212A - Method and device for detecting pinhole defect and contamination defect on transparent object - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To eliminate setting for a dot position by type of glass, and to enhance thereby versatility for a dot portion. <P>SOLUTION: An image processing means 15 processes an image of a transparent object 3 imaged by an image pick-up means 14, and detects a defect of the transparent object 3 containing a dotlike printing portion in its one portion. An original image of the transparent object 3 is formed into blocks in every of m×n picture elements, and a value of the maximum brightness-the minimum brightness within the each block serves as brightness of the block. The blocks of a prescribed brightness level or more are extracted to serve as defect candidates, and the candidates including the blocks of a prescribed number or more are excepted from the candidates. A kind of the defect in the defect candidate is determined based on the brightness of the original image in a periphery of the block of the defect candidate. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

【００５３】
このようにすることにより、検査対象である透明体の一部にドット状のプリント部分が含まれており、設計上のピンホールおよび／または設計上のベタ以外のプリントが含まれている場合であっても、ピンホールおよび汚れ等の欠点を精度良く検出することができる。
【００５４】
（設定値Ｂの設定方法）
以下、ガラス板および欠点の一例を用いて、本発明の設定値Ｂを設定する方法について具体的に説明する。ここで、図１６は、検査対象となるガラス板の一例を説明する概略構成図であり、図１７〜１９は、欠点のサイズと欠点が含まれるブロック数との関係を説明する概念図である。
【００５５】
まず、一般的なガラス板および欠点の仕様について説明する。図１６において示すガラス板は、短辺２００ｍｍの二等辺三角形であり、５０ｍｍのマスキング幅を有する。
【００５６】
したがって、
ガラス板の周囲長≒６８３ｍｍであり、
マスキング内見切り長さ＝１００＋１００＋１００√２≒３４１ｍｍ
となる。
【００５７】
一方、
ドットプリントのドット径 ＜ ３ｍｍ
ピンホールおよび汚れ欠点の直径 ０．８φｍｍ〜５．０φｍｍ
とする。
【００５８】
上述したように、
ｍまたはｎ × 分解能 ＞ ３ｍｍ（最大ドットの直径）
になるように、ｍ，ｎが設定される。
ｍまたはｎ × 分解能は、すなわちブロックの一辺の長さ（単位ｍｍ）であるため、これは、
ブロックの一辺の長さ ＞ ３ｍｍ
を意味する。
【００５９】
上述したように、検出したい欠点のサイズは、０．８φｍｍ〜５．０φｍｍである。また、ブロックの一辺の長さが３ｍｍ以上であるため、欠点のサイズが０．８φｍｍの場合には、φ ＜ Ｍ かつ φ ＜ Ｎとなり、図１７に示すように、欠点が含まれるブロックの数は、１〜４ブロック（最大４ブロック）となる。また、欠点のサイズが５．０φｍｍの場合には、φ ≧ Ｍ かつ φ ≧ Ｎとなり、図１８に示すように、欠点が含まれるブロックの数は、４〜９ブロック（最大９ブロック）となる。
【００６０】
一方、上述したように、マスキング内見切りの周囲長は約３４１ｍｍなので、ブロック数は、
Ｇ ＝ ３４１／３ ≒ １１４ブロックとなる。
【００６１】
このように、欠点のブロック数は１〜９、マスキング内見切りエッジは１１４ブロックと、その差は大きい。したがって、設定値Ｂは両者の間であればよく（すなわち、９＜設定値Ｂ＜１１４）、例えば５０という値が適当である。
【００６２】
さらに説明すると、図１７の例では、４＜設定値Ｂ＜１１４とすればよく、これは、上述した、
（１）φ ＜ Ｍ かつ φ ＜ Ｎの場合 ４ブロック＜ 設定値Ｂ ＜ Ｇ
の一例を示すものである。
【００６３】
また、図１８の例では、９＜設定値Ｂ＜１１４とすればよく、これは、上述した、
（４）φ ≧ Ｍ かつ φ ≧ Ｎの場合 ９ブロック＜ 設定値Ｂ ＜ Ｇ
の一例を示すものである。
【００６４】
さらに、ブロックの辺の長さ、Ｍ，ＮがＭ≠Ｎでない場合について説明する。図１９に示すように、ＭまたはＮのいずれか一方が欠点の直径φ以上の場合（図１９では、Ｎ＝６ｍｍ，φ＝５ｍｍ，Ｎ≧φ）、欠点が含まれるブロックの数は、２〜６ブロック（最大６ブロック）となる。したがって、設定値Ｂは、６＜設定値Ｂ＜１１４であればよい。これは、上述した、
（２）φ ≧ Ｍ かつ φ ＜ Ｎの場合 ６ブロック＜ 設定値Ｂ ＜ Ｇ
（３）φ ＜ Ｍ かつ φ ≧ Ｎの場合 ６ブロック＜ 設定値Ｂ ＜ Ｇ
の一例を示すものである。
【００６５】
以上の説明から、設定値Ｂの好適な範囲は、
４〜９ブロック ＜ 設定値Ｂ ＜ Ｇ
とすることができる。
【００６６】
このＧは、図面または実測から得たマスキング内見切りの周囲長と、分解能と、ｍおよびｎと、から求めることができる。具体的には、
Ｇ＝マスキング内見切りの周囲長／ブロックの一辺の長さ
＝マスキング内見切りの周囲長／（ｍまたはｎ × 分解能）
によって求めることができる。
【００６７】
また、Ｇは整数値なので、本発明においてはＧ≧１１であれば理論上検査が可能となる。実際には測定のばらつきがあるので、これを考慮して、Ｇは９よりもある程度（例えば１０ブロック程度）大きいことが望ましい。
【００６８】
なお、例えば、自動車で実際に使用されている一般的な最小のガラス板には、図１６に示すようなものがある。このガラス板の寸法は、上述したように短辺２００×２００、マスキング幅５０ｍｍである。このガラス板ではマスキング内見切りの周囲長が１１４ブロック（Ｇ＝１１４ブロック）であり、欠点のブロック数との差が十分に大きいので、容易に検査をすることができる。すなわち、自動車のガラス板を例にした場合には、マスキング部分の寸法が実用的な範囲で最小の場合であっても、マスキング部分を欠点から区別し、マスキング部分を非欠点として確実に除外することができる。
【００６９】
このようにすることにより、透明体の画像にドット状のプリント部分が含まれる場合であっても、ドット部分を欠点候補から自動的に外すことができる。
【００７０】
なお、上記の説明においては、透射光を用いて検査対象物の欠陥を検出する例を説明したが、本発明は反射光を用いて検査対象物の欠陥を検出する場合にも適用できる。そして、自動車用ガラス等のガラス板を検査対象として本発明を説明してきたが、本発明はこれに限られるものではない。ガラス板の他に、樹脂等の他の透明体についても本発明を適用できる。
【００７１】
また、本発明が適用可能な透明体は、必ずしも平板に限られるものではなく、パネルなどのように、ゆるやかな曲率を有している板でもよい。また、検査対象としては、透光性を有していれば半透明な板にも適用可能である。
【００７２】
【発明の効果】
以上説明したように、本発明によれば、ガラスの品種によらずドット部分を自動的に検査対象から外すため、品種ごとのドット位置の設定を不要とすることができ、プリントのばらつき、搬送のばらつきに自動的に対応することができる。
【００７３】
また、圧縮画像での判定を行うため、処理時間を１／（ｍ×ｎ）に短縮することができる。
【図面の簡単な説明】
【図１】従来の欠点検出方法を説明する概略図である。
【図２】欠点を伴う自動車用ガラスの一例を示す図である。
【図３】ドット状プリントを含む自動車ガラスの一例を示す図である。
【図４】ドット状プリントを含む自動車ガラスの一例を示す図である。
【図５】従来の欠点検出方法を説明する概略図である。
【図６】本発明の実施の形態に係る欠点検出装置を示す概略構成図である。
【図７】本発明の実施の形態に係る欠点検出装置の動作を説明するフローチャートである。
【図８】本発明の実施の形態に係る欠点検出装置の動作を説明するフローチャートである。
【図９】取り込まれたガラス板の画像の一例を示す図である。
【図１０】高低差フィルタ処理を説明する概念図である。
【図１１】高低差フィルタ処理を説明する概念図である。
【図１２】高低差フィルタ処理を説明する概念図である。
【図１３】ガラス板全体の圧縮画像の一例を示す図である。
【図１４】原画像の切り出しを説明する概念図である。
【図１５】輝度算出の基礎となる画素の配置例を説明する図である。
【図１６】検査対象となるガラス板の一例を説明する概略構成図である。
【図１７】欠点のサイズと欠点が含まれるブロック数との関係を説明する概念図である。
【図１８】欠点のサイズと欠点が含まれるブロック数との関係を説明する概念図である。
【図１９】欠点のサイズと欠点が含まれるブロック数との関係を説明する概念図である。
【符号の説明】
２ 照明
３ 透明体
４ 撮像手段
５，５ｂ ピンホール欠点
５ａ 設計上のピンホール
６ インク汚れ欠点
７ ベタプリント部分
８ ドット状プリント部分
９ 透明部分
１２ 照明手段
１４ 撮像手段
１５ 画像処理手段
１６ 表示手段
２１ ベタ部外側エッジ
２２ ベタ内部
２３ ドット部
２４ 透明部
２５ インク汚れ
２６ ピンホール
３０ 画素[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a method and an apparatus for detecting a defect on a transparent body such as a glass plate, and in particular, for a transparent body including a dot-shaped printed portion such as an automobile glass, a pinhole defect existing in a solid printed portion and a transparent member. The present invention relates to a method and an apparatus for detecting contamination defects present in a part.
[0002]
[Prior art]
Conventionally, as a method of detecting a defect included in a transparent body such as a glass plate, for example, a method disclosed in Patent Document 1 is known. On the other hand, in the case of automotive glass, a black print may be applied around the glass. Here, if there is a dropout in the printed portion (pinhole) or if ink adheres to the transparent portion (ink adherence), a drawback occurs.
[0003]
As a method of detecting such a defect, there is a method of using a transmission bright field method and detecting a locally dark portion (pinhole) with a dark surrounding and a locally dark portion (ink stain) with a bright surrounding.
[0004]
Hereinafter, a conventional defect detection method will be described with reference to the drawings. FIG. 1 is a schematic view for explaining a conventional defect detection method, FIG. 2 is a view showing an example of an automobile glass having a defect, and FIGS. 3 and 4 are diagrams of an automobile glass including a dot-shaped print. FIG. 5 is a diagram illustrating an example, and FIG. 5 is a schematic diagram illustrating a conventional defect detection method.
[0005]
In FIG. 1, reference numeral 2 denotes illumination such as a fluorescent lamp, and reference numeral 4 denotes an imaging unit such as a line sensor camera. In this example, a transparent body 3 such as a glass plate to be inspected flows between the illumination 2 and the imaging unit 4 in a certain direction. Then, the transparent body 3 is irradiated with light by the illumination 2, and the transparent body 3 is sequentially imaged by the imaging means 4.
[0006]
As an example of the transparent body 3, as shown in FIG. 2, a pinhole defect 5 exists in a solid print portion and an ink stain defect 6 exists in a transparent portion, as shown in FIG. 2. Such a transparent body 3 is imaged by the line sensor camera 4 and image processing is performed on the imaged image to detect a pinhole defect in a solid print portion and an ink stain defect in a transparent portion.
[0007]
On the other hand, some automotive glass prints include not only solid prints but also dot prints, mainly for design reasons. For example, as shown in FIGS. 3 and 4, there is a glass having a solid print portion 7 having a fixed width on the outer peripheral portion of the glass and a dot-shaped print portion 8 having a fixed width inside the glass.
[0008]
In the case of a glass plate partially including such a dot-shaped print portion, if the above-described conventional technique is used, light and darkness are mixed in the dot portion, so that erroneous determination (oversight) increases. In order to avoid this, a program process is often performed so as to exclude the dot portion from the inspection target. Specifically, as shown in FIG. 5, the dot print area 8 in the image of the glass plate is excluded from the inspection target by the processing of the program.
[0009]
[Patent Document 1]
JP-A-2002-83303
[0010]
[Problems to be solved by the invention]
However, since the shape of the glass and the shape of the masking (print) differ depending on the type, it is necessary to make settings for all types. That is, the setting of the image area to be excluded from the inspection target must be performed separately for each product type. Further, even for the same type, the position of the dot masking (print) portion in the image captured by the camera is slightly different due to variations in print position accuracy and transport accuracy. Therefore, if the exclusion setting range is strictly performed on the dot range, the dot image area will be out of the range, and the dot range will be inspected. Conversely, if the exclusion setting range is loosened (larger), the dot range is less likely to deviate from the setting range, but the portion to be inspected becomes smaller, and the image area including the defective portion to be detected becomes the inspection target. There is a problem that there is a possibility of being excluded from the.
[0011]
An object of the present invention is to provide a method and an apparatus for detecting pinhole defects and dirt defects, which do not require setting of dot positions for each type of glass, and therefore have high versatility for dot portions.
[0012]
[Means for Solving the Problems]
Therefore, the present invention does not set the range to be excluded from the inspection target in advance, but sets the entire range of the image of the transparent body including the dot portion and the defect portion as the inspection target, and identifies only the dot portion thereon. It is excluded from defect candidates.
[0013]
Specifically, the present invention provides an apparatus for detecting a defect of a transparent body partially including a dot-shaped print portion. The apparatus includes imaging means for imaging the transparent body, and image processing means for processing an image of the transparent body imaged by the imaging means, wherein the image processing means converts the image of the transparent body into an original image. The original image is divided into a plurality of blocks each of m pixels vertically by n pixels horizontally (m and n are natural numbers), and a height difference filter value that is a value of a maximum luminance minus a minimum luminance of a pixel is obtained for each block. A height difference filter means for setting the obtained height difference filter value to the brightness of the block to generate a compressed image obtained by compressing the original image; and setting the brightness of the block and a predetermined brightness level for the compressed image. Defect extracting means for comparing a high-luminance area in the compressed image having luminance equal to or higher than the luminance level and extracting the identified high-luminance area as a defect candidate It is determined whether or not the number of blocks constituting each defect candidate is equal to or greater than a predetermined block reference number for the defect candidates, and a defect candidate formed of blocks equal to or greater than the block reference number is excluded. Defect candidate exclusion means.
[0014]
With the above configuration, even when a dot-shaped print portion is included in a part of the transparent body to be inspected, the dot portion can be automatically removed from the defect candidate.
[0015]
Further, the image processing means cuts out the original image around the defect candidate area, compares the luminance of the cut out original image with the luminance level, and sets the luminance of the cut out original image to the luminance level. In the above case, the defect candidate is determined as a stain defect, and when the luminance of the cut-out original image is lower than the luminance level, a defect type identification unit that determines the defect candidate as a pinhole defect is provided. Have more.
[0016]
With such a configuration, even when a part of the transparent body to be inspected includes a dot-shaped printed portion and includes both the pinhole defect and the dirt defect, the pinhole defect and the dirt defect are reduced. Can be identified and detected.
[0017]
The image processing means compares the number of determined pinhole defects with the number of designed pinholes, and if the determined number of pinhole defects is larger, a pin is inserted in the transparent body. A pinhole defect determining means for determining that a hole defect exists, and comparing the number of the determined dirt defects with the number of prints other than the solid print design, and when the determined number of the dirt defects is larger, Further includes a stain defect determining means for determining that a stain defect exists in the transparent body.
[0018]
By doing so, it is possible to determine the presence or absence of a defect even when the solid print portion includes a design pinhole and the transparent portion includes a design small print. .
[0019]
Further, the invention of the above-described defect detection device is also realized as an invention of a method. Further, the above-described invention is also realized as a program for causing a computer to realize a predetermined function or a recording medium on which the program is recorded.
[0020]
The means in the present specification can be realized by hardware, software, or a combination of hardware and software. Execution by a combination of hardware and software corresponds to, for example, execution in a computer system having a predetermined program.
[0021]
And even if the function of one means is realized by two or more hardware, software or a combination of hardware and software, the function of two or more means is realized by one hardware, software or hardware and software. It may be realized by a combination.
[0022]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings. First, a defect detection device according to an embodiment of the present invention will be described. FIG. 6 is a schematic configuration diagram showing a defect detection device according to the embodiment of the present invention. As shown in FIG. 6, the defect detection apparatus uses a transmission bright field method as an optical system, and includes a lighting unit 12, an imaging unit 14, an image processing unit 15, and a display unit 16. ing.
[0023]
As the illumination means 12, for example, a fluorescent lamp can be used. The illumination unit 12 irradiates light to the transparent body 3 from below. As the illuminating means 12, besides a fluorescent lamp, for example, a fiber illuminator that uses a halogen lamp as a light source and guides light from the halogen lamp through a fiber may be used. Alternatively, a rod-shaped LED illumination may be used.
[0024]
A transporting unit (not shown) such as a roller for transporting a transparent body 3 such as a glass plate as an object to be inspected is provided above the illumination unit 12. In this example, the transport means is arranged so that the transport direction of the transparent body 3 is orthogonal to the longitudinal direction of the fluorescent lamp 12.
[0025]
The imaging unit 14 is, for example, a line sensor camera that repeats one-dimensional scanning. The imaging means 14 is arranged on the side facing the illumination means 12 with the transparent body 3 conveyed therebetween. It is mounted in a direction perpendicular to the upper surface of the transparent body 3 so that the scanning line of the line sensor camera 14 is in a direction orthogonal to the transport direction of the transparent body 3, and is arranged so as to capture the transparent body 3 as a field of view. You. In addition, since the field of view of the line sensor camera is determined, the number of installations may be determined as appropriate according to the width of the inspection object. The imaging unit 14 is not limited to a line sensor camera as long as it can image the transparent body 3 by the transmission bright field method and output image data, and for example, a matrix camera can be used.
[0026]
The transparent body 3 is transported in a certain direction by the transporting unit between the imaging unit 4 and the illumination unit 12 as described above. In this example, the transport direction is to the right as indicated by the arrow in FIG. The image data output from the imaging unit 14 is input to the image processing unit 15.
[0027]
As the image processing means 15, for example, a computer can be used. The image processing unit 15 includes, for example, a processing control unit realized by a CPU or the like, and a storage unit realized by a cache memory, a main storage device, an auxiliary storage device, or the like. When the output of the line sensor camera 14 is an analog signal, it is necessary to convert the digital signal into a digital signal and take it into the computer. Therefore, the image processing means 15 may further include an image input device having at least an analog / digital conversion function. Required. When the line sensor camera 14 is a digital camera, analog / digital conversion is unnecessary.
[0028]
The image processing unit 15 includes an image capturing unit that captures an image of a transparent body to be inspected, and an image of the captured transparent body as an original image. The original image is m pixels vertically by n pixels horizontally (m and n are natural numbers). Into a plurality of blocks for each block, and for each block, find a height difference filter value that is a value of a maximum brightness minus a minimum brightness of a pixel, and set the found height difference filter value to the brightness of the block to obtain an original image. A height difference filter means for generating a compressed image, and comparing the luminance of the block with a predetermined luminance level for the compressed image to identify a high luminance area in the compressed image having a luminance equal to or higher than the luminance level. A defect candidate extracting means for extracting the identified high-luminance area as a defect candidate; and, for the defect candidate, the number of blocks constituting each defect candidate is determined by a predetermined block reference. Or whether to determine, and a non-defect candidate excluding means excludes the disadvantages candidates composed of the number of blocks or the number of blocks reference.
[0029]
The image processing means 15 further cuts out the original image around the defect candidate area, compares the luminance of the cut out original image with the above-mentioned luminance level, and, when the luminance of the cut out original image is equal to or higher than the luminance level, Is a defect type discriminating means for discriminating the defect candidate as a stain defect, and when the luminance of the cut-out original image is lower than the luminance level, discriminating the defect candidate as a pinhole defect. The number of pinhole defects is compared with the number of pinholes in the design, and if the number of pinhole defects determined is larger, pinhole defect determination means for determining that there is a pinhole defect in the transparent body is determined. Comparing the number of dirt defects with the number of prints other than the solid print in the design, and if the number of determined dirt defects is larger, determining that dirt defects exist in the transparent body; Having. These means are realized, for example, by causing a CPU of a computer to execute a predetermined program. The functions of these units will be described in more detail in the operation of the embodiment described later.
[0030]
The display means 16 is realized by a display device such as a CRT and a liquid crystal display. The display means 16 displays the output from the image processing means 15.
[0031]
Note that the above-described illumination means, imaging means, transport means, and arrangement of these means are merely examples, and do not limit the present invention. The illumination means, the image pickup means, the transport means, and the arrangement of these means of the defect detection apparatus used in the present invention employ a transmission bright field method as an optical system, image the transparent body 3, and convert the image data of the transparent body 3 into images. A configuration capable of outputting can be widely used.
[0032]
(Operation of the embodiment)
Next, the operation of the defect detection device according to the embodiment of the present invention will be described. Here, FIGS. 7 and 8 are flowcharts for explaining the operation of the defect detection device according to the embodiment of the present invention, and FIG. 9 is a diagram showing an example of the captured image of the glass plate. , 11, and 12 are conceptual diagrams illustrating height difference filter processing, FIG. 13 is a diagram illustrating an example of a compressed image of the entire glass plate, and FIG. 14 is a conceptual diagram illustrating cutout of an original image. FIG. 15 is a diagram illustrating an example of the arrangement of pixels serving as a basis for luminance calculation.
[0033]
As shown in FIG. 7, first, an image of a glass plate is captured by an apparatus as shown in FIG. 6 (step 102). Specifically, the glass plate 3 is caused to flow in a certain direction, and the imaging unit 14 sequentially captures images. Then, the captured image of the glass plate is captured by the image capturing unit of the image processing unit 15 and stored in the storage unit. In this step, the pitch (resolution) at which the image is captured needs to be smaller than the defect to be detected. This pitch can be set arbitrarily based on the size of the defect to be detected. Generally, however, it is desirable that the pitch is about の of the minimum defect to be detected.
[0034]
FIG. 9 shows an example of an image (hereinafter, referred to as an “original image”) of the glass plate captured in step 102. The glass plate of the image in FIG. 9 has a solid portion 7, a dot portion 8, and a transparent portion 9 in order from the outermost periphery. In this example, it is assumed that the transparent portion 9 has no print. The transparent portion 9 has an ink stain defect 6 and the solid portion 7 has pinholes 5a and 5b. The pinhole 5a is a pinhole intentionally provided in design and is not a defect. On the other hand, the pinhole 5b is a disadvantage. Note that this image is assumed as an example for describing the present embodiment, and the glass plate to which the present invention can be applied and the configuration of the image are not limited thereto.
[0035]
Next, as shown in FIG. 7, a height difference filter process is performed on the original image (step 104). Specifically, as shown in FIG. 10, the original image is divided into m × n pixels. Here, m and n are natural numbers, and m and n may be the same number (m = n) as shown in FIG. Then, for each block, (maximum luminance−minimum luminance) in each block is calculated and stored. This processing is referred to as “height difference filter processing”. In this process, the original image is formed into a matrix of blocks each including m pixels vertically by n pixels horizontally, and in each block, the luminance difference is calculated by subtracting the minimum luminance value from the maximum luminance value of the pixel. Is referred to as a “height difference filter value”), the original image is compressed by setting the brightness of the block using the obtained height difference filter value, and the compressed image is stored in the storage means. This processing replaces m × n pixels with one block, so that the information amount of the original image can be compressed to 1 / (m × n). Such processing can be performed by the height difference filter means of the image processing means 15.
[0036]
This will be described more specifically with reference to FIGS. In these figures, it is assumed that the brightness of the transparent portion is 255 and the brightness of the printed portion is zero. Blocks with a height difference filter value of 255 are represented by solid colors, and blocks with a height difference filter value of 0 are represented by white.
[0037]
In the example of these figures, the values of m and n are
m or n × resolution> maximum dot diameter
When m and n are set so as to be as shown in FIGS. 11 and 12, a) solid portion outer edge 21, b) solid inside 22, c) dot portion 23, d) transparent portion 24, e) ink stain 25, f) The height difference filter value of the block included in each area of the pinhole defect and the design cutout 26 is as follows.
a) Solid outer edge: 255-0 = 255
b) Solid inside: 0-0 = 0
c) Dot part: 255-0 = 255
d) Transparent part: 255-255 = 0
e) Ink stain: 255-0 = 255
f) Pinhole defect and design cut-out: 255-0 = 255
[0038]
When the luminance of each block is set using the height difference filter value thus obtained, a compressed image as shown in FIGS. 11 and 12 is obtained. FIG. 13 shows the obtained compressed image of the entire glass.
[0039]
Next, defect candidates are extracted (step 106). Specifically, an appropriate setting value A, that is, an appropriate luminance level is set, and an image having a block luminance equal to or higher than the setting value A is extracted. In the example of FIG. 13, a solid portion outer edge region 21, a dot portion region 23, an ink smear region 25, a pinhole defect, and a cutout region 26 in design are extracted. These regions are set as defect candidates. Here, the set value A can be an arbitrary value. To determine the set value A, once the image is captured, the luminance of the transparent portion and the print portion is measured,
A = (luminance of transparent portion−luminance of solid print portion) / 2
Is appropriate.
[0040]
The above-described defect candidate extraction processing can be performed by the defect candidate extraction unit in the image processing unit 15. For example, the defect candidate extraction unit compares the compressed image with the set value A, and identifies a high luminance area in the compressed image having a luminance equal to or more than the set value A. Then, the identified high-luminance area is extracted as a defect candidate area, an identification number is given to each defect candidate, and stored in the storage means. Note that, in order to identify individual defect candidates, a range in which blocks equal to or higher than the set luminance level are continuous is defined as one defect candidate region. More specifically, in FIG. 13, each of the ink stain 25 and the pinhole 26 is composed of one block, and each one block is set as one defect candidate. On the other hand, each of the solid portion outer edge 21 and the dot portion 23 is composed of a block that is continuous in a rectangular shape, and a range in which the block is continuous in a rectangular shape is set as one defect candidate. Such ranges of individual defect candidates can be recognized by, for example, labeling processing by software.
[0041]
Next, non-defect candidates included in the defect candidates are excluded. In this example, a process of excluding a masking (solid portion) outer edge and a dot portion from defect candidates is performed. As can be seen from FIG. 13, the number of blocks of the solid portion outer edge 21 and the dot portion 23 is much larger than that of the ink stain 25 and the pinhole 26. Therefore, an appropriate number of blocks (set value B) is set, and those with more blocks are excluded from defect candidates. In this processing, since it is necessary to exclude the minimum masking portion while leaving the defect candidate of the maximum block, the setting value B used here is, for example, when the number of blocks of the dot portion is represented by G,
Maximum number of blocks including defects <set value B <G
Preferably, specifically,
4 to 9 blocks <Set value B <G
It is preferred that
[0042]
More specifically, the setting value B used here can be set based on, for example, the size of the defect to be detected and the length of one side of the block. When the size of the defect to be detected is represented by its diameter φ, the length of one side of the block is represented by M or N (M = m × resolution, N = n × resolution), and the number of blocks in the dot portion is represented by G The value B may be set as follows. A specific description of a method for determining the set value B will be described later.
(1) When φ <M and φ <N 4 blocks <setting value B <G
(2) When φ ≧ M and φ <N 6 blocks <setting value B <G
(3) When φ <M and φ ≧ N, 6 blocks <setting value B <G
(4) When φ ≧ M and φ ≧ N 9 blocks <set value B <G
[0043]
Referring to FIG. 7, the identified defect candidates are sequentially targeted, and it is determined whether or not the number of blocks constituting the defect candidate is equal to or greater than a block reference number (set value B) (step 108). In the above case, it is determined as a break or a dot portion of the masking (printed portion) and is excluded from the defect candidates (step 110). On the other hand, when the number of blocks constituting the defect candidate is less than the block reference number, the defect candidate is left as a defect candidate, the defect candidate in the storage unit is updated, and the process proceeds to the next process. By such non-defect candidate exclusion processing, only the ink stain 25 and the pinhole 26 remain as defect candidates in the example of FIG. Note that such non-defect candidate exclusion processing can be performed by the non-defect candidate exclusion means in the image processing means 15.
[0044]
Next, the type of defect is identified. Here, an original image around the defect candidate is extracted, and pinholes and ink stains are identified. As can be seen from FIG. 13, in the compressed image, the pinhole and the ink stain cannot be distinguished. In order to make this distinction, an original image around the defect candidate is cut out and it is determined whether the periphery is dark and the center is bright (pinhole) or conversely the periphery is bright and the inside is dark (ink smear). Specifically, as shown in FIG. 14, it is appropriate that the range to be cut out is one block at a time around the defective block in the compressed image. FIG. 14A shows the cut-out range of the original image when the number of defective blocks in the compressed image is one, and FIG. 14B shows the cut-out range of the original image when the number of defective blocks in the compressed image is two. FIG. 14 is a diagram in the case of ink contamination, and the black and white of the original image is reversed in the pinhole.
[0045]
Next, as shown in FIG. 15, the average of the luminance of the pixels 30 at the four corners of the extracted original image is obtained. Then, it is determined whether or not the average luminance value is equal to or higher than the above-mentioned set value A (set luminance level). Judge as a hole defect.
[0046]
Referring to FIG. 7, first, an original image of one block around a defect candidate is extracted (step 112). Next, the average brightness of the four corner pixels of the cut-out original image is determined, and it is determined whether or not this average brightness value is equal to or greater than the set value A (step 114). If the average luminance value is equal to or more than the set value A, it is determined that the defect is an ink stain (step 118). If the average value is less than the set value A, it is determined that the defect is a pinhole defect (step 116). The defect type is stored in the storage means. Then, it is determined whether or not the last defect candidate has been processed (step 120). If the last defect candidate has not been processed, the process proceeds to the next defect candidate (step 122). Proceed to the next. Note that such defect type identification processing can be performed by the defect type identification unit of the image processing unit 15.
[0047]
By doing so, even if a part of the transparent body to be inspected includes a dot-shaped printed portion, it is possible to accurately detect defects such as pinhole defects and dirt defects. .
[0048]
Further, even when the position, shape, and the like of the dot region in the image of the transparent body change, the defect can be detected without resetting the program. Accordingly, detection can be performed without requiring high accuracy for the printing device and the transport mechanism of the detection device, and the cost of the device can also be reduced.
[0049]
Next, as shown in FIG. 8, the number of detected pinhole defects is compared with the number of pinholes present in the design. For example, as shown in FIG. 9, there is a case where a pinhole 5a exists in a solid portion in design. On the other hand, in the above-mentioned step 116, all of them including the design pinholes are detected as defects. Therefore, the number of detected pinhole defects is compared with the number of designed pinholes to determine whether the former is larger (step 202). If the former is larger, the pinhole defect is included in the glass plate. (Step 204), otherwise, it is determined that no defect is included. Note that such processing can be performed by the pinhole defect determining means in the image processing means 15.
[0050]
On the other hand, the number of detected ink stain defects is compared with the number of prints of the transparent portion which is present in the design. As in the case of the pinhole, a print may be present in the transparent portion by design. Therefore, similarly to the above, the number of detected ink stain defects is compared with the number of prints other than the solid print in design to determine whether the former is greater (step 206). If so, it is determined that there is an ink stain defect in the glass plate (step 208); otherwise, it is determined that no defect is included. Such processing can be performed by the stain defect determining means in the image processing means 15.
[0051]
The following table determines whether there is a defect other than the design pinholes and prints as described above in the case of the glass plate of FIG.
[0052]
[Table 1]

[0053]
In this manner, when a part of the transparent object to be inspected includes a dot-shaped print portion and a pinhole in design and / or a print other than solid in design is included. Even if it is, defects such as pinholes and dirt can be accurately detected.
[0054]
(Setting method of setting value B)
Hereinafter, a method of setting the set value B of the present invention will be specifically described using an example of a glass plate and a defect. Here, FIG. 16 is a schematic configuration diagram illustrating an example of a glass plate to be inspected, and FIGS. 17 to 19 are conceptual diagrams illustrating the relationship between the size of a defect and the number of blocks including the defect. .
[0055]
First, the specification of a general glass plate and a defect will be described. The glass plate shown in FIG. 16 is an isosceles triangle with a short side of 200 mm and a masking width of 50 mm.
[0056]
Therefore,
The perimeter of the glass plate is 683 mm,
Masking internal parting length = 100 + 100 + 100√2 ≒ 341 mm
It becomes.
[0057]
on the other hand,
Dot diameter of dot print <3mm
Diameter of pinhole and dirt defect 0.8φmm ～ 5.0φmm
And
[0058]
As mentioned above,
m or n × resolution> 3 mm (maximum dot diameter)
M and n are set so that
Since m or nx resolution is the length of one side of the block (in mm),
Length of one side of block> 3mm
Means
[0059]
As described above, the size of the defect to be detected is 0.8 φmm to 5.0 φmm. Further, since the length of one side of the block is 3 mm or more, when the size of the defect is 0.8 φmm, φ <M and φ <N, and as shown in FIG. Is 1 to 4 blocks (up to 4 blocks). When the size of the defect is 5.0 mm, φ ≧ M and φ ≧ N, and as shown in FIG. 18, the number of blocks including the defect is 4 to 9 blocks (maximum 9 blocks). .
[0060]
On the other hand, as described above, the perimeter of the inside masking parting is about 341 mm, so the number of blocks is
G = 341/3 ≒ 114 blocks.
[0061]
As described above, the number of defective blocks is 1 to 9 and the number of parting edges in the masking is 114, which is a large difference. Therefore, the set value B may be between the two (that is, 9 <set value B <114), and for example, a value of 50 is appropriate.
[0062]
More specifically, in the example of FIG. 17, 4 <set value B <114 may be satisfied.
(1) When φ <M and φ <N 4 blocks <setting value B <G
FIG.
[0063]
In addition, in the example of FIG. 18, it is sufficient that 9 <set value B <114.
(4) When φ ≧ M and φ ≧ N 9 blocks <set value B <G
FIG.
[0064]
Further, a case where the lengths of the sides of the block, M and N, are not M ≠ N will be described. As shown in FIG. 19, when either M or N is larger than the diameter φ of the defect (in FIG. 19, N = 6 mm, φ = 5 mm, N ≧ φ), the number of blocks including the defect is 2 ６6 blocks (up to 6 blocks). Therefore, the set value B may be 6 <set value B <114. This is described above,
(2) When φ ≧ M and φ <N 6 blocks <setting value B <G
(3) When φ <M and φ ≧ N, 6 blocks <setting value B <G
FIG.
[0065]
From the above description, the preferable range of the set value B is:
4 to 9 blocks <Set value B <G
It can be.
[0066]
This G can be determined from the perimeter of the masking inner part obtained from the drawing or actual measurement, the resolution, and m and n. In particular,
G = perimeter of masking inside part / length of one side of block
= Perimeter of the masking parting / (m or n × resolution)
Can be determined by:
[0067]
Further, since G is an integer value, in the present invention, inspection is theoretically possible if G ≧ 11. Actually, since there is a variation in measurement, in consideration of this, it is desirable that G is somewhat larger than 9 (for example, about 10 blocks).
[0068]
Note that, for example, a typical minimum glass plate actually used in an automobile is as shown in FIG. The dimensions of this glass plate are, as described above, a short side of 200 × 200 and a masking width of 50 mm. In this glass plate, the perimeter of the inside masking partition is 114 blocks (G = 114 blocks), and the difference from the number of defective blocks is sufficiently large, so that the inspection can be easily performed. That is, in the case of an automobile glass plate as an example, even when the size of the masking portion is the smallest in a practical range, the masking portion is distinguished from the defect, and the masking portion is surely excluded as a non-defect. be able to.
[0069]
This makes it possible to automatically remove the dot portion from the defect candidates even when the transparent image includes a dot-shaped print portion.
[0070]
In the above description, an example in which a defect of an inspection target is detected using transmitted light has been described. However, the present invention can also be applied to a case where a defect of an inspection target is detected using reflected light. Although the present invention has been described with reference to a glass plate such as a glass for automobiles, the present invention is not limited to this. In addition to the glass plate, the present invention can be applied to other transparent bodies such as a resin.
[0071]
The transparent body to which the present invention can be applied is not necessarily limited to a flat plate, but may be a plate having a gentle curvature, such as a panel. Further, as a test object, a translucent plate can be applied as long as it has translucency.
[0072]
【The invention's effect】
As described above, according to the present invention, since the dot portion is automatically excluded from the inspection target regardless of the type of glass, it is not necessary to set the dot position for each type, and variations in printing and transport Can be automatically handled.
[0073]
Further, since the determination is performed on the compressed image, the processing time can be reduced to 1 / (m × n).
[Brief description of the drawings]
FIG. 1 is a schematic diagram illustrating a conventional defect detection method.
FIG. 2 is a diagram showing an example of an automotive glass having a defect.
FIG. 3 is a diagram illustrating an example of an automobile glass including a dot print.
FIG. 4 is a diagram illustrating an example of an automobile glass including a dot print.
FIG. 5 is a schematic diagram illustrating a conventional defect detection method.
FIG. 6 is a schematic configuration diagram showing a defect detection device according to an embodiment of the present invention.
FIG. 7 is a flowchart illustrating an operation of the defect detection device according to the embodiment of the present invention.
FIG. 8 is a flowchart illustrating an operation of the defect detection device according to the embodiment of the present invention.
FIG. 9 is a diagram illustrating an example of a captured image of a glass plate.
FIG. 10 is a conceptual diagram illustrating height difference filter processing.
FIG. 11 is a conceptual diagram illustrating height difference filter processing.
FIG. 12 is a conceptual diagram illustrating height difference filter processing.
FIG. 13 is a diagram illustrating an example of a compressed image of the entire glass plate.
FIG. 14 is a conceptual diagram illustrating clipping of an original image.
FIG. 15 is a diagram illustrating an example of the arrangement of pixels serving as a basis for luminance calculation.
FIG. 16 is a schematic configuration diagram illustrating an example of a glass plate to be inspected.
FIG. 17 is a conceptual diagram illustrating the relationship between the size of a defect and the number of blocks including the defect.
FIG. 18 is a conceptual diagram illustrating the relationship between the size of a defect and the number of blocks including the defect.
FIG. 19 is a conceptual diagram illustrating the relationship between the size of a defect and the number of blocks including the defect.
[Explanation of symbols]
2 Lighting
3 Transparent body
4 Imaging means
5,5b Pinhole defect
5a Designed pinhole
6 Ink stain defects
7 Solid print part
8 dot print
9 Transparent parts
12 Lighting means
14 Imaging means
15 Image processing means
16 Display means
21 Solid outer edge
22 Inside Solid
23 dots
24 Transparent part
25 Ink stain
26 Pinhole
30 pixels

Claims (9)

An apparatus for detecting a defect of a transparent body partially including a dot-shaped print portion,
Imaging means for imaging the transparent body;
Image processing means for processing the image of the transparent body imaged by the imaging means,
The image processing means,
The image of the transparent body is used as an original image, and the original image is divided into a plurality of blocks each having m pixels vertically by n pixels horizontally (m and n are natural numbers). A height difference filter value, and a height difference filter means for generating a compressed image obtained by compressing the original image by setting the obtained height difference filter value to the luminance of the block,
For the compressed image, the luminance of the block is compared with a predetermined luminance level, a high-luminance area in the compressed image having luminance equal to or higher than the luminance level is identified, and the identified high-luminance area is identified as a defect. Defect candidate extracting means for extracting as a candidate,
For the defect candidates, a non-defect that determines whether the number of blocks constituting each defect candidate is equal to or greater than a predetermined block reference number and excludes the defect candidates composed of blocks equal to or greater than the block reference number A defect detection device comprising: a candidate exclusion unit. The image processing means,
Cut out the original image around the defect candidate area, compare the luminance of the cut out original image with the luminance level, and if the luminance of the cut out original image is equal to or higher than the luminance level, 2. The defect type identification unit according to claim 1, further comprising: defect type discriminating means for discriminating the candidate as a stain defect and, when the luminance of the cut-out original image is lower than the luminance level, discriminating the defect candidate as a pinhole defect. Defect detection device. The image processing means,
The number of pinhole defects determined is compared with the number of design pinholes, and if the number of determined pinhole defects is larger, the pin is determined to have a pinhole defect in the transparent body. Hole defect determining means,
The number of the determined stain defects is compared with the number of prints other than the solid prints in the design, and if the number of the determined stain defects is larger, the number of the stain defects determined to be present in the transparent body is determined. 3. The defect detection device according to claim 2, further comprising a defect determination unit. A method for detecting a defect of a transparent body partially including a dot-shaped print portion,
The image of the transparent body is used as an original image, and the original image is divided into a plurality of blocks each having m pixels vertically by n pixels horizontally (m and n are natural numbers). Calculating a height difference filter value, and setting the obtained height difference filter value to the luminance of the block to generate a compressed image obtained by compressing the original image,
For the compressed image, the luminance of the block is compared with a predetermined luminance level, a high-luminance area in the compressed image having luminance equal to or higher than the luminance level is identified, and the identified high-luminance area is identified as a defect. Extracting as candidates;
For the defect candidate, determining whether the number of blocks constituting each defect candidate is equal to or greater than a predetermined block reference number, and excluding the defect candidate composed of blocks equal to or greater than the block reference number; And a defect detection method. The original image around the area of the defect candidate is cut out, and the luminance of the cut out original image is compared with the luminance level. If the luminance of the cut out original image is equal to or higher than the luminance level, the defect 5. The defect detection method according to claim 4, further comprising the step of: determining the candidate as a stain defect; and, when the luminance of the cut-out original image is lower than the luminance level, determining the defect candidate as a pinhole defect. . Comparing the number of determined pinhole defects with the number of designed pinholes, and determining that there is a pinhole defect in the transparent body if the determined number of pinhole defects is larger. When,
Comparing the number of the determined stain defects with the number of prints other than the solid print on the design, and if the determined number of the stain defects is larger, determining that the stain defect exists in the transparent body; The defect detection method according to claim 5, further comprising: A program for detecting a defect of a transparent body partially including a dot-shaped print portion, the computer comprising:
The image of the transparent body is used as an original image, and the original image is divided into a plurality of blocks each having m pixels vertically by n pixels horizontally (m and n are natural numbers). Height difference filter value, and calculates the height difference filter value to the brightness of the block to generate a compressed image of the original image, height difference filter means,
For the compressed image, the luminance of the block is compared with a predetermined luminance level, a high-luminance area in the compressed image having luminance equal to or higher than the luminance level is identified, and the identified high-luminance area is identified as a defect. Defect candidate extracting means for extracting as a candidate,
For the defect candidates, a non-defect that determines whether the number of blocks constituting each defect candidate is equal to or greater than a predetermined block reference number and excludes the defect candidates composed of blocks equal to or greater than the block reference number Program for functioning as candidate exclusion means. Computer
Cut out the original image around the defect candidate area, compare the luminance of the cut out original image with the luminance level, and if the luminance of the cut out original image is equal to or higher than the luminance level, 8. The method according to claim 7, further comprising: determining a candidate as a stain defect; and, when the luminance of the cut-out original image is less than the luminance level, the defect candidate further serving as a defect type identification unit for determining the defect candidate as a pinhole defect. The program described in. Computer
The number of pinhole defects determined is compared with the number of design pinholes, and if the number of determined pinhole defects is larger, the pin is determined to have a pinhole defect in the transparent body. Hall defect determination means,
The number of the determined stain defects is compared with the number of prints other than the solid prints in the design, and if the number of the determined stain defects is larger, the number of the stain defects determined to be present in the transparent body is determined. 9. The program according to claim 8, further causing the program to function as a defect determination unit.

Images