JP7444171B2 - Surface defect discrimination device, appearance inspection device and program - Google Patents

Description

The present invention relates to a surface defect discriminating device for discriminating surface defects of an object to be inspected, such as a product or component having a surface with strong specular reflection properties, an external appearance inspection device equipped with this surface defect discriminating device, and a program.

Scratches on the surface of products and parts impair their appearance. Further, if a deposition plate for forming a thin film such as a film has irregularities due to scratches or the like, the irregularities are transferred to the manufactured thin film, resulting in defects in the thin film.

Therefore, visual inspection apparatuses have been proposed for detecting surface defects in various products, parts, film-formed plates, and the like.

For example, Patent Document 1 discloses a feature in which a part is photographed while switching light sources in multiple directions, and by analyzing the direction of the illumination light source and the photographed image, it is determined whether a shadow in the image is a defect or dirt. A technique for doing so has been disclosed.

Japanese Patent Application Publication No. 11-118450

However, the invention described in Patent Document 1 assumes that the object to be inspected is stationary. For example, a drum-driven belt component, which is difficult to control stationary, is inspected while the belt is moving to detect surface defects. It is impossible to determine.

Therefore, there is a need for a technology that can identify surface defects while moving the object to be inspected relative to a lighting device or a line sensor that takes images.

The present invention was made in view of such a technical background, and provides a surface defect determination device that can determine surface defects while moving an object to be inspected relative to a lighting device and a line sensor. The purpose is to provide visual inspection equipment and programs.

ただし、ｉ：サブピクセル推定位置のインデックス
ｊ：点灯している照明装置の識別番号
（９）前記判別手段は、前記位置合わせ手段により位置合わせされたサブピクセル画像において、各照明装置に対応する明点が重複せずかつ各明点が予め設定された範囲内にある場合は、被検査物の表面に凹欠陥または凸欠陥が存在すると判定する前項１、３、前項１または３を引用する前項４～７、８のいずれかに記載の表面欠陥判別装置。
（１０）前記判別手段は、前記位置合わせ手段により位置合わせされたサブピクセル画像において、各照明装置に対応する明点の位置が照明装置の配置位置と逆である場合は凹欠陥が存在し、逆でない場合は凸欠陥が存在すると判定する前項９に記載の表面欠陥判別装置。
（１１）前記判別手段は、前記位置合わせ手段により位置合わせされたサブピクセル画像において、各照明装置に対応する明点が重複しているときは、被検査物の表面にゴミまたはほこりが存在すると判定する前項１、３、前項１または３を引用する前項４～７、８、９、１０のいずれかに記載の表面欠陥判別装置。
（１２）全体の受光量が所定の閾値を超える画素を欠陥候補画素として検出し、検出された欠陥候補画素について、前記位置合わせ手段によりサブピクセル画像を位置合わせし、かつ判定手段により被検査物の表面欠陥を判別する前項１、３、前項１または３を引用する前項４～７、８、９、１０、１１のいずれかに記載の表面欠陥判別装置。
（１３）前記照明装置の光源として、ＬＥＤまたは可視光半導体レーザーが用いられる前項１～１２のいずれかに記載の表面欠陥判別装置。
（１４）前記照明装置は３個以上であり、前記ラインセンサを中心とする円周上でかつ、３６０度÷照明装置の数、の角度差で配置されている前項１～１３のいずれかに記載の表面欠陥判別装置。
（１５）異なる位置に配置された複数の照明装置と、各照明装置から被検査物に照射された照明光の反射光を受光可能なラインセンサと、前記被検査物を、前記照明装置及びラインセンサに対して相対的に移動させる移動手段と、各照明装置からの照明光を１つずつ所定の周期で切り替えて被検査物に照射させる照明制御手段と、前記移動手段により、前記被検査物を前記照明装置及びラインセンサに対して相対的に移動させながら、前記照明制御手段により、各照明装置からの照明光が切り替えられる毎に、被検査物からの反射光を受光して撮影を行うように、前記ラインセンサを制御するラインセンサ制御手段と、前項１～１４のいずれかに記載の表面欠陥判別装置と、を備えた外観検査装置。
（１６）異なる位置に配置された照明装置及びラインセンサに対して被検査物を相対的に移動させながら、前記各照明装置からの照明光を１つずつ切り替えて被検査物に照射させたときに、各照明装置からの照明光が切り替えられる毎に、被検査物からの反射光をラインセンサで受光して撮影することにより、前記照明光の切り替え分だけそれぞれ位置ずれした状態で複数の画像を取得する画像取得ステップと、前記画像取得ステップにより取得された、各照明装置に対応する画像を位置合わせする位置合わせステップと、前記位置合わせステップにより位置合わせされた画像から、被検査物の表面欠陥を判別する判別ステップと、をコンピュータに実行させ、前記ラインセンサの各画素の一部は、１つの前記照明装置による照明光が被検査物に照射されることによる今回の撮影と前回の撮影とで撮影範囲が重複する重複領域となっており、１つの画素における前記重複領域を除く部分をサブピクセルとするとき、今回の撮影での画素全体の受光量から前記重複領域の受光量を差し引くことで、今回のサブピクセルの受光量を推定しサブピクセル画像を作成するサブピクセル画像作成ステップを前記コンピュータにさらに実行させ、前記位置合わせステップでは、前記サブピクセル画像作成ステップにより作成された各照明装置に対応するサブピクセル画像を位置合わせする処理を前記コンピュータに実行させるためのプログラム。
（１７）前記サブピクセル画像作成ステップでは、前記重複領域の受光量を領域毎に補正した状態で、画素全体の受光量から差し引く処理を前記コンピュータに実行させる前項１６に記載のプログラム。
（１８）前記サブピクセル画像作成ステップでは、前記重複領域の受光量を、前回以前に推定されたサブピクセルの受光量の和から求め、求めた受光量を画素全体の受光量から差し引いて今回のサブピクセルの受光量を推定する処理を前記コンピュータに実行させる前項１６または１７に記載のプログラム。
（１９）前記サブピクセル画像作成ステップでは、撮影開始後の最初の画素全体の受光量を１画素あたりのサブピクセルの数で割った平均値を、最初のサブピクセルの受光量と推定する処理を前記コンピュータに実行させる前項１８に記載のプログラム。
（２０）前記サブピクセル画像作成ステップでは、画素全体の受光量が所定の閾値を超えない場合、画素全体の受光量を１画素あたりのサブピクセルの数で割った平均値を、今回のサブピクセルの受光量と推定し、画素全体の受光量が所定の閾値を超える場合、画素全体の受光量から前記重複領域の受光量を差し引くことで、今回のサブピクセルの受光量を推定する処理を前記コンピュータに実行させる前項１６～１９のいずれかに記載のプログラム。
（２１）前記位置合わせステップでは、前記サブピクセル画像作成ステップにより作成された各照明装置に対応するサブピクセル画像の位置合わせを、下記式により輝度値Ｋijを補正値Ｋ'ijに補正することにより行う処理を前記コンピュータに実行させる前項１６～２０のいずれかに記載のプログラム。

ただし、ｉ：サブピクセル推定位置のインデックス
ｊ：点灯している照明装置の識別番号
（２２）前記判別ステップでは、前記位置合わせステップにより位置合わせされたサブピクセル画像において、各照明装置に対応する明点が重複せずかつ各明点が予め設定された範囲内にある場合は、被検査物の表面に凹欠陥または凸欠陥が存在すると判定する処理を前記コンピュータに実行させる前項１６～２１のいずれかに記載のプログラム。
（２３）前記判別ステップでは、前記位置合わせステップにより位置合わせされたサブピクセル画像において、各照明装置に対応する明点の位置が照明装置の配置位置と逆である場合は凹欠陥が存在し、逆でない場合は凸欠陥が存在すると判定する処理を前記コンピュータに実行させる前項２２に記載のプログラム。
（２４）前記判別ステップでは、前記位置合わせステップにより位置合わせされたサブピクセル画像において、各照明装置に対応する明点が重複しているときは、被検査物の表面にゴミまたはほこりが存在すると判定する処理を前記コンピュータに実行させる前項１６～２３のいずれかに記載のプログラム。
（２５）全体の受光量が所定の閾値を超える画素を欠陥候補画素として検出し、検出された欠陥候補画素について、サブピクセル画像を位置合わせする前記位置合わせステップを前記コンピュータに実行させ、かつ判定ステップにより被検査物の表面欠陥を判別する処理を前記コンピュータに実行させる前項１６～２４のいずれかに記載のプログラム。
（２６）前記照明装置の光源として、ＬＥＤまたは可視光半導体レーザーが用いられる前項１６～２５のいずれかに記載のプログラム。
（２７）複数個の前記照明装置は、前記ラインセンサを中心とする円周上かつ３６０度／（照明装置の個数）の位置に配置されている前項１６～２６のいずれかに記載のプログラム。
The above objective is achieved by the following means.
(1) When the object to be inspected is moved relative to the illumination devices and line sensors placed at different positions, and the illumination light from each of the lighting devices is switched one by one to illuminate the object to be inspected. Each time the illumination light from each illumination device is switched, the line sensor receives and photographs the reflected light from the object to be inspected, thereby creating multiple images with respective positions shifted by the switching amount of the illumination light. an image acquisition means for acquiring images corresponding to each illumination device; and an alignment means for aligning images corresponding to each illumination device acquired by the image acquisition means; and a determination means for determining a defect, wherein a part of each pixel of the line sensor is photographed in the current photographing and the previous photographing by irradiating the inspection object with illumination light from one of the illuminating devices. When the area is an overlapping area where the range overlaps, and the part of one pixel excluding the overlapping area is defined as a sub-pixel, by subtracting the amount of light received in the overlapping area from the amount of light received by the entire pixel in the current shooting, It further includes sub-pixel image creation means for estimating the amount of light received by the current sub-pixel and creates a sub-pixel image, and the positioning means is configured to generate a sub-pixel image corresponding to each illumination device created by the sub-pixel image creation means. A surface defect discrimination device that aligns the surface .
( 2 ) When the object to be inspected is moved relative to the illumination devices and line sensors placed at different positions, and the illumination light from each of the lighting devices is switched one by one to illuminate the object to be inspected. The image acquisition means acquires a plurality of images for each illumination light by receiving and photographing reflected light from the object to be inspected with a line sensor each time the illumination light from each illumination device is switched. In preparation, a part of each pixel of the line sensor becomes an overlapping area where the photographing ranges of the current photographing and the previous photographing overlap due to illumination light from one of the illuminating devices being irradiated onto the inspected object. Therefore, when the part of one pixel excluding the overlapping area is defined as a subpixel, the amount of light received by the current subpixel is estimated by subtracting the amount of light received in the overlapping area from the amount of light received by the entire pixel in the current shooting. A surface defect further comprising: a sub-pixel image creating means for creating a sub-pixel image; and a determining means for determining a surface defect of an object to be inspected based on the sub-pixel image created by the sub-pixel image creating means. Discrimination device.
( 3 ) The surface defect discriminating device according to item 2, further comprising positioning means for positioning subpixel images corresponding to each illumination device created by the subpixel image creation means.
( 4 ) The surface defect discriminating device according to any one of items 1 to 3 , wherein the sub-pixel image creation means subtracts the amount of light received in the overlapping region from the amount of light received in the entire pixel after correcting the amount of light received in the overlapping region for each region.
( 5 ) The sub-pixel image creation means calculates the amount of light received in the overlapping area from the sum of the amounts of light received by the sub-pixels estimated before the previous time, subtracts the calculated amount of received light from the amount of light received by the entire pixel, and calculates the amount of light received in the overlapping area in this time. 5. The surface defect determination device according to any one of items 1 to 4 , which estimates the amount of light received by a subpixel.
( 6 ) The sub-pixel image creation means estimates the amount of light received by the first sub-pixel as the average value obtained by dividing the amount of light received by the entire first pixel after the start of imaging by the number of sub-pixels per pixel. The surface defect determination device described in .
( 7 ) If the amount of light received by the entire pixel does not exceed a predetermined threshold, the subpixel image creation means calculates the average value obtained by dividing the amount of light received by the entire pixel by the number of subpixels per pixel, and determines the amount of light received by the current subpixel. If the amount of light received by the entire pixel exceeds a predetermined threshold, the amount of light received by the current sub-pixel is estimated by subtracting the amount of light received by the overlapping area from the amount of light received by the entire pixel. 6. The surface defect determination device according to any one of 6 .
( 8 ) The alignment means aligns the sub-pixel images corresponding to each illumination device created by the sub-pixel image creation means by correcting the brightness value Kij to the correction value K'ij using the following formula. 8. The surface defect discriminating device according to any one of the preceding sections 4 to 7 , which cites the preceding sections 1 and 3 .

However, i: index of estimated sub-pixel position
j: Identification number of lighting device that is lit ( 9 ) The discrimination means determines whether the bright points corresponding to each lighting device do not overlap and each bright point in the subpixel image aligned by the alignment means. If it is within a preset range, it is determined that a concave defect or a convex defect exists on the surface of the object to be inspected.Described in any of the preceding sections 1 and 3 , or any of the preceding sections 4 to 7 and 8 that cite the preceding section 1 or 3 . surface defect determination device.
( 10 ) The determining means determines that a concave defect exists if the position of the bright point corresponding to each illumination device is opposite to the arrangement position of the illumination device in the sub-pixel image aligned by the alignment means; The surface defect discriminating device according to item 9 above, which determines that a convex defect exists if the reverse is not the case.
( 11 ) When the bright points corresponding to each illumination device overlap in the sub-pixel images aligned by the alignment unit, the discrimination unit determines that dirt or dust is present on the surface of the object to be inspected. The surface defect discriminating device according to any one of the preceding sections 4 to 7 , 8 , 9 , and 10 , which cites the preceding section 1 or 3 .
( 12 ) Pixels for which the total amount of light received exceeds a predetermined threshold are detected as defective candidate pixels, and the sub-pixel images of the detected defective candidate pixels are aligned by the alignment means, and the determination means 1. The surface defect discriminating device according to any one of the preceding sections 4 to 7 , 8 , 9 , 10 , and 11 that cites the preceding section 1 or 3 .
( 13 ) The surface defect determination device according to any one of items 1 to 12 , wherein an LED or a visible light semiconductor laser is used as a light source of the illumination device.
( 14 ) In any one of the preceding clauses 1 to 13 , the lighting devices are three or more, and are arranged on a circumference centered on the line sensor with an angular difference of 360 degrees divided by the number of lighting devices. The surface defect determination device described.
( 15 ) A plurality of illumination devices arranged at different positions, a line sensor capable of receiving reflected light of the illumination light irradiated onto the object to be inspected from each illumination device, and the object to be inspected is connected to the illumination device and the line sensor. A moving means for moving the object to be inspected relative to the sensor, an illumination control means for switching the illumination light from each illumination device one by one at a predetermined cycle and irradiating the object to be inspected, and the moving means to move the object to be inspected. is moved relative to the illumination device and the line sensor, and each time the illumination light from each illumination device is switched, the illumination control means receives reflected light from the object to be inspected and photographs the object. 15. A visual inspection apparatus comprising: line sensor control means for controlling the line sensor; and the surface defect discriminating apparatus according to any one of items 1 to 14 above.
( 16 ) When the object to be inspected is moved relative to the illumination devices and line sensors arranged at different positions, and the illumination light from each illumination device is switched one by one to illuminate the object to be inspected. Each time the illumination light from each illumination device is switched, the line sensor receives and photographs the reflected light from the object to be inspected, thereby creating multiple images with respective positions shifted by the switching amount of the illumination light. an alignment step of aligning the images corresponding to each illumination device acquired in the image acquisition step; and a A determination step of determining a defect is executed by a computer , and a part of each pixel of the line sensor is divided into the current photographing and the previous photographing by irradiating the object with illumination light from one of the illuminating devices. When the photographing range is an overlapping area, and the part of one pixel excluding the overlapping area is defined as a sub-pixel, the amount of light received in the overlapping area is subtracted from the amount of light received by the entire pixel in the current shooting. This causes the computer to further execute a sub-pixel image creation step of estimating the amount of light received by the current sub-pixel and creating a sub-pixel image, and in the alignment step, each illumination created in the sub-pixel image creation step A program for causing the computer to perform a process of aligning subpixel images corresponding to devices .
( 17 ) The program according to item 16 , wherein, in the sub-pixel image creation step, the computer executes a process of subtracting the amount of light received in the overlapping region from the amount of light received by the entire pixel after the amount of light received in the overlapping region is corrected for each region.
( 18 ) In the sub-pixel image creation step, the amount of light received in the overlapping area is calculated from the sum of the amounts of light received by the sub-pixels estimated before the previous time, and the calculated amount of received light is subtracted from the amount of light received by the entire pixel to obtain the current amount of light received. 18. The program according to item 16 or 17 , which causes the computer to execute a process of estimating the amount of light received by a sub-pixel.
( 19 ) In the sub-pixel image creation step, the average value obtained by dividing the amount of light received by the entire first pixel after the start of imaging by the number of sub-pixels per pixel is estimated as the amount of light received by the first sub-pixel. 19. The program according to item 18 , which is executed by the computer.
( 20 ) In the subpixel image creation step, if the amount of light received by the entire pixel does not exceed a predetermined threshold, the average value obtained by dividing the amount of light received by the entire pixel by the number of subpixels per pixel is calculated as the current subpixel. If the amount of light received by the entire pixel exceeds a predetermined threshold, the amount of light received by the current sub-pixel is estimated by subtracting the amount of light received by the overlapping area from the amount of light received by the entire pixel. 20. The program according to any one of the preceding items 16 to 19 , which is executed by a computer.
(21) In the alignment step, the alignment of the sub-pixel images corresponding to each lighting device created in the sub-pixel image creation step is performed by correcting the brightness value Kij to the correction value K'ij using the following formula. 21. The program according to any one of items 16 to 20 , which causes the computer to execute the processing to be performed.

However, i: index of estimated sub-pixel position
j: Identification number of lighting device that is lit ( 22 ) In the discrimination step, in the sub-pixel images aligned in the alignment step, bright points corresponding to each lighting device do not overlap and each bright point 22. The program according to any one of the preceding items 16 to 21 , which causes the computer to execute a process of determining that a concave defect or a convex defect exists on the surface of the object to be inspected if the defect is within a preset range.
( 23 ) In the determination step, in the sub-pixel image aligned in the alignment step, if the position of the bright point corresponding to each illumination device is opposite to the arrangement position of the illumination device, a concave defect exists; 23. The program according to item 22 , which causes the computer to execute a process of determining that a convex defect exists if the reverse is not the case.
( 24 ) In the discrimination step, if bright points corresponding to each illumination device overlap in the sub-pixel images aligned in the alignment step, it is determined that dirt or dust is present on the surface of the object to be inspected. 24. The program according to any one of items 16 to 23 , which causes the computer to execute a process for determining.
( 25 ) Detecting pixels whose total amount of received light exceeds a predetermined threshold as defective candidate pixels, and causing the computer to execute the alignment step of aligning sub-pixel images for the detected defective candidate pixels, and making a determination. 25. The program according to any one of items 16 to 24 , which causes the computer to execute a process of determining surface defects of an object to be inspected in steps.
( 26 ) The program according to any one of items 16 to 25 above, wherein an LED or a visible light semiconductor laser is used as a light source of the lighting device.
( 27 ) The program according to any one of items 16 to 26 , wherein the plurality of lighting devices are arranged on a circumference around the line sensor at positions of 360 degrees/(number of lighting devices).

According to the invention described in the preceding paragraph (1), while the object to be inspected is moved relative to the lighting devices and line sensors arranged at different positions, the illumination light from each lighting device is switched one by one. The object to be inspected is irradiated. Each time the illumination light from each illumination device is switched, a line sensor receives and photographs the reflected light from the object to be inspected, and multiple images are acquired with each position shifted by the switching amount of illumination light. Ru. The acquired images corresponding to each illumination device are aligned, and then surface defects of the object to be inspected are determined from the aligned images.

In this way, since the object to be inspected is moving relative to the lighting device and the line sensor, the position of each object is shifted by the amount of switching of the lighting light when the lighting light from each lighting device is switched. Then, the multiple images corresponding to each illumination device acquired from the line sensor are aligned, and surface defects on the inspected object are determined in this aligned state, so the inspection object can be moved relatively. However, it is possible to identify surface defects on the object to be inspected.

In addition, a part of each pixel of the line sensor is an overlapping area where the photographing range of the current photographing and the previous photographing overlaps due to the illumination light from one illumination device being irradiated onto the inspected object. When the part of one pixel excluding the overlapping area is defined as a subpixel, the amount of light received by the current subpixel is estimated by subtracting the amount of light received in the overlapping area from the amount of light received by the entire pixel in the current shooting, and the amount of light received by the subpixel is estimated. An image is created. Then, the created sub-pixel images corresponding to each illumination device are aligned, and surface defects of the object to be inspected are detected from the aligned images. Here, since the sub-pixel is a portion of a pixel excluding an overlapping region, it is smaller than one pixel, and therefore the resolution of the visual inspection is improved and finer surface defects can be detected.

In other words, when photographing a relatively moving non-inspection object with a line sensor, the distance between the line sensor and the non-inspection object's imaging surface is unstable, and the depth of field must be set deep. There is a trade-off relationship between increasing the depth of field and decreasing resolution, and it may not be possible to inspect even the smallest defects. However, by using subpixel images smaller than one pixel, resolution can be improved without increasing the depth of field. This has the advantage of increasing the accuracy and allowing inspection of even finer defects.

According to the invention described in the above item ( 2 ), since defects are determined from sub-pixel images smaller than one pixel, the resolution of appearance inspection is improved and finer surface defects can be detected.

According to the invention described in item ( 3 ) above, by aligning the sub-pixel images corresponding to each illumination device, more accurate defect detection is possible.

According to the invention described in the previous section ( 4 ), the amount of light received in the overlapping area is corrected for each area and then subtracted from the amount of light received in the entire pixel. The amount of light received by a pixel can be estimated, which in turn enables more accurate defect determination.

According to the invention described in the preceding paragraph ( 5 ), the amount of light received in the overlapping area is calculated from the sum of the amounts of light received by sub-pixels estimated before the previous time, and the calculated amount of light received is subtracted from the amount of light received by the entire pixel to calculate the current amount of light received. Since the amount of light received by each sub-pixel is estimated, the estimation process is simplified.

According to the invention described in the preceding paragraph ( 6 ), the average value obtained by dividing the amount of light received by the entire first pixel after the start of shooting by the number of subpixels per pixel is estimated as the amount of light received by the first subpixel. , it is possible to smoothly perform estimation processing of the amount of light received by sub-pixels from the next time onwards.

According to the invention described in the preceding paragraph ( 7 ), when the amount of light received by the entire pixel does not exceed a predetermined threshold value, in other words, when there is a high possibility that no surface defect exists, the amount of light received by the entire pixel is determined by the amount of light received per pixel. The average value divided by the number of sub-pixels is estimated as the amount of light received by the current sub-pixel. On the other hand, if the amount of light received by the entire pixel exceeds a predetermined threshold, in other words, if there is a high possibility that a surface defect exists, the amount of light received by the overlapping area is subtracted from the amount of light received by the entire pixel. The amount of light received by the pixel is estimated. Thereby, the defect determination process can be executed while concentrating on areas where there is a high possibility that surface defects exist.

According to the invention described in the preceding item ( 8 ), it is possible to accurately align the sub-pixel images corresponding to each illumination device, and as a result, it is possible to perform highly accurate defect determination.

According to the invention described in the preceding item ( 9 ), concave defects or convex defects such as scratches on the surface of the object to be inspected can be determined.

According to the invention described in the preceding item ( 10 ), it is possible to discriminate between concave defects and convex defects on the surface of the object to be inspected.

According to the invention described in the preceding paragraph ( 11 ), dirt or dust on the surface of the object to be inspected can be determined.

According to the invention described in the preceding section ( 12 ), pixels whose total amount of light reception exceeds a predetermined threshold are detected as defective candidate pixels, subpixel images are aligned with respect to the detected defective candidate pixels, and the pixels to be inspected are Since the surface defects of the object are determined, the defect determination process can be concentrated on areas where surface defects are likely to exist.

According to the invention described in the preceding item ( 13 ), since an LED or a visible light semiconductor laser is used as the light source of the lighting device, each lighting device can be switched at high speed.

According to the invention described in the preceding paragraph ( 14 ), there are three or more lighting devices, and they are arranged on the circumference with the line sensor as the center and with an angular difference of 360 degrees ÷ the number of lighting devices. A lighting device in which the proof light is not perpendicular to scratches, etc. is always ensured, and surface defects such as scratches can be identified with high accuracy.

According to the invention described in the preceding paragraph ( 15 ), it is possible to accurately determine surface defects on non-inspected objects while moving the object to be inspected relative to the illumination device and line sensor arranged at different positions. This is an excellent visual inspection device.

According to the invention described in the preceding sections (16) to (27) , the surface defect determination process of the non-inspected object is performed while the object to be inspected is moved relative to the illumination device and line sensor arranged at different positions. It can be executed by a computer.

1 is a configuration diagram of an appearance inspection apparatus according to an embodiment of the present invention. (A) and (B) are diagrams for explaining the arrangement relationship of a plurality of lighting devices. FIG. 3 is a diagram for explaining the relative positional relationship between a photographing range and pixels when photographing while switching between a plurality of illumination devices. FIG. 7 is a diagram for explaining the relative positional relationship between the photographing range and pixels when the first to fourth photographs are taken using one illumination device. 3 is a diagram for explaining a method for estimating the amount of light received by a sub-pixel 23. FIG. FIG. 3 is a diagram for explaining alignment of sub-pixel images for a plurality of lighting devices. FIG. 3 is a diagram showing an example of sensitivity distribution of pixels. FIG. 7 is a diagram illustrating an example of a corrected light reception amount estimated value of each region of a pixel, which is calculated in consideration of weighting of the sensitivity distribution of the pixel. FIG. 7 is a diagram for explaining that the amount of received reflected light also differs depending on the region of the pixel because the distribution of the amount of irradiated light from a plurality of illumination devices is different. It is a figure for explaining the discrimination method of a void defect. It is a figure for explaining the discrimination method of a convex defect. FIG. 3 is a diagram for explaining a method for determining scratch defects. FIG. 3 is a diagram for explaining a method for determining dust and dirt. FIG. 3 is a diagram schematically showing an image that is determined to be a void defect by combining a plurality of sub-pixel images. FIG. 6 is a diagram schematically showing an image that is determined to be a convex defect by combining a plurality of sub-pixel images. FIG. 3 is a diagram schematically showing an image that is determined to be a flaw defect by combining a plurality of sub-pixel images. FIG. 3 is a diagram schematically showing an image that is determined to be dust or dirt by combining a plurality of sub-pixel images.

Embodiments of the present invention will be described below based on the drawings.
[Configuration of appearance inspection device]
FIG. 1 is a configuration diagram of a visual inspection apparatus according to an embodiment of the present invention. As shown in FIG. 1, the appearance inspection apparatus includes a line sensor 1, two lighting devices 2a and 2b, a lighting control section 8 that controls each lighting device 2a and 2b, and a line sensor control unit that controls the line sensor 1. 9, transport drums 3 for transporting the object to be inspected 5, a drum encoder 4 for detecting the photographing position of the object to be inspected 5, a display device 6, a computer 10, and a controller for controlling the transport speed of the object to be inspected 5. For control purposes, a conveyance drum control section 11 for controlling the rotation speed of the conveyance drum 3 is provided.

The computer 10 processes images taken by the line sensor 1 to determine defects, and also synchronously controls the illumination devices 2a and 2b and the line sensor 1. Furthermore, the display device 6 displays images subjected to defect determination processing by the computer 10, processing results, and the like.

The object to be inspected 5 is in the form of a belt with high reflectance, and is installed in a roll using a transport drum 3, and is transported in the Y direction by rotation of the transport drums 3, 3 in the direction of the arrow. The photographing position of the inspection object 5 by the line sensor 1 is detected by the drum encoder 4.

The line sensor 1 extends in the X direction perpendicular to the moving direction Y of the object to be inspected 5, and the two illumination devices 2a and 2b are connected to the line sensor when viewed from above, as shown in FIG. 2(A). They are arranged with an angle difference of 180 degrees at target positions centered on 1, and can be illuminated from two different directions. The direction in which the lighting devices 2a and 2b face each other may be the X direction, the Y direction, or any other direction. In this embodiment, two lighting devices 2a and 2b are used, but three or more may be used. In the case of three or more lighting devices, as shown in FIG. 2(B), they are arranged on the circumference centered on the line sensor 1 when viewed from above, and with an angular difference of 360 degrees ÷ the number of lighting devices. It is desirable to ensure that an illumination device that is not perpendicular to surface defects such as linear scratches is ensured, which allows surface defects such as scratches to be identified with high accuracy. Note that FIG. 2B shows the case of three lighting devices 2a, 2b, and 2c, which are arranged with an angular difference of 120 degrees from each other.

Each of the lighting devices 2a and 2b can be turned on and off at any timing under the control of the lighting control section 8.

If the photographing range of one line of the line sensor 1 is small compared to the object to be inspected 5, after photographing the object to be inspected once in the Y direction, the sensor illumination conveyance unit 12 connects the illumination devices 2a and 2b to the line. The sensor 1 may be moved in the X direction by the length of the line sensor 1 and photographed once again in the Y direction, and this may be sequentially repeated to photograph the entire inspection object 5.

The line sensor 1 receives reflected light when the object to be inspected 5 is illuminated by switching on/off of each lighting device 2a, 2b while moving the object to be inspected 5 in the Y direction. The line sensor 1 and each illumination device 2a, 2b are not directly facing each other, and the line sensor 1 receives reflected light obtained by diffusely reflecting the illumination light from each illumination device 2a, 2b on the object to be inspected 5. Therefore, the image taken by the line sensor 1 becomes a dark field image.

Since the surface of the object to be inspected 5 has a high reflectance, if there are concave defects, convex defects, scratch-like defects, dust, dirt, etc. at the illumination position, the reflected light diffusely reflected by these defects, dust, dirt, etc. will be reflected as a line. incident on sensor 1.

The line sensor control section 9 and the illumination control section 8 are connected to the computer 10, and the line sensor 1 and the illumination device 2a, and the line sensor 1 and the illumination device 2b are synchronously emitted and photographed, respectively.

In this embodiment, the line rate of the line sensor 1 is set to 100 kHz (shutter speed 0.01 ms) based on the specifications of general line sensors. In other words, it is preferable to use an LED light source or an LD (visible light semiconductor laser) light source that can be alternately switched at high speed every 0.01 ms as the illumination device 2a and the illumination device 2b.
[Surface defect determination processing]
Next, the surface defect determination processing of the inspection object 5 by the computer 10 will be explained. Note that the computer 10 is equipped with a CPU, a RAM, a storage device, etc., and the surface defect determination process is executed by the CPU operating according to an operation program stored in the storage device, etc.

As described above, photographing by the line sensor 1 is performed by alternately turning on and off a plurality of (two in this embodiment) illumination devices 2a and 2b while moving the object 5 to be inspected. The computer 10 sequentially acquires images taken by the line sensor 1.
<Creation of subpixel image>
FIG. 3 is a diagram for explaining the relative positional relationship between the photographing range and the pixel 20 when photographing is performed while switching between the illumination devices 2a and 2b.

As shown in FIG. 3, the size of the defect 30 to be detected is 12A, the resolution of the line sensor 1 (length of one pixel 20) is 6A, and the illumination devices 2a and 2b are switched every time the object 5 to be inspected is fed by A. Photographs shall be taken. The resolution of the line sensor 1 is 6A, which corresponds to one imaging area of one pixel. Therefore, as shown in FIG. 3, the first photographing is performed using the illumination light of the illumination device 2a, and when the inspection object 5 is sent A, the second photographing is performed, and the photographing is performed using the illumination light of the illumination device 2b. be exposed. Between the first photographing and the second photographing, the photographing area of the object to be inspected 5 has moved by A. The same applies to the third and subsequent shootings. In the example of FIG. 3, for convenience of explanation, a state is shown in which the pixel 20 moves by A every time switching shooting is performed.

FIG. 4 is a diagram for explaining the relative positional relationship between the photographing range and the pixel 20 when the first to fourth photographs are taken using one lighting device 2a.

As shown in FIG. 4, focusing on the illumination device 2a, the illumination device 2a is turned on every time the object to be inspected 5 moves by 2A, and starts irradiating the object to be inspected with the illumination light. Photographed by 1. That is, every time the object to be inspected 5 moves by 2A, a photograph is taken corresponding to the illumination light from the illumination device 2a. The irradiation time of the illumination light of the illumination device 2a, in other words, the light reception time of each pixel 20 of the line sensor 1 is a time corresponding to the moving distance A. These points also apply to the lighting device 2b.

In addition, as shown in FIG. 4, when photographing with the illumination device 2a, 4A, which is a part of the sensor resolution 6A, is obtained by photographing the same photographic range of the inspected object 5 in the current photographing and the previous photographing. This is an overlapping area where the shooting ranges overlap. In other words, if the pixel 20 is divided into three regions in the length direction, the first region 21, the second region 22, and the third region 23, the length per region is 2A, and the second region 22 in the previous shooting The third area 23 and the first area 21 and second area 22 of the current imaging are overlapping areas with the same imaging range. For example, the first region 21 and the second region 22 of the fourth imaging overlap with the second region 22 and the third region 23 of the third imaging, respectively. In FIG. 4, the overlapping area between the current imaging and the previous imaging is displayed in gray.

Regarding the third region 23 of the current photographing, the photographing range does not overlap with the previous photographing and is updated as a new photographing range, and this is defined as a sub-pixel. Hereinafter, the third area will also be referred to as a subpixel.

FIG. 5 is a diagram for explaining a method for estimating the amount of light received by the sub-pixel 23, and shows the relative relationship between the photographing range and the pixel 20 when the illumination device 2a performs the i-th photograph and several times before and after the i-th photograph. It shows the positional relationship.

As shown in FIG. 5, the amount of light received by the sub-pixel 23 in the i-th imaging is determined from the amount of light received by the entire one pixel 20 (6A) in the i-th imaging. It is necessary to calculate and estimate by subtracting the amount of light received for 4A in the area 21 and the second area 22.

In comparison with the (i-3)th image, the (i-2)th image is updated by 2A of the subpixel 23, and each time the number of shots increases from the (i-1)th time to the ith time, the subpixel 23 is updated in order. It is updated every 2A minutes. The updated new sub-pixel 23 becomes an overlapping area at the next imaging, remains as an overlapping area at the next imaging, and is removed from the overlapping area at the next imaging. In other words, the overlapping area between the current imaging and the previous imaging is the sub-pixel 23 from the previous two imagings, the previous one and the time before the previous one. Therefore, the amount of light received by the sub-pixel 23 in the i-th shooting is determined from the total amount of light received for 1 pixel 6A in the i-th shooting, and the estimated amount of light received by the sub-pixel 23 in the previous (i-1)th shooting. , is the value obtained by subtracting the sum of the estimated amount of light received by the sub-pixels 23 at the time of the (i-2)th photographing time before the previous one. In other words,
(Estimated value of the amount of light received by the i-th sub-pixel) = (Total amount of received light of the i-th time) - {(Estimated value of the amount of light received by the (i-1)th sub-pixel) + ((i-2)th estimated value of the amount of light received by the sub-pixel)}.

The numerical values written in the first to third regions 21 to 23 of each pixel 20 in FIG. 5 are examples of the estimated amount of light received in that region, and are the same as the numerical values of the sub-pixel 23 from the previous time or the time before the previous time. The numerical value to the right of the pixel 20 is the total amount of light received by one pixel. In the example of FIG. 5, the total amount of light received for one pixel 6A in the i-th shooting is 3.8, and the estimated amount of light received by the sub-pixel 23 in the previous (i-1)th shooting is 1.3. Since the estimated amount of light received by the sub-pixel 23 during the (i-2)th shooting time is 0.5, the estimated amount of light received by the sub-pixel 23 during the i-th shooting is [3.8-(1 .3+0.5)}=2.0.

However, this process of estimating the amount of light received by the sub-pixel 23 only needs to be carried out for the pixel 20 detected as a defective candidate pixel with a high possibility that a defect exists, and the estimated position of the detected defective candidate pixel is It is sufficient to create a sub-pixel image by setting i based on the information of the drum encode 4, storing the amount of light received by the sub-pixel 23 at that time in association with position information. As a result, the defect determination process can be executed in a concentrated manner in areas where there is a high possibility that surface defects exist, resulting in improved efficiency. Regarding defective candidate pixels, pixels 20 whose total amount of received light exceeds a predetermined threshold may be detected as defective candidate pixels.

Note that for pixels 20 whose total amount of light received does not exceed a predetermined threshold, there is a low possibility that a defect exists, so the average value of the pixel light amount for 2 A is calculated as 1/3 of the amount of light received by the entire pixel. , may be estimated as the amount of light received by the sub-pixel 23 (for example, the (i-4)th or (i-3)th time in FIG. 5).

In addition, for the first imaging after the start of the inspection, since there is no estimated amount of light received by the previous sub-pixel 23, the average value obtained by dividing the amount of received light of the entire pixel by the number of sub-pixels 23 per pixel is calculated for the first image. It is sufficient to estimate the amount of light received by the sub-pixel 23 and use this amount of light to estimate the amount of light received by the sub-pixel 23 in subsequent shooting.

In this way, a sub-pixel image consisting of the amount of light received by 1/3 pixel (an area of 2 A) is created around the defective candidate pixel instead of an image of pixel 20. This increases the resolution of the line sensor 1 by three times, making it possible to detect and discriminate fine surface defects with high accuracy. In other words, when photographing a non-inspection object 5 moving with respect to the line sensor 1 and illumination devices 2a and 2b using the line sensor 1, the distance between the photographing planes of the line sensor 1 and the non-inspection object 5 is not stable, and the object It is necessary to set the depth of field deep, but there is a trade-off relationship where increasing the depth of field lowers the resolution, so it may not be possible to inspect even the smallest defects, but it is possible to use subpixel images smaller than one pixel. This increases resolution without increasing the depth of field, making it possible to inspect even finer defects.

Although a sub-pixel image has been generated for the illumination device 2a, a sub-pixel image of 1/3 pixel can also be generated for the illumination device 2b in the same manner.
<Alignment of subpixel images for illumination devices 2a and 2b>
As shown in FIG. 6, the positions of sub-pixel images for one pixel 20 corresponding to the illumination device 2a are a1, a2, a3..., and the sub-pixel images for one pixel 20 corresponding to the illumination device 2b are Assuming that the positions of are b1, b2, b3..., the object to be inspected 5 is moving relative to the line sensor 1 and the illumination devices 2a, 2b at the time of photographing, so the sub-pixel image and the illumination for the illumination device 2a are The sub-pixel images for the device 2b are alternately shifted by a moving distance A corresponding to the illumination light switching time, such as a1, b1, a2, b2, a3, b3, . . . .

Therefore, this positional shift is corrected. Specifically, if the position of the sub-pixel of the illumination device 2b corresponding to the position a2 of the illumination device 2a is b2',
Amount of light received at position b2' (brightness value) = (amount of light received at position b1 + amount of light received at position b2)/2
The positional shift is corrected by correcting the amount of received light (brightness value). The same applies to the sub-pixel positions b3', b4', . . . of the illumination device 2b corresponding to the positions a3, a4, . . . of the illumination device 2a.

Further, the position of the lighting device 2a may be aligned to correspond to the positions b1, b2, b3, . . . of the lighting device 2b.

The above correction formula is for a case where there are two illumination devices, but a correction formula that can be applied to two, three or more illumination devices is expressed by the following formula.

However, i: index of estimated sub-pixel position
j: Identification number of the lighting device that is lit <Correction when estimating the amount of light received by sub-pixels>
The amount of light received by the sub-pixel 23 was estimated on the assumption that all areas corresponding to one pixel 6A have the same light-receiving sensitivity. However, in reality, as shown in the sensitivity distribution in FIG. 7, the light receiving sensitivity differs depending on each part of the pixel 20, with the central part having relatively high light receiving sensitivity and the both ends having low light receiving sensitivity. In FIG. 7, it is shown that the sensitivity of the hatched portion is high, and the sensitivity of the third region is higher than that of the first region even at both end portions.

As explained in FIG. 5, if the current shooting is the i-th shooting, then in the previous shooting (i-1), the sub-pixel is in the third area 23 at the right end of one pixel. Yes, the light receiving sensitivity is low. In the current i-th photographing, this sub-pixel 23 overlaps with the second region 22 of 2A in the center, and this region has high light-receiving sensitivity. Therefore, the amount of light received by the central second region 22 in the i-th photographing should be greater than the amount of light received by the sub-pixel 23 in the (i-1)th photographing.

Furthermore, the amount of light received by the first area 21 in the i-th imaging is the amount of light received by the sub-pixel 23 in the (i-2)-th imaging; Therefore, the amount of light received by the first region 21 in the i-th photographing should actually be smaller.

Therefore, in order to correct the amount of light received in each region 21 to 23 of the pixel 20, weighting is performed according to each region 21 to 23, and a weighting coefficient is set for each region 21 to 23. Specifically, the weighting coefficient of the first area 21 at the left end of one pixel 20 is ε1, the weighting coefficient of the second area 22 at the center is ε2, and the weighting coefficient of the third area 23 at the right end is ε3, and i The amount of light received by the sub-pixel 23 in the second photographing is calculated using the following formula.
(Estimated value of the amount of light received by the i-th sub-pixel) = (Total amount of received light of the i-th time) - {(Estimated value of the amount of light received by the (i-1)th sub-pixel) * ε2 / ε3 + ((i-2 ) estimated value of the amount of light received by the sub-pixel)*ε1/ε3}
As a specific example of ε1, ε2, and ε3, in this embodiment, ε1=1/3, ε2=1, and ε3=2/3 are set. FIG. 8 shows an example of the corrected light reception amount estimated values for each of the regions 21 to 23 calculated in consideration of weighting.

In the example of FIG. 8, when the amount of light received by the sub-pixel 23 during the (i-2)th photographing is set to 0.3, the amount of light received in the second area 22 during the (i-1)th photographing is The amount of received light is corrected and increased to 0.5, and in the first region 21 at the time of the i-th photographing, the amount of received light is corrected and reduced to 0.2. Furthermore, when the amount of light received by the sub-pixel 23 during the (i-1)th shooting is set to 0.9, the amount of light received in the second area 22 during the i-th shooting is corrected and increases to 1.3. are doing.

Further, since not only the light receiving sensitivity of the pixel 20 but also the distribution of the amount of irradiation light of the illumination devices 2a and 2b differ, as shown in FIG. 9, the amount of received reflected light 40 may also differ depending on the area of the pixel 20. The example in FIG. 9 shows that the amount of reflected light is 1 for the first region 21 and the third region 23 at the ends of the pixel 20, while the amount of reflected light for the second region 22 at the center is twice as much. Therefore, in order to correct the difference in the amount of reflected light, weighting coefficients ε1 to ε3 may be set based on each region 21 to 23 of the pixel 20. For example, ε1=1/2, ε2=1, and ε3=1/2 may be set.

In this way, with the amount of received light corrected for each area 21 to 23 of the pixel 20, the amount of received light in the overlapping area is subtracted from the amount of received light for the entire pixel, so the more accurate amount of received light in the overlapping area is subtracted and The amount of light received by the sub-pixel 23 can be estimated, and as a result, defects can be determined with higher accuracy.
<Defect determination>
Surface defects are determined based on mutually aligned sub-pixel images.

Among the concave defects, for a spherical concave defect 51 called a "void defect," as shown in FIG. The positional relationship between each position of the devices 2a and 2b and the reflection position is reversed. That is, in the aligned sub-pixel image 61, the bright points 61a, 61b corresponding to each illumination device 2a, 2b do not overlap, each bright point 61a, 61b is within a preset range, and is bright. If the positions of the points 61a and 61b are in the opposite positional relationship to the arrangement positions of the illumination devices 2a and 2b, it is determined that the defect is a concave defect 51.

On the other hand, regarding the "convex defect", as shown in FIG. 11, the illumination lights from the illumination devices 2a and 2b to the convex defect 52 do not cross, and the positions of the illumination devices 2a and 2b and the reflection position are in a positional relationship. are the same. Therefore, in the aligned sub-pixel image 62, the bright points 62a, 62b corresponding to each illumination device 2a, 2b do not overlap, each bright point 62a, 62b is within a preset range, and is bright. If the positions of the points 62a and 62b have the same positional relationship as the arrangement positions of the illumination devices 2a and 2b, it is determined that the defect is a convex defect 52.

In addition, in FIGS. 10 and 11, bright points 61a and 62a corresponding to the lighting device 2a are shown by double hatching, and bright points 61b and 62b corresponding to the lighting device 2b are shown by broken line hatching. The same applies to FIG. 12 and subsequent figures.

Among the concave defects, for a plane defect 53 called a "scratch defect", as shown in FIG. The positional relationship between each position and the reflection position is reversed. In addition, since the directions of the scratched surfaces are uneven, the reflection of the illumination light from the illumination device 2a and the reflection of the illumination light from the illumination device 2b occur together. However, since the plane is a high reflectance surface, each illumination light is not mixed and reflected. Therefore, in the aligned sub-pixel image 63, the bright point 63a corresponding to the illumination device 2a and the bright point 63b corresponding to the illumination device 2b do not overlap, and the bright points 63a and 63b are mixed, and the bright point 63a , 63b are opposite to the arrangement positions of the illumination devices 2a, 2b, it is determined that the flaw defect 53 is present on the surface of the object to be inspected 5.

Since the surface of "dust" and "garbage" is a diffuse reflection surface, it diffusely reflects the illumination light from the illumination devices 2a and 2b, and therefore, as shown in FIG. 13, each illumination light is mixed and reflected by the defect 54. be done. Therefore, in the aligned sub-pixel image 64, when the bright points 64a, 64b corresponding to the respective illumination devices 2a, 2b overlap, it is determined that dirt or dust is present on the surface of the inspection object 5.

Each sub-pixel image produced by the illumination devices 2a and 2b is a dark-field image, and uneven defects, scratch defects, dust, dirt, etc. appear as white spots. Detection of defect candidates on an image is performed as follows.

That is, when the size of the defect candidate is set in terms of image area to be W1 or more and W2 or less, and the brightness is set to be B2 or more, each sub-pixel image is binarized at B2, and the discrete pixels are aggregated by expansion/contraction processing. Furthermore, each pixel set is labeled by color coding or the like.

Only the pixel sets with labels where the aggregate area S satisfies W1≦S≦W2 are left, and the other pixel sets are deleted from each sub-pixel image. W1 does not simply indicate the minimum defect size, but also indicates the minimum size that can be considered as a "defect part."

Then, when the size of the defect is defined as the number of pixels of If there is a pixel set within the range of coordinates Vi ± It is classified into four types.

Specifically, as shown in FIG. 14, when aligned sub-pixel images SPb corresponding to illumination device 2a and sub-pixel images SPb corresponding to illumination device 2b are combined as shown in the figure on the right, each illumination The bright points 61a, 61b corresponding to the devices 2a, 2b do not overlap, each bright point 61a, 61b is within the range of coordinates Vi±X/2, and the positions of the bright points 61a, 61b are in the lighting device 2a, Since the positional relationship is opposite to that of 2b, it is determined that it is a void defect.

As shown in FIG. 15, when the aligned sub-pixel image SPb corresponding to illumination device 2a and sub-pixel image SPb corresponding to illumination device 2b are combined as shown in the figure on the right, each illumination device 2a, 2b is The corresponding bright points 62a, 62b do not overlap, each bright point 62a, 62b is within the range of coordinates Vi±X/2, and the positions of the bright points 62a, 62b are the same as the arrangement positions of the lighting devices 2a, 2b. Since they have the same positional relationship, it is determined that it is a convex defect.

As shown in FIG. 16, when the aligned sub-pixel image SPb corresponding to illumination device 2a and sub-pixel image SPb corresponding to illumination device 2b are combined as shown in the figure on the right, each illumination device 2a, 2b is The corresponding bright points 62a, 62b are within the range of coordinates Vi±X/2, the bright points 62a, 62b are mixed without overlapping, and the positions of the bright points 63a, 63b are the locations of the lighting devices 2a, 2b. Since it is the opposite, it is determined that it is a scratch defect.

As shown in FIG. 17, when the aligned sub-pixel image SPb corresponding to illumination device 2a and sub-pixel image SPb corresponding to illumination device 2b are combined as shown in the figure on the right, each illumination device 2a, 2b Since the corresponding bright spots 64a and 64b overlap, it is determined that it is dirt or dust.

As described above, surface defects on the moving object 5 to be inspected can be detected and discriminated.

The detection results are displayed on the display device 6. The display preferably includes the image after alignment of the two sub-pixel images SPa and SPb shown on the right side of FIGS. 14 to 17, as well as the type of defect determined and the range of coordinates Vi±X/2 for each defect. may also be displayed together.

Although one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment. For example, the line sensor 1 and the lighting devices 2a and 2b are fixed and the image is taken while the inspected object 5 is moved. Photographing may be performed while moving the object, as long as at least one of the inspected object 5, the line sensor 1, and the illumination devices 2a and 2b moves relative to the other.

In addition, although the relative moving distance per photographing of the non-inspection object 5 is A, and the length of the sub-pixel 23 is 2A, the current photographing and the previous photographing with respect to one illumination device are It is sufficient if the photographing ranges overlap and sub-pixels can be formed. For this reason, it is preferable that the relative moving distance of the non-inspection object 5 per photographing be set to 1/2 or less of one pixel.

In addition, although two lighting devices 2a and 2b were used, as described above, three or more lighting devices were sequentially switched and used, and three or more types of sub-pixel images corresponding to each lighting device were compared to identify defects. It is preferable to detect and discriminate from the viewpoint that the directions of the irradiated light are diverse and surface defects can be detected and discriminated with higher accuracy.

INDUSTRIAL APPLICATION This invention can be utilized when determining the surface defect of an object to be inspected, such as a product or a component having a surface with strong specular reflection properties.

1 Line sensor 2a, 2b Illumination device 4 Drum encoder 5 Object to be inspected 6 Illumination device 8 Illumination control section 9 Line sensor control section 10 Computer 11 Drum conveyance control section 20 Pixel 21 First area 22 Second area 23 Sub-pixel (third region)
30 Defect 51 Concave defect (void defect)
52 Convex defect 53 Scratch defect 54 Dust or dirt 61a to 64a Bright spots caused by lighting device 2a 61b to 64b Bright spots caused by lighting device 2b

Claims (27)

When the object to be inspected is moved relative to the illumination devices and line sensors arranged at different positions, and the illumination light from each illumination device is switched one by one and irradiated onto the object to be inspected, each Each time the illumination light from the illumination device is switched, a line sensor receives and photographs the reflected light from the object to be inspected, thereby obtaining a plurality of images with respective positions shifted by the switching amount of the illumination light. an image acquisition means;
Aligning means for aligning images corresponding to each illumination device acquired by the image acquisition means;
Discrimination means for determining surface defects of the object to be inspected from the images aligned by the alignment means;
Equipped with
A part of each pixel of the line sensor is an overlapping area where the photographing ranges of the current photographing and the previous photographing overlap due to illumination light from one illumination device being irradiated onto the inspected object,
When the part of one pixel excluding the overlapping area is defined as a subpixel, the amount of light received by the current subpixel is estimated by subtracting the amount of light received by the overlapping area from the amount of light received by the entire pixel in the current shooting. further comprising subpixel image creation means for creating a pixel image;
The alignment means is a surface defect determination device that aligns the sub-pixel images corresponding to each illumination device created by the sub-pixel image creation means . When the object to be inspected is moved relative to the illumination devices and line sensors arranged at different positions, and the illumination light from each illumination device is switched one by one and irradiated onto the object to be inspected, each comprising an image acquisition means for acquiring a plurality of images for each of the illumination lights by receiving and photographing reflected light from the object to be inspected with a line sensor each time the illumination light from the illumination device is switched;
A part of each pixel of the line sensor is an overlapping area where the photographing ranges of the current photographing and the previous photographing overlap due to illumination light from one illumination device being irradiated onto the inspected object,
When the part of one pixel excluding the overlapping area is defined as a subpixel, the amount of light received by the current subpixel is estimated by subtracting the amount of light received by the overlapping area from the amount of light received by the entire pixel in the current shooting. subpixel image creation means for creating a pixel image;
A determining means for determining a surface defect of the object to be inspected based on the sub-pixel image created by the sub-pixel image creating means;
A surface defect discrimination device further equipped with. 3. The surface defect determination apparatus according to claim 2, further comprising positioning means for positioning subpixel images corresponding to each illumination device created by said subpixel image creation means. 4. The surface defect discriminating device according to claim 1 , wherein the sub-pixel image creation means subtracts the amount of light received in the overlapping region from the amount of light received in the entire pixel after correcting the amount of light received in each region. The sub-pixel image creation means calculates the amount of light received in the overlapping area from the sum of the amounts of light received by the sub-pixels estimated before the previous time, subtracts the calculated amount of received light from the amount of light received by the entire pixel, and calculates the amount of light received by the current sub-pixel. The surface defect determination device according to any one of claims 1 to 4 , which estimates the amount of received light. 6. The sub-pixel image creation means estimates the amount of light received by the first sub-pixel as the average value obtained by dividing the amount of light received by the entire first pixel after the start of imaging by the number of sub-pixels per pixel. surface defect determination device. If the amount of light received by the entire pixel does not exceed a predetermined threshold, the sub-pixel image creation means calculates the average value obtained by dividing the amount of light received by the entire pixel by the number of sub-pixels per pixel as the current amount of light received by the sub-pixel. If the amount of light received by the entire pixel exceeds a predetermined threshold, the amount of light received by the current sub-pixel is estimated by subtracting the amount of light received by the overlapping area from the amount of light received by the entire pixel. The surface defect determination device according to any one of the above. 3. The positioning means performs positioning of the sub-pixel images corresponding to each illumination device created by the sub-pixel image creation means by correcting the brightness value Kij to the correction value K'ij according to the following formula. 1 , 3 , the surface defect discriminating device according to any one of claims 4 to 7 , which refers to claim 1 or 3 .

However, i: index of estimated sub-pixel position
j: Identification number of the lighting device that is lit In the sub-pixel image aligned by the alignment unit, if the bright points corresponding to each illumination device do not overlap and each bright point is within a preset range, the discrimination unit determines that the object to be inspected is 9. The surface defect discriminating device according to claim 1 , wherein the device determines that a concave defect or a convex defect exists on the surface of the surface of the substrate. The determining means determines that, in the sub-pixel image aligned by the alignment means, if the position of the bright spot corresponding to each illumination device is opposite to the arrangement position of the illumination device, a concave defect exists, and if the position is not reversed, a concave defect exists. The surface defect discriminating device according to claim 9 , which determines that a convex defect exists. The determining means determines that dirt or dust is present on the surface of the inspected object when bright points corresponding to each illumination device overlap in the sub-pixel images aligned by the alignment means. Claim 1 , Claim 3 , The surface defect discriminating device according to any one of claims 4 to 7 , 8 , 9 , and 10 which refers to claim 1 or 3 . Pixels whose total amount of light reception exceeds a predetermined threshold are detected as defective candidate pixels, the alignment means aligns sub-pixel images of the detected defective candidate pixels, and the determination means detects defects on the surface of the object to be inspected. The surface defect discriminating device according to claim 1 , 3 , or any one of claims 4 to 7 , 8 , 9 , 10 , and 11 referring to claim 1 or 3 . The surface defect determination device according to any one of claims 1 to 12 , wherein an LED or a visible light semiconductor laser is used as a light source of the illumination device. 14. The lighting devices are three or more, and are arranged on a circumference centered on the line sensor with an angular difference of 360 degrees divided by the number of lighting devices. Surface defect determination device. multiple lighting devices placed at different positions;
a line sensor capable of receiving reflected light of the illumination light irradiated onto the inspected object from each illumination device;
a moving means for moving the object to be inspected relative to the illumination device and the line sensor;
illumination control means that switches the illumination light from each illumination device one by one at a predetermined period to irradiate the object to be inspected;
While the moving means moves the object to be inspected relative to the illumination device and the line sensor, the illumination control means controls the amount of light from the object to be inspected each time the illumination light from each illumination device is switched. Line sensor control means for controlling the line sensor so as to receive reflected light and perform photography;
The surface defect determination device according to any one of claims 1 to 14 ,
Appearance inspection equipment equipped with When the object to be inspected is moved relative to the illumination devices and line sensors arranged at different positions, and the illumination light from each illumination device is switched one by one and irradiated onto the object to be inspected, each Each time the illumination light from the illumination device is switched, a line sensor receives and photographs the reflected light from the object to be inspected, thereby obtaining a plurality of images with respective positions shifted by the switching amount of the illumination light. an image acquisition step;
an alignment step of aligning images corresponding to each illumination device acquired in the image acquisition step;
a determination step of determining a surface defect of the object to be inspected from the images aligned in the alignment step;
make the computer run
A part of each pixel of the line sensor is an overlapping area where the photographing ranges of the current photographing and the previous photographing overlap due to illumination light from one illumination device being irradiated onto the inspected object,
When the part of one pixel excluding the overlapping area is defined as a subpixel, the amount of light received by the current subpixel is estimated by subtracting the amount of light received by the overlapping area from the amount of light received by the entire pixel in the current shooting. further causing the computer to perform a sub-pixel image creation step of creating a pixel image;
In the positioning step, the program causes the computer to perform a process of positioning subpixel images corresponding to each lighting device created in the subpixel image creation step . 17. The program according to claim 16, wherein, in the sub-pixel image creation step, the computer executes a process of subtracting the amount of light received in the overlapping region from the amount of light received in the entire pixel after the amount of light received in the overlapping region is corrected for each region . In the sub-pixel image creation step, the amount of light received in the overlapping area is calculated from the sum of the amounts of light received by the sub-pixels estimated before the previous time, and the calculated amount of received light is subtracted from the amount of light received by the entire pixel to calculate the amount of light received by the current sub-pixel. The program according to claim 16 or 17 , which causes the computer to execute a process of estimating the amount of received light. In the sub-pixel image creation step, the computer performs a process of estimating the average value obtained by dividing the amount of light received by the entire first pixel by the number of sub-pixels per pixel as the amount of light received by the first sub-pixel after the start of imaging. The program according to claim 18 , which is executed. In the subpixel image creation step, if the amount of light received by the entire pixel does not exceed a predetermined threshold, the average value obtained by dividing the amount of light received by the entire pixel by the number of subpixels per pixel is calculated as the amount of light received by the current subpixel. If the amount of light received by the entire pixel exceeds a predetermined threshold, the computer executes a process of estimating the amount of light received by the current sub-pixel by subtracting the amount of light received by the overlapping area from the amount of light received by the entire pixel. The program according to any one of claims 16 to 19 . In the alignment step, the sub-pixel images corresponding to each illumination device created in the sub-pixel image creation step are aligned by correcting the brightness value Kij to a correction value K'ij using the following formula. The program according to any one of claims 16 to 20 , which is executed by the computer.

However, i: index of estimated sub-pixel position
j: Identification number of the lighting device that is lit In the discrimination step, if the bright points corresponding to each illumination device do not overlap in the sub-pixel images aligned in the alignment step and each bright point is within a preset range, the inspection object is detected. 22. The program according to claim 16, which causes the computer to execute processing for determining that a concave defect or a convex defect exists on the surface of the computer. In the determination step, in the sub-pixel image aligned in the alignment step, if the position of the bright spot corresponding to each illumination device is opposite to the arrangement position of the illumination device, a concave defect exists, and if it is not reversed, a concave defect exists. 23. The program according to claim 22 , which causes the computer to execute processing for determining that a convex defect exists. In the determination step, if bright points corresponding to each illumination device overlap in the sub-pixel images aligned in the alignment step, it is determined that dirt or dust is present on the surface of the object to be inspected. 24. The program according to claim 16 , which causes the computer to execute the program. A pixel whose total amount of received light exceeds a predetermined threshold value is detected as a defective candidate pixel, and the computer executes the alignment step of aligning sub-pixel images for the detected defective candidate pixel, and the determination step The program according to any one of claims 16 to 24 , which causes the computer to execute a process of determining a surface defect of an object to be inspected. 26. The program according to claim 16 , wherein an LED or a visible light semiconductor laser is used as a light source of the lighting device. 27. The program according to claim 16 , wherein the plurality of lighting devices are arranged at positions of 360 degrees/(number of lighting devices) on a circumference centered on the line sensor.

Images