JP2014109447A - Defect inspection device - Google Patents

Abstract

To provide a defect inspection apparatus capable of accurately erasing inherent noise of a camera-side device from image data captured by a camera and accurately inspecting a defect of an inspection object.
An image processing unit includes a correction data storage unit that stores correction data for correcting inherent noise of a device on the camera, and an image data correction that corrects image data captured by the camera. Part 34. The image data correction unit 34 generates correction image data by performing correction according to the image data for each image data captured by the camera 12 based on the correction data.
[Selection] Figure 1

Description

  The present invention includes a camera that images an inspection object and an image processing unit that processes an image captured by the camera, and inspects a defect of the inspection object based on the image processed by the image processing unit. The present invention relates to a defect inspection apparatus.

  As a conventional defect inspection apparatus, for example, Patent Document 1 includes a CCD camera that captures an image of a solar battery, and a control arithmetic device that processes an image captured by the CCD camera, and an image that has been subjected to image processing by the control arithmetic device. Describes a defect inspection apparatus for inspecting defects of solar cells. In the technique described in Patent Document 1, a difference image is created from two images of a reference solar cell image and a solar cell image to be inspected, and a defect of the solar cell is determined based on the difference image. It is configured as follows. Thereby, it is supposed that the defect of a solar cell can be determined easily.

  For example, Patent Document 2 includes a camera that captures an image of a photovoltaic power generation device, a computer that controls the camera, and a monitor that displays an image captured by the camera, and the captured image is confirmed on the monitor. Thus, a solar power generation element inspection device for inspecting a solar power generation element for defects is described. The technique described in Patent Document 2 does not perform image processing on an image captured by a camera, but acquires a clear image of a solar power generation element by using an autofocus function of the camera or a device of a light source or a filter. It is intended to make it easy to detect a defective portion of an element.

JP 2008-26113 A (refer to FIG. 7, paragraphs “0015” and “0037” in particular) JP 2011-29477 A (refer to FIG. 1, paragraphs “0020” and “0035” in particular)

  The technique described in Patent Document 1 described above determines a solar cell defect based on a difference image created using a reference solar cell image, but serves as a basis for determining a solar cell defect. In some cases, intrinsic noise of camera equipment (eg intrinsic noise caused by the manufacturing process of the camera's image sensor) is reflected in the difference image, and the reflected intrinsic noise accurately inspects solar cell defects. There was a problem that it was difficult to do. When the present inventor made an analysis on this problem, for example, when the imaging condition for imaging the solar cell to be inspected is changed, the difference between the imaging condition for the reference solar cell and the imaging condition for the solar cell to be inspected, etc. Inherent noise cannot be accurately erased on the difference image, and the inherent noise is reflected in the difference image, which is the basis for determining the defect of the solar cell, preventing accurate inspection of the defect of the solar cell. found. In particular, in a defect inspection apparatus that inspects a defect of a solar cell as an object to be inspected, the weak noise emitted from the solar cell is often imaged to inspect the defect of the solar cell. It has also been found that it tends to hinder accurate inspection of solar cell defects. In addition, even when trying to acquire a clear image of the photovoltaic power generation device as in the technique described in Patent Document 2 described above, the above-described inherent noise may be reflected in the image, and the reflected inherent characteristic may be reflected in the image. There has been a problem that noise tends to hinder accurate inspection of defects in solar cells.

  An object of the present invention is to provide a defect inspection apparatus capable of accurately erasing inherent noise of a device on the camera side from image data captured by a camera and accurately inspecting a defect of an inspection object. There is.

A first invention includes a camera that images an inspection object and an image processing unit that processes an image captured by the camera, and inspects a defect of the inspection object based on an image processed by the image processing unit A defect inspection device,
The image processing unit is picked up by the camera based on correction data stored in the correction data storage unit and correction data storage unit that stores correction data for correcting intrinsic noise of the device on the camera side An image data correction unit that corrects image data and generates corrected image data,
The image data correcting unit generates the corrected image data by performing correction according to image data for each image data captured by the camera based on the correction data.

  According to the above configuration, for example, even when the imaging condition for imaging the inspection object is changed, the correction is performed on each piece of image data captured by the camera using the correction data with high accuracy. Corrected image data can be generated, and the inherent noise of the camera device can be accurately erased from the image data captured by the camera. Thereby, it is possible to accurately inspect the defect of the inspection object using the corrected image data from which the inherent noise of the device on the camera side is accurately erased.

  According to a second aspect of the present invention, in the defect inspection apparatus of the first aspect, the correction data is acquired from a correction image captured under a correction data acquisition imaging condition set so that the inherent noise clearly appears on the image. It is data.

  According to the above configuration, the image data captured by the camera can be corrected with higher accuracy using the correction data that accurately and significantly reflects the characteristics of the inherent noise, and the inherent noise can be further accurately corrected from the image data captured by the camera. Can be erased well.

  According to a third aspect of the present invention, in the defect inspection apparatus of the first or second aspect, the correction data includes noise shape data representing the shape of the inherent noise and the distribution of noise intensity for each pixel.

  According to the above configuration, by using the noise shape data, it is possible to correct the image data with high accuracy by applying a correction that matches or substantially matches the inherent noise generated in the image data, and generates corrected image data with higher accuracy. be able to. As a result, the inherent noise can be erased more accurately from the image data captured by the camera.

  According to a fourth aspect of the present invention, in the defect inspection apparatus according to any one of the first to third aspects, the image processing unit calculates a noise intensity of intrinsic noise that appears in image data captured by the camera, and calculates the calculated noise. A correction value calculation unit that calculates a correction value according to the magnitude of the intensity is provided, and the image data correction unit generates the corrected image data using the correction value.

  According to the above configuration, for example, even when the imaging condition for imaging the inspection object is changed, the noise intensity is used for the inherent noise whose intensity that appears on the image data differs depending on the imaging condition etc. Corrected image data can be generated with higher accuracy by performing correction according to the strength of the inherent noise, and the inherent noise of the camera-side device can be erased more accurately from the image data captured by the camera. .

  According to a fifth aspect of the present invention, in the defect inspection apparatus of the fourth aspect, the correction data includes reference pixel coordinates referred to for calculating the noise intensity, and the correction value calculation unit refers to the reference pixel coordinates. Then, a plurality of pixels corresponding to the reference pixel coordinates are selected from among a large number of pixels of the image data captured by the camera, and the noise intensity is calculated based on the luminance of the selected plurality of pixels.

  According to the above configuration, it is possible to quickly calculate the noise intensity of the inherent noise that appears in the image data using the reference pixel coordinates, and to quickly generate corrected image data with high accuracy.

  According to a sixth aspect of the present invention, in the defect inspection apparatus according to any one of the first to fifth aspects of the present invention, the inherent noise is noise caused by a manufacturing process of an image sensor of the camera.

  According to the above configuration, noise caused by the manufacturing process of the camera image sensor can be eliminated, and the noise caused by the manufacturing process of the camera image sensor hardly hinders accurate inspection of the defect of the inspection object. Become.

  A seventh aspect of the present invention is the defect inspection apparatus according to any one of the first to sixth aspects, wherein the inspection object is a solar cell, and the solar cell emits light by applying a current to the solar cell element of the solar cell. Is imaged with the camera in a dark room.

  According to the above configuration, even if the inherent noise is easily reflected in the image captured by the defect inspection apparatus that inspects the defect of the solar cell by imaging the weak light emitted from the solar cell, the image captured by the camera The inherent noise can be erased from the data with high accuracy, and the solar cell can be inspected for defects with high accuracy.

The block diagram which shows the whole structure of the defect inspection apparatus which concerns on this invention Flow chart showing specific contents of correction data acquisition control Explanatory drawing explaining the processing content of a correction data acquisition part Flow chart showing specific contents of inspection image correction control Explanatory drawing explaining the processing content of a correction value calculating part

[Overall configuration of defect inspection system]
The overall configuration of the defect inspection apparatus S according to the present invention will be described with reference to FIG. As shown in FIG. 1, the defect inspection apparatus S controls an imaging unit 1 that images a solar cell panel P as an inspection target, and various devices of the imaging unit 1 and processes an image captured by the imaging unit 1. The control device C includes a display unit 4 that displays an image processed by the control device C, and an input unit 5 that inputs various control commands to the control device C. In the present embodiment, as a defect inspection method for the solar cell panel P, an example in which an EL method for applying a forward bias to the power generation element of the solar cell panel P will be described. The EL method is an abbreviation for the electroluminescence method, and when a forward bias is applied, near infrared rays are emitted from a normal power generation element portion, but from a portion where a defect such as a crack or defective electrode is generated. The property of the power generation element that the near infrared rays are not emitted is observed in a dark room to inspect defects such as cracks in the power generation element and electrode defects.

  The imaging unit 1 is disposed at a position that is a predetermined distance away from the support base 11 that positions the solar cell panel P at a predetermined position and supports the solar cell panel P on the floor, and the solar cell panel P supported by the support base 11. The camera 12 and the light source 13 which irradiates light to the surface of the solar cell panel P are provided. The camera 12 is composed of a digital still camera having an autofocus function, and is supported on the floor portion via a support member 14 composed of a tripod or the like. The light source 13 irradiates the surface of the solar battery panel P with light including wavelengths in the near-infrared band in synchronization with the exposure timing of the camera 12 by the autofocus function, and is supported on the ceiling via a stay or the like. Has been. Various devices of the imaging unit 1 are arranged inside the dark room 15 in order to accurately detect the near-infrared emission state of the solar battery panel P while avoiding the influence of light from the outside.

  The camera 12 includes an image sensor 12A configured by a two-dimensional image sensor such as a silicon-based CMOS or CCD, and a filter 12B disposed on the solar cell panel P side with respect to the image sensor 12A. The filter 12B cuts light other than the wavelength in the near-infrared band, and transmits only light in the near-infrared band wavelength to the image sensor 12A side.

  The solar cell panel P is connected with a current application device 16 as a power source. The current application device 16 applies a forward direct current to the power generation element of the solar cell panel P to emit near infrared rays from the power generation element of the solar cell panel P. In the present embodiment, the current application device 16 is configured to be able to change the applied current value of the direct current applied to the power generation element of the solar cell panel P, and emits near-infrared rays that are emitted from the power generation element of the solar cell panel P. The amount can be changed.

  The control device C includes an imaging unit control unit 2 that controls the camera 12, the light source 13, and the current application device 16 of the imaging unit 1, and an image processing unit 3 that processes an image captured by the camera 12. The imaging unit control unit 2 is connected to the camera control unit 21 that controls the imaging timing and imaging time of the camera 12, the light source control unit 22 that controls the irradiation timing and irradiation time of the light source 13, and the direct current to the power generation element of the solar battery panel P. And a power supply control unit 23 that controls the current application device 16 so as to control the application timing, the application time, and the applied current value for applying the current. The camera control unit 21, the light source control unit 22, and the power supply control unit 23 are comprehensively controlled by the overall control unit 24.

  The overall control unit 24 includes an imaging condition setting unit 25 that sets imaging conditions for imaging by the imaging unit 1. The imaging condition setting unit 25 can set an imaging condition for inspection for imaging the solar battery panel P during actual inspection and an imaging condition for acquiring correction data for acquiring correction data described later. As imaging conditions for inspection, for example, camera imaging conditions such as imaging timing and imaging time, light source irradiation conditions such as irradiation timing and irradiation time, and application timing and application of applying a direct current to the power generation element of the solar battery panel P Examples include current application conditions such as time and applied current value. These imaging conditions can be individually set by an operator manually inputting from the input unit 5 connected to the control device C. The imaging conditions for inspection set by the input unit 5 are stored in the imaging condition setting unit 25, and when the operator inputs an inspection start command from the input unit 5, the overall control unit 24 controls the camera control unit 21 and the light source control unit. 22 and the power supply control unit 23, and these control units control the camera 12, the light source 13, and the current application device 16 so as to satisfy predetermined imaging conditions for inspection. Thereby, the image for a test | inspection of the solar cell panel P is imaged on the predetermined imaging condition for a test | inspection.

  The imaging conditions for acquiring correction data include camera imaging conditions, light source irradiation conditions, current application conditions, and the like, as with the imaging conditions for inspection. These imaging conditions are set in advance as conditions in which the inherent noise N of the device on the camera 12 side clearly appears on the captured image. Here, as the inherent noise N possessed by the device on the camera 12 side, for example, inherent noise caused by the manufacturing process of the image sensor 12A of the camera 12 (for example, an error in the mounting position or a fine shape due to a manufacturing error or the like occurs. Fixed pattern noise generated in the manufacturing process due to errors), inherent noise due to output unevenness of the image sensor 12A of the camera 12, inherent noise due to manufacturing error of the filter 12B, and the like. In short, it is various noises that enter an image captured by a device interposed between the solar cell panel P as the inspection object and the image sensor 12A. Unlike the imaging conditions for inspection, the imaging conditions for acquiring correction data are set in advance and stored in the imaging condition setting unit 25, and the overall control unit is input by an operator inputting a correction data acquisition command from the input unit 5. 24, output commands are output to the camera control unit 21, the light source control unit 22, and the power supply control unit 23, so that these control units satisfy predetermined imaging conditions for acquiring correction data. The application device 16 is controlled. As a result, an image of the solar cell panel PA is captured under a predetermined correction data acquisition imaging condition. In addition, when acquiring correction data, in order to image the solar cell panel without a defect in order to detect the intrinsic noise N accurately, this solar cell panel without a defect which becomes this reference | standard panel becomes a test object A symbol PA different from that of the panel P is given.

  The image processing unit 3 stores an image input processing unit 31 that converts the output value output from the image sensor 12A of the camera 12 into image data, and correction data for correcting the inherent noise N of the device on the camera 12 side. A correction data storage unit 32; a correction value calculation unit 33 for calculating a correction value R based on the correction data stored in the correction data storage unit 32 and the inspection image data Da input by the image input processing unit 31; An image data correction unit 34 that corrects the inspection image data Da input by the image input processing unit 31 based on the correction value R calculated by the correction value calculation unit 33 and generates corrected image data; The correction data acquisition unit 35 that acquires correction data based on the correction image data input by the processing unit 31 and the corrected image data generated by the image data correction unit 34 And a display image generating unit 36 for generating a display image to be displayed on the display unit 4 Zui.

  The control device C is connected to an input unit 5 configured with a keyboard, switches, and the like, and a display unit 4 configured with a display or the like. The operator operates the defect inspection apparatus S at the input unit 5, and the actual visual defect inspection by the operator is performed based on the display image displayed on the display unit 4.

[Details of correction data acquisition control]
Specific contents of the correction data acquisition control by the control device C will be described with reference to FIGS. First, as a preparatory work for obtaining correction data, an operator positions and supports a solar cell panel PA having no defect serving as a reference panel on the support base 11. Next, as shown in FIG. 2, when the operator inputs a correction data acquisition command from the input unit 5 (step S <b> 1, YES), the overall control unit 24 to the camera control unit 21, the light source control unit 22, and the power supply control unit 23. The control unit controls the camera 12, the light source 13, and the current application device 16 so as to satisfy a predetermined correction data acquisition imaging condition. As a result, a correction image of the solar battery panel PA is captured under predetermined correction data acquisition imaging conditions (step S2). When the capturing of the correction image of the solar battery panel PA is completed, the image input processing unit 31 inputs the output value output from the image sensor 12A of the camera 12 into image data, and the input correction image data is processed. It outputs to the correction data acquisition part 35 (step S3). The correction data acquisition unit 35 acquires correction data according to the procedure shown in FIG. 3 based on the acquired correction image data (step S4). In the present embodiment, an example will be described in which correction data for the gate-shaped inherent noise N is acquired when the gate-shaped inherent noise N resulting from the manufacturing process of the image sensor 12A occurs. Such a gate-shaped intrinsic noise N is considered to be generated by an adhesive that adheres the image sensor 12A to the substrate. The position and size of each camera 12 differs, and the noise inherent to the camera is different in imaging conditions. In this embodiment, noise specific to the camera is corrected by image processing.

  As illustrated in FIG. 3A, the correction data acquisition unit 35 acquires noise shape data D as correction data from the correction image data input by the image input processing unit 31. The noise shape data D is data obtained by digitizing the shape of the inherent noise N and the noise intensity distribution for each pixel. Here, “noise intensity” is a numerical value of the in-plane distribution of the noise intensity of the pixel as a value representing the relative noise intensity of the pixel of the image data for correction. This is calculated from the luminance of the pixel of the image data by a predetermined arithmetic expression. The “luminance” of the pixel of the correction image data indicates the luminance in the HLS color space (the same applies to the inspection image data Da described later). Therefore, in the noise shape data D, the noise intensity of the pixel is expressed as a larger value as the inherent noise N is larger. For example, in FIG. 3A, the position of the pixel having the noise intensity of “1.0” is represented. In the position of the pixel having the largest intrinsic noise N and the noise intensity “0.0”, the intrinsic noise N does not exist. In FIG. 3A, a part of the noise shape data D of the inherent noise N is enlarged and displayed. Also, the hatched portion indicated by N in FIG. 3A represents a portion where the inherent noise N in the noise shape data D is present by hatching.

  Next, as shown in FIG. 3B, the correction data acquisition unit 35 uses a plurality of (n, 17 in the example of FIG. 3B) reference pixels as correction data from the noise shape data D that has already been acquired. The coordinates En (n = 1, 2, 3...) Are acquired. The reference pixel coordinate En is a coordinate of a pixel selected from a pixel group represented as a place where the inherent noise N exists in the pixel of the noise shape data D. In FIG. 3B, one reference pixel coordinate En is enlarged and displayed. In this embodiment, as the plurality of reference pixel coordinates En, a pixel at each corner of the inherent noise N and a pixel at the end of the inherent noise N are selected, and the inherent noise is determined from a pixel group between these pixels. Pixels are selected at predetermined intervals in the longitudinal direction of N. One reference pixel coordinate En is a coordinate of a peak position An where the noise intensity is highest in the short direction crossing the inherent noise N, and a peripheral portion of the inherent noise N in the short direction crossing the inherent noise N. It consists of a pair (An, Bn, Cn) with the coordinates of two base positions Bn, Cn. As a method for selecting the reference pixel coordinate En, it is only necessary to select a pixel at a representative position on the inherent noise N, and it may be selected at equal intervals or set at unequal intervals. In addition, when the shape of the inherent noise N is not a gate shape but different shapes such as a linear shape, an L shape, a U shape, or a rectangle, for example, an end portion or a corner portion of the inherent noise according to the shape of the inherent noise. These pixels may be selected, and pixels may be selected at predetermined intervals from a pixel group between these pixels. In the case where the shape of the inherent noise N is a dot shape in which a plurality of pixels are gathered, for example, the central pixel of the inherent noise may be selected, and the outer peripheral pixels of the inherent noise may be selected at a predetermined interval.

  The reference pixel coordinate En is set in advance by setting a predetermined interval for selecting the reference pixel coordinate En in the correction data acquisition unit 35, a threshold value of noise intensity as to whether it is a peripheral portion of the inherent noise N, and the like. The data acquisition unit 35 may be configured to automatically select the reference pixel coordinates En. Further, the noise shape data D is displayed on the display unit 4 via the display image generation unit 36, and the operator operates the input unit 5 during the noise shape data D displayed on the display unit 4 to manually reference pixel coordinates. It may be configured such that En is designated and the manually designated reference pixel coordinate En is stored in the correction data storage unit 32 together with the noise shape data D.

  Next, as illustrated in FIG. 3C, the correction data acquisition unit 35 acquires pixel axis coordinates F as correction data from the already acquired noise shape data D. The pixel axis coordinate F is the coordinate of the noise center pixel closest to the target pixel in the noise shape data D. Here, the noise center pixel is a pixel having the highest noise intensity in the short direction crossing the inherent noise N in the pixel group having the inherent noise N in the pixels of the noise shape data D. In this embodiment, as shown by a circle in FIG. 3C, the pixel column or row passes through the center of the inherent noise N. Here, an example of obtaining the pixel axis coordinate F will be described in detail. For example, as shown by a square mark in FIG. 3C, if the pixel of interest is a pixel having a noise intensity of “0.2”, the pixel Is obtained as the pixel axis coordinate F. For example, if the pixel of interest is a pixel having a noise strength of “0.4”, the noise closest to that pixel is obtained. The coordinates of the noise center pixel having the strength “1.0” are acquired as the pixel axis coordinates F. In short, the coordinates of the noise center pixel closest to each pixel of interest are acquired as the pixel axis coordinates F. It should be noted that the pixel axis coordinates F are all the pixels where the inherent noise N is present in the noise shape data D (for example, other than those having a noise intensity of “0.0” among the pixels of the noise shape data D). Pixel) is extracted as a target pixel.

  As shown in FIG. 2, the correction data acquired by the correction data acquisition unit 35, that is, the noise shape data D, the reference pixel coordinates En, and the pixel axis coordinates F are output from the correction data acquisition unit 35 to the correction data storage unit 32. And stored in the correction data storage unit 32 (step S5). The correction data stored in the correction data storage unit 32 is displayed on the display unit 4 from the correction data storage unit 32 via the display image generation unit 36 when the operator operates the input unit 5, and the correction data is stored. The content can be confirmed, and the content of the correction data stored in the correction data storage unit 32 can be corrected or changed by the operator operating the input unit 5.

  In principle, the correction data acquisition control described above may be executed once at the time of manufacture or shipment of the defect inspection apparatus S. However, when the camera 12 or the parts of the camera 12 are replaced, the support table 11 is used. Positioning and supporting the solar panel PA having no defect as a reference panel, and the operator inputting a correction data acquisition command from the input unit 5 (step S1), the above-described series of correction data acquisition control is executed again. (Steps S2 to S5).

  As described above, the control device C is configured to execute correction data acquisition control, whereby correction data used for inspection image correction control, which will be described later, can be acquired quickly with a simple operation, and the camera 12 and the camera 12 Re-acquisition of correction data in the case where twelve parts are replaced can be performed easily and quickly.

[Specific contents of inspection image correction control]
The specific contents of the inspection image correction control by the control device C will be described with reference to FIGS. First, an operator positions and supports the solar cell panel P to be inspected on the support base 11 as a preparation operation for defect inspection. Next, as shown in FIG. 4, when the operator inputs an inspection start command from the input unit 5 (step S <b> 10, YES), the overall control unit 24 sends the camera control unit 21, the light source control unit 22, and the power supply control unit 23. An output command is output, and these control units control the camera 12, the light source 13, and the current application device 16 so as to satisfy predetermined imaging conditions for inspection. As a result, an inspection image of the solar cell panel P is captured under predetermined inspection imaging conditions (step S11). When the imaging of the inspection image of the solar battery panel P is completed, the image input processing unit 31 inputs the output value output from the imaging element 12A of the camera 12 into image data, and the inspection image data Da that has been input processed. Is output to the correction value calculation unit 33 and the image data correction unit 34 (step S12). The correction value calculation unit 33 reads the correction data stored in the correction data storage unit 32 and corrects the inspection image data Da according to the procedure shown in FIG. 5 based on the acquired inspection image data Da and the read correction data. The correction value R to be calculated is calculated (step S13).

  As shown in FIG. 5A, the correction value calculator 33 calculates a noise intensity K from the inspection image data Da and the reference pixel coordinates En of the correction data. In FIG. 5A, one reference pixel coordinate En is enlarged and displayed, and a portion corresponding to the inherent noise N in the inspection image data Da is displayed as Na. Here, the noise intensity K is an evaluation value for evaluating how much the inherent noise N is reflected in the inspection image actually taken, and the luminance of the portion where the inherent noise N exists in the inspection image data Da. The degree of amplification is calculated by the following (Equation 1). That is, as shown in FIG. 5A, the correction value calculation unit 33 performs the luminance Xn of the coordinates of the peak position An of the reference pixel coordinate En and the luminance Yn of the coordinates of the base positions Bn and Cn in the inspection image data Da. , Zn are automatically measured, and the difference between the luminance Xn of the coordinates of the peak position An and the average value of the luminances Yn, Zn of the coordinates of the base positions Bn, Cn is automatically calculated, and these measurements and calculations are performed. The calculation is performed for all the reference pixel coordinates En, and the average value of the calculated values is automatically calculated. This average value is the noise intensity K (“K = 25” in the example of FIG. 5A), and is a single calculated value.

  Next, as shown in FIG. 5B, the correction value calculation unit 33 calculates the standard luminance L from the inspection image data Da and the reference pixel coordinates En of the correction data. In FIG. 5B, a part of the inspection image data Da is enlarged and a portion corresponding to the inherent noise N in the inspection image data Da is displayed as Na. Here, the reference brightness L is a reference brightness for calculating the correction value R of each pixel to be corrected, and the brightness Xn of the coordinate at the peak position An of the reference pixel coordinate En in the inspection image data Da. Calculate based on That is, as indicated by the thick circle in FIG. 5B, the correction value calculation unit 33 sets all the reference pixel coordinates En in the inspection image data Da (“17” in the example of FIG. 5B). The average value obtained by removing the maximum value and the minimum value from the luminance Xn (in the example of FIG. 5B, “183”, “199”, “180”... The brightness L is automatically calculated (“L = 192” in the example of FIG. 5B). This reference luminance L is also a single calculated value as with the noise intensity K.

  Next, as shown in FIG. 5C, the correction value calculator 33 calculates the noise intensity K and the reference luminance L that have already been calculated, the image data for inspection Da, the noise shape data D of the correction data, and the pixel axis coordinates. From F, the correction value R is calculated using the following (formula 2). In FIG. 5C, a part of the inspection image data Da and a part of the noise shape data D are displayed. In other words, if the noise intensity of the noise shape data D is the correction value R of the pixel having the noise shape data D of “0.2”, the correction value calculation unit 33 uses the pixel of the inspection image data Da corresponding to the pixel (the luminance is “167”). The brightness of the pixel axis coordinate F of “pixel”, that is, the axis coordinate brightness M is obtained (“178” in the example of FIG. 5B), and in addition to this axis coordinate brightness M, the already calculated noise intensity K ( = “25”), the reference luminance L (= “192”), and the noise intensity (= “0.2”) of the noise shape data D are substituted into (Expression 2), and the correction value R of the pixel (= “4.6”) is automatically calculated. These operations are performed on all pixels extracted as pixels having inherent noise N from the pixels of the noise shape data D (for example, the noise intensity of the pixels of the noise shape data D is “0.0”. And the correction value R of all the pixels to be corrected is automatically calculated. FIG. 5D shows a part of the calculation result of the correction value R of all the pixels to be corrected. The correction value R indicated by a circle in FIG. This is a correction value R of a pixel having a noise intensity of “0.2” in the shape data D. Then, the correction value calculation unit 33 outputs the calculated correction value R to the image data correction unit 34.

  As shown in FIG. 4, the image data correction unit 34 subtracts the correction value R from the inspection image data Da input by the image input processing unit 31 to generate corrected image data (step S14). That is, the value of the correction value R corresponding to each pixel is subtracted from the luminance of each pixel of the inspection image data Da. Then, the image data correcting unit 34 outputs the generated corrected image data to the display image generating unit 36, and the display image generating unit 36 generates a display image (step S15), and the corrected image obtained by correcting the inherent noise N is obtained. It is displayed on the display unit 4 (step S16).

  Next, when a corrected image in which the inherent noise N is corrected is displayed on the display unit 4, the operator checks the corrected image and inspects the solar cell panel P for defects. Then, the inspected solar cell panel P is removed from the support base 11 and the support base 11 is positioned and supported for the solar cell panel P to be newly inspected, and when the operator inputs an inspection start command from the input unit 5 (step) S10) The above-described series of inspection image correction control is executed again (steps S11 to S16). Thereby, the defect inspection of the solar cell panel P can be continuously and efficiently performed.

  As described above, the control device C is configured to execute the inspection image correction control, whereby the inspection imaging conditions for imaging the solar battery panel P (for example, the above-described camera imaging conditions, light source irradiation conditions, current application) Even if the condition) is changed, the inherent noise N can be accurately erased from the inspection image data Da, and the defect inspection of the solar battery panel P is performed from the corrected image from which the inherent noise N has been accurately erased. Can be performed with high accuracy. In addition, a series of inspection image correction control can be executed by a simple operation by simply operating the input unit 5, and a corrected image can be quickly displayed on the display unit 4 by a simple control process.

[Another embodiment]
(1) In the above embodiment, an example in which the noise shape data D, the reference pixel coordinate En, and the pixel axis coordinate F are adopted as the correction data stored in the correction data storage unit 32 has been described. One of these correction data For example, only the noise shape data D may be adopted as the correction data, and the correction value may be calculated based on the noise shape data D and the inspection image data Da. Specifically, for example, the sum of the noise intensities of all the pixels of the noise shape data D or the sum of the noise intensities (W1) of all the pixels where the inherent noise N of the noise shape data D exists, and for inspection The sum of the luminances of all the pixels of the image data Da or the sum of the luminances of all the pixels where the inherent noise N of the inspection image data Da exists (W2) is calculated, and the ratio of these sums (W2 / W1) ) As the amplification factor of the inherent noise N, the amplification factor may be multiplied by the noise shape data D to calculate a correction value, and the correction value may be subtracted from the inspection image data Da. In this case, the calculation process takes some time, but it is possible to calculate a correction value with high accuracy in consideration of the luminance values of all the pixels or the luminance values of all the pixels where the inherent noise N exists. It becomes.

(2) In the above embodiment, the noise shape data D is obtained by digitizing the shape of the inherent noise N and the noise intensity distribution for each pixel, and the pixel of the correction image data is used as the noise intensity of each pixel. Although the example using the numerical value of the in-plane distribution of the noise intensity is shown, if the characteristic of the shape of the intrinsic noise N and the noise intensity distribution appear, for example, the correction image data The luminance of the pixel multiplied by a constant, the luminance obtained by subtracting a predetermined value from the luminance of the pixel of the correction image data, or the luminance of the pixel of the correction image data itself may be used.

(3) In the above embodiment, only the pixels having the inherent noise N are extracted from the pixels of the noise shape data D, and the corrected image data is generated by correcting the extracted pixels. Although the example configured as described above is shown, the corrected image data may be generated by correcting all of the pixels of the noise shape data D.

(4) In the above embodiment, the control device C is configured to execute the correction data acquisition control. However, the correction data acquisition unit 35 and the like are abolished and the correction data acquisition control is not executed. May be. In this case, the correction data acquired in the same procedure as described above using a measuring device (not shown) for acquiring correction data different from the defect inspection apparatus S is input from the input unit 5 to the control device C. What is necessary is just to comprise so that it may memorize | store in the correction data memory | storage part 32. FIG.

(5) In the above embodiment, an example is shown in which the operator visually inspects the defect based on the display image displayed on the display unit 4. However, the controller C has the defect of the solar battery panel P. And a determination unit (not shown) for determining the presence / absence and defect state (defect type, size, etc.) using a predetermined threshold value, etc., and this determination unit is automatically based on the corrected image data It may be configured to perform defect determination and display the determination result by the determination unit on the display unit 4.

(6) The defect inspection apparatus S shown in the above embodiment is shown as an example. The defect inspection apparatus S includes an upper lid that can be freely opened and closed at the top of the main body, and a solar cell panel P is placed between the main body and the upper lid. The present invention can be similarly applied to a different type of defect inspection apparatus such as a so-called copying machine type defect inspection apparatus (not shown) for inspecting a defect of a battery panel.

(7) In the above embodiment, the case where the EL method in which a forward bias is applied to the power generation element of the solar cell panel P is used as a defect inspection method for the solar cell panel P is described as an example. Different defect inspection methods such as a PL (photoluminescence) method in which light is emitted using a light emitting element such as an LED and light emission from the resulting solar cell panel is detected by a camera or the like may be employed.

  The defect inspection apparatus of the present invention can be used as a defect inspection apparatus for inspecting defects of electronic panels such as an organic EL panel and a touch panel in addition to the solar battery panel P as an object to be inspected.

3 Image Processing Unit 12 Camera 12A Image Sensor 15 Dark Room 32 Correction Data Storage Unit 33 Correction Value Calculation Unit 34 Image Data Correction Unit D Noise Shape Data (Correction Data)
Da Image data for inspection (image data)
En Reference pixel coordinates (correction data)
F Pixel axis coordinates (correction data)
K Noise intensity N Intrinsic noise P Solar panel (inspection object)
R correction value S Defect inspection system

Claims (7)

A defect inspection apparatus that includes a camera that images an inspection object and an image processing unit that processes an image captured by the camera, and inspects a defect of the inspection object based on an image subjected to image processing by the image processing unit. There,
The image processing unit is picked up by the camera based on correction data stored in the correction data storage unit and correction data storage unit that stores correction data for correcting intrinsic noise of the device on the camera side An image data correction unit that corrects image data and generates corrected image data,
The defect inspection apparatus, wherein the image data correction unit generates the corrected image data by performing correction according to image data for each image data captured by the camera based on the correction data.   The defect inspection apparatus according to claim 1, wherein the correction data is data acquired from a correction image imaged under a correction data acquisition imaging condition set so that the inherent noise appears clearly on the image.   The defect inspection apparatus according to claim 1, wherein the correction data includes noise shape data representing the shape of the inherent noise and the distribution of noise intensity for each pixel. The image processing unit includes a correction value calculation unit that calculates the noise intensity of the inherent noise that appears in the image data captured by the camera, and calculates a correction value according to the calculated noise intensity.
The defect inspection apparatus according to claim 1, wherein the image data correction unit generates the corrected image data using the correction value. The correction data includes reference pixel coordinates referred to for calculating the noise intensity,
The correction value calculation unit selects a plurality of pixels corresponding to the reference pixel coordinates from among a large number of pixels of image data captured by the camera with reference to the reference pixel coordinates, and selects the plurality of selected pixels. The defect inspection apparatus according to claim 4, wherein the noise intensity is calculated based on luminance.   The defect inspection apparatus according to claim 1, wherein the inherent noise is noise caused by a manufacturing process of an image sensor of the camera. The said test target object is a solar cell, The said solar cell which light-emitted by applying an electric current to the solar cell element of the said solar cell is imaged with the said camera in a dark room. Defect inspection equipment.

Images