JP5422620B2 - Defect detection apparatus and method for patterned retardation film, and manufacturing method - Google Patents

Description

  The present invention relates to a defect detection apparatus and method for a patterned retardation film, and a manufacturing method.

  One of the display methods of a three-dimensional (3D) television whose market has been expanding in recent years is a space division method. In the space division method, only the image light for the right eye in the odd-numbered column of the display screen, for example, reaches only the image light for the left eye in the even-numbered column to the left eye. As a display method based on such a space division method, one in which a patterned retardation film is arranged on the surface of a display panel is known. A typical patterned retardation film has a function of a quarter wave plate.

  The above-mentioned patterned retardation film has two stripe-shaped retardation regions arranged alternately at a fixed pitch, and each of the two retardation regions has an optical axis (fast axis or slow axis). Have a λ / 4 wavelength plate function. Image light of linearly polarized light is incident on one of the phase difference areas, and the even number of lines are incident on the other phase difference area. Thus, for example, odd-numbered image light is converted into clockwise circularly polarized light, and even-numbered image light is converted into counterclockwise circularly polarized light and emitted. When the observer sees through circularly polarized glasses with different polarization rotation directions on the left and right, only the right-eye image in one row is delivered to the right eye, and only the left-eye image is delivered to the left eye. As a result, a 3D image is observed.

  Although the manufacturing process of the patterned retardation film is under strict control, it is difficult to eliminate defects completely due to various factors in the manufacturing process. Defects include foreign matter, uneven light distribution, and scratches. For this reason, it is necessary to carry out so-called on-line inspection for performing inspection on the production line, grasp the position of the patterned phase difference where the above-described defect has occurred, and remove the defective portion. In some cases, it is necessary to remove the cause of the defect.

  There is currently no known defect detection technique for patterned retardation films. In addition, from the viewpoint of a transparent film or a retardation film, for example, defect inspection apparatuses such as Patent Documents 1 and 2 are known. The inspection apparatus of Patent Document 1 uses a transparent film as an inspection object, places the transparent film between the first and second polarizing plates, emits light through the first polarizing plate, and is emitted from the transparent film. Light is received through the second polarizing plate. Moreover, the inspection apparatus of patent document 2 makes a retardation film into a test | inspection object, makes a phase difference film arrange | positioned between two polarizing plates similarly to patent document 1, and receives light with a light receiver. In the inspection apparatus of Patent Document 2, the polarizing plates are arranged in a crossed Nicol arrangement, the vertical light receiving angle θ1 of the light receiver is 0 ° <θ1 <90 °, and the parallel light receiving angle θ2 is 0 ° <θ2 <90 °. By doing so, it is possible to detect a retardation defect due to retardation in the thickness direction of the retardation film.

Japanese Patent Laid-Open No. 6-148095 JP 2008-267991 A

  By the way, as described above, the patterned retardation film has a large number of alternately provided retardation regions having different optical axis directions. For this reason, the luminance observed in each phase difference region is different only by arranging the polarizing plates as in Patent Documents 1 and 2. For this reason, there is a problem that defects cannot be detected with high accuracy.

  The present invention has been made in view of the above problems, and provides a defect detection apparatus and method for a patterned retardation film and a production method capable of accurately detecting defects in the patterned retardation film. With the goal.

In order to achieve the above object, in the defect detection apparatus for the patterned retardation film of the present invention, the first and second stripes each functioning as a λ / 4 wavelength plate and whose slow axes are substantially orthogonal to each other. a first and second polarizer retardation region is crossed Nicols so as to sandwich the patterned retardation film in which a plurality aligned alternately arranged, the inspection light on the patterned retardation film through said first polarizer A light source unit that irradiates the image, a photographing device that photographs the patterned retardation film through the second polarizing plate to obtain a luminance image, and a defect detection unit that detects a defect from the luminance image, 1 and the second polarizer, in a state where one of polarization transmission axis is substantially parallel with either one of the optical axes of the first and second retardation region, the first and second retardation regions normal Taken with the camera As each luminance becomes the same level in the extinction state near is obtained by adjusting the direction of the one polarization transmission axis.

When the first and second retardation regions are formed by laminating a retardation layer on a support made of a transparent film, the patterned retardation film and the patterned position are formed. It is preferable to arrange a retardation compensator that cancels the retardation characteristics of the support between the retardation film or between the second polarizing plate and the patterned retardation film. As the retardation compensator, it is convenient to use the same transparent film as the support.

A binarizing circuit shadows have been the luminance image is binarized with a predetermined threshold value, for each pixel in the dark pixel is lower than or more bright pixels and the threshold threshold shooting, a plurality of light pixels an image processing unit which have a defect candidate extracting circuit for an area defect candidate regions linked is, the defect detecting section, to the defect candidate region, between the other adjacent the defect candidate regions The number of pixels is counted in the direction in which the boundary line in the luminance image corresponding to the boundary between the first and second phase difference areas is aligned. If the counted number of pixels is equal to or less than a predetermined value, the defect candidate area is determined as a defect. It is preferable that

  An image processing unit that erases a boundary line in the luminance image corresponding to a boundary between the first and second phase difference regions, and the defect detection unit detects a defect based on the luminance image from which the boundary line is erased. It is preferable to detect.

  Further, the image processing unit binarizes the captured luminance image with a predetermined threshold value, and makes each pixel a bright pixel equal to or higher than the threshold value and a dark pixel lower than the threshold value. And a contraction processing circuit for performing contraction processing for contracting bright pixel regions in the direction in which the boundary lines are arranged on the binarized luminance image by the number of times corresponding to the width of the boundary lines, and erasing the boundary lines It is preferable to have.

  The image processing unit performs an expansion process for expanding the bright pixel area in the same direction as the contraction process on the contracted luminance image, and sets the bright pixel area remaining in the contraction process to a size before the contraction process. It is preferable that an expansion processing circuit to be returned is included, and the defect detection unit detects a defect based on the luminance image subjected to the expansion processing.

  In addition, the image processing unit, with respect to a pixel on the luminance image, uses a pixel separated from the pixel by at least the number of pixels according to the boundary line interval in a direction in which the boundary line is arranged as a counterpart pixel. Difference image generation that generates a difference image in which the boundary line is substantially eliminated by performing difference processing for each pixel on the luminance image by subtracting the luminance of the other pixel from the luminance to a new value of the pixel. It is preferable to have a circuit and a binarization circuit that binarizes the difference image with a predetermined threshold value and makes each pixel a bright pixel that is equal to or higher than the threshold value and a dark pixel that is lower than the threshold value.

  The difference image generation circuit generates a difference image in each direction by performing difference processing between the other pixels in different directions on the luminance image, and the binarization circuit includes the plurality of differences. Preferably, each of the images is binarized with a predetermined threshold value to generate a plurality of binarized images, and the defect detection unit performs defect detection on the plurality of binarized images.

In the defect detection method for a patterned retardation film of the present invention, stripes each functioning as a λ / 4 wavelength plate and whose slow axes are substantially orthogonal to each other between the first and second polarizing plates arranged in crossed Nicols. When the patterned retardation film is irradiated with inspection light through the first polarizing plate in a state in which a patterned retardation film in which a plurality of first and second retardation regions are arranged alternately is arranged. , as the luminance of the first and second retardation regions normal to be photographed through the second polarizing plate is the same level in the light off state near either one of the first and second polarizing plates An adjustment step of adjusting the direction of the polarization transmission axis of the first and second retardation regions in a range substantially parallel to one of the optical axes of the first and second phase difference regions; the path between the second polarizer A patterned retardation film is provided, the patterned retardation film is irradiated with inspection light through the first polarizing plate, and a luminance image is obtained by photographing the retardation film through the second polarizing plate. And a detection step of detecting a defect based on the acquired luminance image.

When the patterned retardation film has the first and second retardation regions formed by laminating a retardation layer on a support made of a transparent film, in the adjustment step and the photographing step, A retardation compensation plate is disposed between the first polarizing plate and the patterned retardation film or between the second polarizing plate and the patterned retardation film to cancel the retardation characteristics of the support. It is preferable. As the retardation compensator, it is convenient to use the same transparent film as the support.

A binarization step shadows have been the luminance image is binarized with a predetermined threshold value, for each pixel in the dark pixel is lower than or more bright pixels and the threshold threshold shooting, a plurality of light pixels A defect candidate extraction step in which a region connected to each other is defined as a defect candidate region, and the detection step determines the number of pixels between the defect candidate region and another adjacent defect candidate region as the first. In addition , counting in the direction in which the boundary lines in the luminance image corresponding to the boundary of the second phase difference region are arranged, and when the number of counted pixels is equal to or less than a predetermined value, it is preferable that the defect candidate region is a defect.

  And an erasing step of erasing a boundary line in the luminance image corresponding to a boundary between the first and second phase difference regions, wherein the detecting step detects a defect based on the luminance image from which the boundary line is erased. It is good to do.

  The erasing step binarizes the captured luminance image with a predetermined threshold value, and makes each pixel a bright pixel equal to or higher than the threshold value and a dark pixel lower than the threshold value; and A shrinking step of shrinking a region of bright pixels in a direction in which the border lines are arranged, performing a shrinking process on the binarized luminance image at a number of times corresponding to the width of the border line, and erasing the border line. Is preferred.

  The erasing step performs an expansion process for expanding the bright pixel area in the same direction as the contraction process on the luminance image subjected to the contraction process, and returns the area of the bright pixel remaining in the contraction process to the size before the contraction process. Preferably, the method includes an expansion processing step, and the detection step detects a defect based on the luminance image subjected to the expansion processing.

  In the erasing step, with respect to a pixel on the luminance image, a pixel separated from the pixel by at least the number of pixels corresponding to the boundary line interval in the direction in which the boundary line is arranged is used as a partner pixel from the luminance of the pixel A difference image generation step for generating a difference image in which the boundary line is substantially eliminated by performing a difference process for each pixel on the luminance image by subtracting the luminance of the pixel as a new value of the pixel; It is preferable to have a binarization step that binarizes the difference image with a predetermined threshold value and makes each pixel a bright pixel that is equal to or greater than the threshold value and a dark pixel that is lower than the threshold value.

  The difference image generation step generates a difference image in each direction by performing a difference process between the other pixels in different directions on the luminance image, and the binarization step includes the plurality of differences. It is preferable that each of the images is binarized with a predetermined threshold value to generate a plurality of binarized images, and the detection step performs defect detection on the plurality of binarized images.

In the method for producing a patterned retardation film of the present invention, a plurality of stripe-shaped first and second retardation regions, each of which functions as a λ / 4 wavelength plate and whose slow axes are substantially perpendicular to each other, are arranged alternately. and a production step of producing a patterned retardation film, arranged said patterned retardation film between the first and second polarizer direction is adjusted in polarization transmission axis, said first polarizer Irradiating inspection light onto the patterned retardation film via the second polarizing plate, obtaining a luminance image by photographing the retardation film via the second polarizing plate, and detecting defects based on the obtained luminance image a detection step of, is performed prior to the acquisition step, wherein the first and second polarizing plates, in a state which arranged the patterned retardation film, the putter through the first polarizer When the inspection retardation film is irradiated with the inspection light, the luminance of the normal first and second retardation regions photographed through the second polarizing plate is the same level in the vicinity of the extinction state. An adjustment step of adjusting the direction of the polarization transmission axis of one of the first and second polarizing plates within a range substantially parallel to the optical axis of one of the first and second retardation regions. is there.

  According to the present invention, the patterned retardation film is irradiated with the inspection light through the first polarizing plate in a state where the patterned retardation film is arranged between the first and second polarizing plates in the crossed Nicols arrangement. When a defect is detected from a luminance image obtained by photographing the retardation film through the second polarizing plate, normal first and second retardation regions are photographed through the second polarizing plate. The direction of the polarization transmission axis of one of the first and second polarizing plates is approximately the same as the optical axis of one of the first and second retardation regions so that each luminance is at the same level in the vicinity of the extinction state. Since the adjustment is performed within the parallel range, the defective portion should be imaged with high brightness with respect to the normal portions of the first retardation region and the second retardation region of the patterned retardation film that are photographed at a low luminance level. And the brightness of the normal part is the first phase difference. Since the same in frequency and a second phase difference territory, it is possible to accurately detect the defect.

It is a perspective view which shows the structure of the defect detection apparatus which implemented this invention. It is sectional drawing which shows the layer structure of a patterned retardation film. It is explanatory drawing which shows the inclination of the slow axis of each phase difference area | region. It is a block diagram which shows the structure of a defect detection unit. It is a flowchart which shows the process sequence for a test | inspection. It is explanatory drawing which shows the change of the luminance image by the process of an image process part. It is a block diagram which shows the defect detection unit in the example which erase | eliminates a boundary line by producing | generating a difference image. It is explanatory drawing which shows the distance of the other party pixel at the time of producing | generating a difference image. It is a block diagram which shows the defect detection unit in the example which determines whether it is a defect according to the pixel count to an adjacent defect candidate. It is explanatory drawing which shows the example which determines with a defect based on the pixel count to the defect candidate adjacent to a Y direction. It is explanatory drawing which shows the example which provided the phase difference compensation board.

[First Embodiment]
In FIG. 1, the defect detection apparatus 10 is provided on the production line of a patterned retardation film. The defect detection apparatus 10 is continuously supplied with a long patterned retardation film (hereinafter simply referred to as a retardation film) 12 from the retardation film forming step, and detects defects on the retardation film 12. Do. In the following description, the longitudinal direction of the retardation film 12 is the X direction, and the width direction of the retardation film 12 perpendicular to the X direction is the Y direction.

  In the retardation film 12, first retardation regions 14 and second retardation regions 15 having different retardation characteristics of the retardation layer 12 b are alternately arranged in the Y direction. As shown in FIG. 2, the retardation film 12 is formed on the surface of a support 12 a made of a transparent film such as TAC (triacetyl cellulose) in the retardation film forming step. The phase difference layer 12b that expresses the phase difference is formed.

  The first retardation region 14 and the second retardation region 15 are stripes extending along the longitudinal direction of the retardation film 12 and are alternately arranged at a constant pitch in the Y direction. The width (length in the Y direction) of each of the phase difference regions 14 and 15 is, for example, 270 μm, and the phase difference regions 14 and 15 are arranged in the Y direction at the same pitch as this width. In FIG. 1, the width of each of the phase difference regions 14 and 15 is exaggerated.

  As shown in FIG. 3, with the X direction as a reference (0 °), the first phase difference region 14 has an optical axis (slow axis or fast axis), for example, a slow axis As1 of “−45 ° (= Is formed so as to function as a λ / 4 wavelength plate tilted by −θ1) ”, and the second phase difference region 15 has a slow axis As2 tilted by“ + 45 ° (= + θ1). Therefore, the retardation film 12 is formed so that the slow axes As1 and As2 of the first retardation region 14 and the second retardation region 15 are orthogonal to each other. Such a retardation film 12 is attached to, for example, a liquid crystal screen and used for observing a stereoscopic image.

  As will be described in detail later, the slow axes As1, As2 of the retardation film 12 do not have to be strictly tilted ± 45 ° with respect to the X direction, and the reference direction is also arbitrary, If it is a patterned phase difference film in which the slow axes of the phase difference regions 14 and 15 are substantially orthogonal to each other, it becomes an inspection object. In addition, a patterned retardation film in which each retardation region is formed so that the fast axes are orthogonal to each other instead of the slow axis is also an inspection target.

  As shown in FIG. 1, the defect detection device 10 includes a light source unit 16, a photographing device 17, first and second polarizing plates 18 and 19, and a defect detection unit 20. Assuming that a 3D television is assumed as a final product in which the retardation film 12 is used and the viewer performs normal observation, for example, it is only necessary to detect a defect having a diameter of about 100 μm. The specifications of each part are so decided. This is because a TV viewer having a size of 100 μm or less is not identified as a defect and thus has no practical problem. In addition, examples of the defect include foreign matter, dirt, scratches, uneven light distribution (local disturbance in phase difference characteristics), and the like.

  In the defect detection apparatus 10, the retardation film 12 supplied from the retardation film forming step is continuously conveyed in the X direction by a conveyance mechanism (not shown). Two guide rollers (not shown) are arranged at a predetermined interval in the conveyance path of the retardation film 12, and the retardation film 12 is hung on these guide rollers, so that the space between the guide rollers is reduced. The retardation film 12 is held flat on the inspection stage.

  On the inspection stage, a light source unit 16, a photographing device 17, and first and second polarizing plates 18 and 19 are arranged. The light source unit 16 and the photographing device 17 are arranged on opposite sides of the conveyance path. The light source unit 16 includes a light source such as a halogen lamp. In this example, the inspection light is uniformly in the width direction from the lower side of the conveyance path to the lower surface of the retardation film 12 via the first polarizing plate 18. Irradiate. In addition, as a light source of the light source part 16, other lamps, such as a metal halide lamp, may be sufficient as LED (light emitting diode).

  The photographing device 17 photographs a luminance image of the retardation film 12 through the second polarizing plate 19 from the normal direction of the retardation film 12 on the upper side of the conveyance path. This photographing device 17 is a linear array camera composed of a photographing lens 17a and a CCD line sensor 17b composed of a large number of light receiving elements arranged in a line, and the width of the phase difference film 12 in one photographing. Take one line parallel to the direction. The value of each pixel of the luminance image represents the amount of light emitted from the corresponding retardation film 12 and transmitted through the second polarizing plate, that is, the luminance observed through the second polarizing plate. Each time the retardation film 12 is conveyed for one line, the photographing device 17 performs one photographing, thereby sequentially photographing luminance images of the retardation film 12 passing through the inspection stage. The pixel resolution of the imaging device 17, that is, the length in the Y direction on the retardation film 12 corresponding to one pixel of the luminance image is, for example, 10 μm / pixel. Note that an area camera that captures a plurality of lines simultaneously may be used as the imaging device 17 instead of the linear array camera.

  The first polarizing plate 18 is between the light source unit 16 and the retardation film 12, and the second polarizing plate 19 is between the retardation film 12 and the photographing device 17, and the posture parallel to the retardation film 12, respectively. It is arranged with. Each of the polarizing plates 18 and 19 is of a linear polarization type, and has a so-called crossed Nicol arrangement in which the polarization transmission axis P1 of the first polarizing plate 18 and the polarization transmission axis P2 of the second polarizing plate 19 are orthogonal to each other. ing. As will be described later, the directions of the polarization transmission axis P1 and the polarization transmission axis P2 are adjusted with respect to the retardation film 12, but the polarization transmission axes P1, P2 are adjusted by interlocking the direction of the polarization transmission axis P2 with the polarization transmission axis P1. Are kept orthogonal to each other.

  When the first polarizing plate 18 and the second polarizing plate 19 photograph the normal first retardation region 14 and the second retardation region 15 having no defect with the photographing device 17, the respective retardation regions 14, 15 The direction of the polarization transmission axis P1 with respect to the retardation film 12 is adjusted so that the same luminance level is obtained. This adjustment is performed in a range in which the polarization transmission axis P1 is substantially parallel to the slow axis As1 and the polarization transmission axis P2 is substantially parallel to the slow axis As2. Accordingly, the first retardation region is in a state close to the extinction state (near the extinction state) in which the inspection light transmitted through the normal first retardation region 14 and the second retardation region 15 does not pass through the second polarizing plate 19. 14 and the second phase difference region 15 have the same luminance.

  The retardation film 12 is formed with the first retardation region 14 and the second retardation region 15 so that the slow axes As1 and As2 are orthogonal to each other. Actually, the slow axes As1 and As2 are formed for various reasons. May deviate from being orthogonal to each other. For this reason, for example, when the polarization transmission axis P1 is adjusted to be parallel to the slow axis As1, the first phase difference region 14 is in an extinction state that does not pass inspection light, but the second phase difference region 15 is in an extinction state. There may be differences in the brightness levels between them. As described above, when a luminance difference is generated in each of the phase difference regions 14 and 15, a defect cannot be accurately detected based on a luminance image obtained by the imaging device 17. Therefore, although not in the extinction state, the brightness of the first phase difference region 14 and the second phase difference region 15 is adjusted to be the same level in the state close to the extinction state as described above, and defect detection is performed with high accuracy. To be able to.

  In this example, the normal phase difference region 14 and the second phase difference region 15 are in a state where the slow axis As1 of the first phase difference region 14 and the polarization transmission axis P1 of the first polarizing plate 18 are substantially parallel. The brightness is adjusted so as to be the same level. The reason why the brightness is almost parallel is to make the state close to the extinction state as can be seen from the above. Here, when the normal brightness of the first phase difference region 14 and the second phase difference region 15 are set to the same level in a state close to the extinction state, the deviation from the angle of 90 ° between the slow axes As1 and As2 is caused. Assuming Δθ, the polarization transmission axis P1 is set to the slow axis As1 and the polarization transmission axis P2 is set to the slow axis As2 while the angle formed by the polarization transmission axes P1 and P2 is maintained at 90 ° (crossed Nicols). On the other hand, it means to adjust to a state shifted by Δθ / 2.

  Further, the polarization transmission axis of either the first polarizing plate 18 or the second polarizing plate 19 is set to be the optical axis (slow axis or fast phase) of either the first retardation region 14 or the second retardation region 15. In this state, the brightness of the normal first phase difference region 14 and the second phase difference region 15 may be adjusted to the same level. Therefore, in the state where the polarization transmission axis P1 of the first polarizing plate 18 is substantially orthogonal to the slow axis As1 of the first phase difference region 14, each luminance of the normal first phase difference region 14 and the second phase difference region 15 is You may adjust so that it may become the same level. In addition, in a state where the polarization transmission axis P1 is substantially parallel to or orthogonal to the fast axis of the first phase difference region 14, the luminances of the normal first phase difference region 14 and the second phase difference region 15 are the same level. You may adjust so that it may become.

  The first polarizing plate 18 and the second polarizing plate 19 are crossed Nicols, and the optical axes of the first retardation region 14 and the second retardation region 15 (for example, the slow axes As1, As2) and Are almost orthogonal. For this reason, making the polarization transmission axis of one polarizing plate and the optical axis of one retardation region substantially parallel each other means that the polarization transmission axis of one polarizing plate and the optical axis of the other retardation region are Make the state almost orthogonal, make the polarization transmission axis of the other polarizing plate and the optical axis of the other retardation region substantially parallel, and make the polarization transmission axis of the other polarizing plate and one retardation region This is the same as making the optical axis of the lens substantially orthogonal.

  The luminance image from the imaging device 17 is sent to the defect detection unit 20. The defect detection unit 20 detects a defect in the retardation film 12 based on the luminance image obtained by the photographing device 17. As shown in FIG. 4, the defect detection unit 20 includes an image processing unit 21 and a defect detection unit 22, and the image processing unit 21 includes a D / A converter 23, a memory 24, a binarization circuit 25, A contraction processing circuit 26, an expansion processing circuit 27, and a noise removal circuit 28 are included.

  The D / A converter 23 converts the luminance level of each pixel of the luminance image into digital data. The memory 24 can store luminance images of a large number of lines, and generates a two-dimensional luminance image by sequentially accumulating the luminance images input line by line. The memory 24 is also used as a work memory for each part of the defect detection unit 20 to perform processing.

  A two-dimensional luminance image having a predetermined number of lines, for example, 400 lines, is generated in the memory 24, and subsequent image processing is performed using the two-dimensional luminance image as one processing unit. A two-dimensional luminance image composed of 400 lines is generated so as to overlap with the preceding and following luminance images, so that defects near the boundary between the two-dimensional luminance images can be detected without omission. Each unit of the image processing unit 21 reads out a luminance image of a predetermined number of lines in the memory 24, reads the luminance image, performs a predetermined process on the luminance image, and stores the luminance image in the memory 24 again.

  Identification data is given to the luminance image. This identification data indicates the history of processing performed by the image processing unit 21, and the content is updated each time processing is performed by each unit of the image processing unit 21. Based on this identification data, each unit of the image processing unit 21 identifies whether or not to perform the corresponding process on the luminance image stored in the memory 24, and if there is a luminance image to be subjected to the corresponding process, stores it from the memory 24. Read and execute the corresponding process. Accordingly, predetermined image processing by the image processing unit 21 is sequentially performed on the two-dimensional luminance image.

  The binarization circuit 25 binarizes each pixel of the luminance image stored in the memory 24 by comparing the luminance level with a predetermined threshold value. By this binarization, a pixel whose luminance level is lower than the threshold value is a dark pixel having a value “0”, and a pixel having a threshold value or more is a bright pixel having a value “1”. In this binarization, the pixels corresponding to the normal portions of the phase difference regions 14 and 15 are dark pixels, and the pixels that are higher in luminance than the normal portions and that are the defect candidates are bright pixels. The threshold value is set based on the luminance obtained from the normal portion after adjusting the direction of the polarization transmission axis P1 of the one polarizing plate 18.

  In the state where the first and second polarizing plates 18 and 19 are adjusted, the brightness of the normal portions of the first and second phase difference regions 14 and 15 is close to the extinction state, and both have the same brightness level. The threshold value for binarization may be a luminance level higher than the luminance level of the normal part, but is determined according to which luminance level or higher the luminance part is a defect candidate. This brightness level is usually determined by setting a defect sample that has been generated in a simulated manner in advance or a defect sample that has actually occurred in the adjusted defect detection apparatus 10 or an experimental machine with an equivalent optical system to measure the brightness of the defect. You can go and decide.

  The contraction processing circuit 26 deletes the boundary line between the phase difference regions 14 and 15 included in the luminance image and noise. Although the boundary between the first phase difference region 14 and the second phase difference region 15 has a narrow width, the phase difference characteristic is lost, and the luminance is increased similarly to the defect. For this reason, on the two-dimensional luminance image, the boundary line has a predetermined number of pixels in the Y direction, for example, two to three pixels, and the luminance on the line along the X direction is high.

  The contraction processing circuit 26 performs contraction processing on the binarized luminance image in order to erase the boundary line. In addition, due to the contraction process, bright pixels that are small noise in the Y direction are simultaneously erased. Each pixel on the boundary line is converted into a bright pixel by binarization. The contraction process is performed so that the width of the bright pixel is reduced in the direction in which the phase difference regions 14 and 15 are arranged, that is, in the Y direction. This contraction process is performed as many times as the number of times that the boundary line can be erased, that is, the number of times corresponding to the width of the boundary line (number of pixels in the Y direction). By erasing the boundary line in this way, the defect can be detected with high accuracy without the influence of the boundary between the phase difference regions 14 and 15.

  In one contraction process, when a pixel on the target luminance image is a dark pixel (value is “0”) on one of the adjacent pixels in the same line, for example, the right side, the target pixel is darkened. Processing for setting a pixel (value is “0”) is performed for each pixel. That is, the pixel at the right end of the bright pixel area on the same line is removed as a dark pixel. For this reason, for example, when the boundary line width is 2 pixels, the contraction process is performed twice, and when the boundary line width is 3 pixels, the boundary line is erased by performing the contraction process three times. The number of times the shrinking process is performed can be determined by the resolution of the photographing device 17 and the width of the boundary line on the retardation film 12, but it is experimentally considered in consideration of variations in the normal range of the boundary line width. It is desirable to determine from the width (number of pixels) of the boundary line on the luminance image.

  In the above shrinking process, the rightmost pixel in the bright pixel area is removed as a dark pixel for each process, but the leftmost pixel in the bright pixel area may be removed as a dark pixel.

  The expansion processing circuit 27 performs expansion processing on the luminance image after the contraction processing. By this expansion processing, the length in the Y direction of the bright pixel area remaining in the luminance image is restored. As a result, the defect size can be accurately determined at the time of defect determination. The expansion process is performed in the Y direction with the inverse logic of the contraction process, and for a pixel of interest on the luminance image, a pixel adjacent to the left side in the same line is a bright pixel (value is “1”). The process of setting the pixel of interest as a bright pixel (value “1”) is performed for each pixel. This expansion process is performed the same number of times as the contraction process.

  In this example, when a defect is detected, expansion processing is performed to facilitate defect extraction and defect determination based on the number of pixels, and the size of the remaining bright pixel region is restored. Therefore, if the defect is determined regardless of the size of the bright pixel region, the expansion process can be omitted. In this example, since the contraction process and the expansion process are performed only in the Y direction, the contraction process and the expansion process may be performed line by line without making the luminance image two-dimensional. Furthermore, in this example, the contraction and expansion processing is performed only in the Y direction, but by performing the processing in the X direction as well, it is possible to simultaneously erase bright pixels that cause noise with a small length in the X direction. .

  The noise removing circuit 28 removes a bright pixel area having a predetermined number of pixels or less from the luminance image as noise. In this example, one isolated bright pixel (isolated point) whose surroundings are all dark pixels is erased as noise after the contraction process and before the expansion process. This erasure is performed by converting a bright pixel into a dark pixel. Note that the noise removal by the noise removal circuit 28 has an area (number of pixels) that is not a problem even if it is not detected as a bright pixel as a noise that appears due to various factors such as electrical noise. A small bright pixel region is erased, and not only a single bright pixel region, but also a bright pixel region in which a plurality of bright pixels are connected may be removed as noise. Further, noise removal may be performed immediately before the defect detection after the brightness image before the shrinkage process or after the expansion process.

  The defect detection unit 22 detects a defect from the luminance image after the expansion process. In the defect detection, the defect detection unit 22 extracts a bright pixel region on the luminance image, and specifies the extracted region as a defect portion. It is also possible to determine the length, size, and shape of the defect in the X and Y directions by examining the number of pixels in the bright pixel region. Of course, it is also possible to determine whether or not a defective portion is to be determined based on the length and size of the bright pixel region. When the defect detection unit 22 detects a defect, the defect detection unit 22 generates and outputs defect information including the position of the defect on the retardation film 12 and the size of the detected defect. This defect information is used as control information for not using the defective part as a product.

  Next, the operation of the above configuration will be described. As shown in FIG. 5, the direction of the polarization transmission axis is adjusted prior to the defect inspection by the defect detection apparatus 10. First, between the first polarizing plate 18 and the second polarizing plate 19 on the inspection stage, a normal portion specified in advance of the retardation film 12 to be inspected is placed. Next, the polarization transmission axes P1 and P2 of the first polarizing plate 18 and the second polarizing plate 19 are orthogonal to each other so as to be in a crossed Nicols state.

  The coarse adjustment is performed while maintaining the crossed Nicol state and the parallel state with the retardation film 12, and the polarization transmission axis P <b> 1 of the first polarizing plate 18 is parallel to the slow axis As <b> 1 of the first retardation region 14 by this rough adjustment. To be. At this time, the polarization transmission axis P1 and the slow axis As1 do not have to be strictly parallel, but may be approximately parallel. In brief, when the first phase difference region 14 or the second phase difference region 15 is observed through the second polarizing plate 19 with the light source unit 16 turned on, the light extinction that minimizes the amount of light transmitted therethrough. What is necessary is just to adjust so that it may be in a state.

  After the coarse adjustment, the light source unit 16 and the imaging device 17 are operated so that the luminance levels of the first phase difference region 14 and the second phase difference region 15 can be referred to based on the luminance image captured by the imaging device 17. . Thereafter, fine adjustment is performed so that the luminance levels of the phase difference regions 14 and 15 are the same.

  In the fine adjustment, the brightness level is set to be the same while maintaining the crossed Nicols state and the parallel state with the retardation film 12. In this fine adjustment, the direction of the polarization transmission axis P1 is not changed greatly, and the direction of the polarization transmission axis P1 is maintained while maintaining the state where the polarization transmission axis P1 is substantially parallel to the slow axis As1 and close to the extinction state. adjust.

  The inspection light from the light source unit 16 enters the retardation film 12 through the first polarizing plate 18 as linearly polarized light. Here, when the polarization transmission axis P1 is exactly parallel to the slow axis As1 of the first phase difference region 14, the linearly polarized inspection light emitted from the first polarizing plate 18 has the first phase difference. A phase difference does not occur even if it passes through the region 14. Therefore, the inspection light is incident on the second polarizing plate 9 as linearly polarized light. The inspection light incident on the second polarizing plate 19 is not emitted from the second polarizing plate 19 because the polarization direction thereof is orthogonal to the polarization transmission axis P <b> 2 of the second polarizing plate 19.

  Further, when the polarization transmission axis P1 is substantially parallel to the slow axis As1 of the first phase difference region 14 but not exactly parallel, the inspection light passes through the first phase difference region 14. A phase difference is generated and becomes elliptically polarized light and enters the second polarizing plate 19. Accordingly, in this case, the polarized light component orthogonal to the polarization transmission axis P2 of the inspection light that has become elliptically polarized light is not emitted from the second polarizing plate 19, but the polarized light component parallel to the polarization transmission axis P2 is not emitted from the second polarizing plate. 19 is ejected. Then, as the deviation of the directional relationship between the polarization transmission axis P1 and the slow axis As1 from parallel increases, the polarization component parallel to the polarization transmission axis P2 increases, so the amount of light emitted from the second polarizing plate 19 increases. As a result, the luminance of the first phase difference region is increased.

  On the other hand, the same applies to the second phase difference region 15, but the relationship between the polarization transmission axes P1 and P2 with respect to the slow axis As2 is opposite to that of the first phase difference region 14. That is, when the polarization transmission axis P1 is exactly perpendicular to the slow axis As2 of the second retardation region 15, the polarization direction of the linearly polarized inspection light from the first polarizing plate 18 is the slow axis As2. Therefore, even if the light passes through the second phase difference region 15, it enters the second polarizing plate 19 without causing a phase difference. The inspection light is not emitted from the second polarizing plate 19 because the polarization direction is orthogonal to the polarization transmission axis P <b> 2 of the second polarizing plate 19.

  In addition, when the polarization transmission axis P1 is substantially orthogonal to the slow axis As1 of the second phase difference region 15, but not exactly orthogonal, the inspection light passes through the second phase difference region 15. A phase difference is produced to become elliptically polarized light. Of the inspection light that has become elliptically polarized light, only the polarization component parallel to the polarization transmission axis P <b> 2 is emitted from the second polarizing plate 19. The greater the deviation from the orthogonality of the direction relationship between the polarization transmission axis P1 and the slow axis As2, the greater the polarization component parallel to the polarization transmission axis P2, the greater the amount of light emitted from the second polarizing plate 19, and the brightness. Get higher.

  When the retardation film 12 in which the first retardation region 14 and the second retardation region 15 are formed so that the slow axes As1 and As2 are accurately orthogonal to each other is an inspection target, the slow axis As1 and The increase / decrease in the shift relative to the polarization transmission axis P1 is the same as the increase / decrease in the shift between the slow axis As2 and the polarization transmission axis P1, and the magnitude of the shift is the same. Accordingly, since the polarization transmission axis P1 is in a state parallel to the slow axis As1, and at the same time, the polarization transmission axis P1 is in a state orthogonal to the slow axis As2, the inspection from the first and second phase difference regions 14 and 15 is performed. Light is not emitted from the second polarizing plate 19, and the luminance of any region can be adjusted to “0” (extinction state).

  Although it is ideal as the retardation film 12 that the slow axes As1 and As2 are accurately orthogonal to each other as described above, formation of a retardation region in which the support 12a exhibits a slight birefringence even though it is slight. In many cases, the slow axis As1 and the slow axis As2 are not exactly orthogonal but slightly deviated from orthogonal due to factors such as an error in the direction of the slow axis. In this case, the polarization transmission axis P1 cannot be brought into a state orthogonal to the slow axis As2 at the same time as the polarization transmission axis P1 is made parallel to the slow axis As1. That is, the first phase difference area 14 and the second phase difference area 15 cannot be finely adjusted so as to be photographed at the luminance level “0” at the same time.

  However, the first and second phase difference regions 14 and 15 have the same displacement between the slow axis As1 and the polarization transmission axis P1 and the orthogonal displacement between the slow axis As2 and the polarization transmission axis P1. The polarization component of the inspection light that is transmitted through the second polarizing plate 19 and emitted from the second polarizing plate 19 can be made the same. Therefore, the luminance of the first and second phase difference regions 14 and 15 can be finely adjusted to the same luminance level at the same time in a state close to the extinction state where the luminance is not “0” but is a low luminance level.

  After performing the fine adjustment of the polarization transmission axis P1 as described above, the luminance level obtained by photographing the normal part of the first phase difference region 14 or the second phase difference region 15 with the photographing device 17 is obtained, and the luminance level is obtained. A slightly higher threshold value is set in the binarization circuit 25. After this, the defect inspection of the retardation film 12 is started.

  When the defect inspection is started, the retardation film 12 is continuously supplied to the defect detection apparatus 10 from the retardation film forming process, and further conveyed downstream through the inspection stage. During the conveyance, the inspection light from the light source unit 16 is irradiated onto the retardation film 12 via the first polarizing plate 18, and for each line of the retardation film 12 sent by one line, the photographing device 17 makes one line. A luminance image is taken.

  Every time photographing is performed, a luminance image for one line is sent from the photographing device 17 to the defect detection unit 20, and the luminance level of each pixel is converted into digital data by the D / A converter 23 and then written to the memory 24. As a result, a two-dimensional luminance image is generated. When a predetermined number of lines of luminance image are written in the memory 24, the binarization circuit 25 reads out the luminance image from the memory 24, and each pixel is compared with the previously set threshold value and binarized. .

  As shown in an example in FIG. 6A, the binarized luminance image I includes a dark pixel region D1 and a boundary line D2 corresponding to normal portions of the first phase difference region 14 and the second phase difference region 15. In addition, bright pixel regions D3 to D7 are included. The normal portions of the first phase difference region 14 and the second phase difference region 15 have a luminance lower than the threshold value, so that the dark pixel region D1 is a dark pixel stripe. Assuming that the width of each of the phase difference regions 14 and 15 is about 270 μm and the pixel resolution of the imaging device 17 is 10 μm / pixel, each dark pixel region D1 has about 27 dark pixels arranged in the Y direction and is long in the X direction. Striped. The boundary line D2 appears as a bright pixel due to the disturbance of the phase difference characteristic of each adjacent dark pixel region D1, as in the defective portion. Each boundary line D2 is, for example, a straight line extending in the X direction with about 2 to 3 bright pixels arranged in the Y direction.

  In the first and second phase difference regions 14 and 15, a defect portion such as a portion in which foreign matter is mixed, a scratched portion, or a portion where light distribution unevenness occurs is disturbed in the phase difference characteristics. When the linearly polarized inspection light is transmitted, the polarization direction is disturbed so that the component emitted from the second polarizing plate 19 is larger than the normal portion. For this reason, the defective portion has high brightness and becomes a bright pixel, and the area size of the bright pixel is in accordance with the size of the defective portion. Even when electrical noise is superimposed on a pixel of a luminance image, the pixel becomes a bright pixel. The bright pixel regions D3 to D7 correspond to such defective portions and noise.

  The binarized luminance image is read from the memory 24 by the contraction processing circuit 26, and contraction processing in the Y direction is performed a predetermined number of times, for example, three times. Thereby, as shown in FIG. 6B, the boundary line D2 is erased as a dark pixel. In addition, the bright pixel areas D4 to D7 corresponding to noise having a small length in the Y direction and minute defect portions are also dark pixels and simultaneously erased. The bright pixel area to be erased by the contraction process is very small and does not cause a problem in practical use. When the boundary line is thicker than the expected width, even if the contraction process is performed a predetermined number of times, the bright pixel of that portion remains, so that it can be detected as a defect.

  Noise removal by the noise removal circuit 28 is performed on the luminance image after the predetermined number of contraction processes. As a result, a small bright pixel region that becomes noise that has not been erased by the shrinking process is erased from the luminance image.

  Thereafter, the expansion processing circuit 27 performs expansion processing a predetermined number of times, for example, three times, and returns the width of the bright pixel area of the luminance image subjected to the contraction processing. As a result, as shown in FIG. 6C, the width of the bright pixel region D3 remaining in the contraction process returns to that before the contraction process.

  After the expansion processing, the defect detection unit 22 extracts a bright pixel area remaining in the luminance image, and specifies the bright pixel area as a defective portion. Accordingly, in FIG. 6C, the bright pixel region D3 is detected as a defective portion. Then, defect information including position information indicating the position of the detected defect portion and defect size is generated and output.

[Second Embodiment]
In the second embodiment, a difference is obtained by comparing the luminance of a pixel of interest with a pixel separated by a predetermined number of images on the luminance image, and the difference is set as a new value of the pixel of interest. The line is erased. In addition, except being demonstrated below, it is the same as that of 1st Embodiment, The same code | symbol is attached | subjected to the substantially same structural member, and the detailed description is abbreviate | omitted.

  As shown in FIG. 7, the image processing unit 21 in this example includes a D / A converter 23, a memory 24, a difference image generation circuit 31, a binarization circuit 25, and a noise removal circuit 28. The difference image generation circuit 31 uses a pixel separated by a predetermined number of images as a counterpart pixel for the pixel of interest on the luminance image of one line digitized by the D / A converter 23, and the difference pixel generation circuit 31 A difference process is performed in which a difference in luminance is taken and the difference is used as a new value of the pixel of interest. A difference image is generated by performing this difference process for each pixel of the luminance image. When obtaining the difference, a value obtained by subtracting the value of the counterpart pixel from the value of the pixel of interest is set as a new value of the pixel of interest.

  The generation of the difference image is for erasing the boundary line between the first and second phase difference regions 14 and 15 in the luminance image, and when the pixel of interest is a pixel on the boundary line, The distance (number of pixels) of the pixels at the time of difference is determined so that becomes a pixel on the boundary line. Since each pixel on the boundary line has substantially the same luminance level, the new value of the pixel on the boundary line is set to substantially “0”, and the boundary line is substantially erased.

  The direction from the pixel of interest to the partner pixel need only be at least separated in the Y direction, and may be both the X direction and the Y direction. As shown in FIG. 8, when the number of pixels in the X direction up to the partner pixel Px2 is “x1” pixels and the number of pixels in the Y direction is “y1” pixels with respect to the target pixel Px1, “y1” The line spacing, that is, n times the number of pixels corresponding to the width of the phase difference region (length in the Y direction) (n = 1, 2, 3,...). “X1” can be an arbitrary value. For example, when “x1” is set to “0”, when the target pixel Px1 is a pixel on the boundary line, the counterpart pixel Px2 is a pixel on another boundary line on the same line.

  Further, “y1” may be set to “n” times the number of pixels corresponding to the width of the phase difference region (between boundary lines), and x1 may be set to “1” or more. In this case, since the difference from the pixels on the line separated from the “x1” line is obtained, when the pixel of interest Px1 is a pixel on the boundary line, the counterpart pixel Px2 is on another boundary line on a different line. It becomes a pixel.

  In addition, when there is a concern about the occurrence of bright spots (hereinafter referred to as streak defects) generated in a streak shape in a certain angle direction with respect to the X direction, the streak defect is not to be erased. The values of x1 and y1 should be set so that the pixels are not compared with each other.

  For example, when a streak defect is likely to appear in a 45 ° direction with respect to the X direction, the counterpart pixel Px2 is not in a 45 ° direction with respect to the pixel of interest Px1 in order to prevent the streak defect from being erased. The combination of “x1” and “y1” is determined. At this time, it is necessary to determine the combination of “x1” and “y1” in consideration of the pixel resolution in the X direction and the Y direction. As in this example, when the imaging device 17 is a linear array camera, the pixel resolution in the Y direction is determined according to the imaging resolution of the imaging device 17, but the pixel resolution in the X direction is 1 of the imaging device 17. It is determined by the line photographing period and the conveyance speed of the retardation film 12. For example, when the pixel resolution in the X direction is 6 μm and the pixel resolution in the Y direction is 3 μm, “x1: y1 = 1: 2” is not satisfied in order not to erase the stripe defect in the 45 ° direction.

  The binarization circuit 25 binarizes the difference image with a predetermined threshold, and makes each pixel a bright pixel and a dark pixel. The noise removing circuit 28 erases a minute bright pixel region as noise from the binarized difference image. The defect detection unit 22 identifies a bright pixel region remaining after the processing of the noise removal circuit 28 as a defective portion.

  When the difference image is generated as described above, unlike the contraction processing of the first embodiment, a small bright pixel region is not erased at the stage of generation of the difference image, and an error or boundary caused by the resolution of the imaging device 17 Many small bright pixel areas remain due to variations in the edge portion of the line itself. For this reason, when generating a difference image and detecting a defect, it is preferable to erase noise.

  According to this example, for each pixel of the luminance image, a difference image is generated by obtaining a difference between the pixel of interest and the partner pixel, but each pixel on the boundary line of the luminance image is on the same or different boundary line. The difference from the partner pixel is obtained. Then, since the luminance of the pixels on each boundary line becomes substantially the same level, the new value of the pixel of interest is almost “0”. Therefore, if this pixel is binarized, it becomes a dark pixel, and as a result, the boundary line is erased.

  On the other hand, when the pixel of interest is a defective pixel, the normal pixel having a considerably low luminance level is the counterpart pixel in most cases, so the new value of the defective pixel does not change much. Bright pixels are obtained by the digitization. Therefore, the defective portion can be detected as a defect without being erased.

  In addition, in order to be able to detect streak-like defects in all directions, such as streak-like defects that occur in random directions, the difference from a plurality of counterpart pixels in different directions with the pixel of interest as a base point is calculated. Alternatively, a difference image in each direction may be generated from the same luminance image, and a defect may be detected from each difference image. For this purpose, a difference image can be generated by using pixels at different positions of x1 and y1 (x1 / y1) as the corresponding pixels. For example, x1 and y1 are set so that the direction of the counterpart pixel from the pixel of interest is 45 ° and a difference image in the 45 ° direction is created, and the direction of the counterpart pixel from the pixel of interest is 30 ° Thus, x1 and y1 are set to create a 30 ° difference image. Then, defect detection is performed on each difference image, and the defect portion specified by the 45 ° difference image and the defect portion specified by the 30 ° difference image are combined as a defect portion of the phase film 12. . In this example, there are two directions of 45 ° and 30 °, but the direction (angle) and the number of directions can be set as appropriate.

[Third Embodiment]
In the third embodiment, a pixel (bright pixel) region with high luminance is extracted as a defect candidate in a two-dimensional luminance image, and the number of pixels up to a defect candidate adjacent in the Y direction is adjacent to each defect candidate. The number of distance pixels is counted, and when the counted number of adjacent distance pixels is equal to or smaller than a predetermined value, the defect candidate is determined as a defect. In addition, except being demonstrated below, it is the same as that of 1st Embodiment, The same code | symbol is attached | subjected to the substantially same structural member, and the detailed description is abbreviate | omitted.

  In this example, as shown in FIG. 9, the image processing unit 21 includes a D / A converter 23, a memory 24, a binarization circuit 25, and a counting circuit 41. Each pixel of the luminance image of one line is digitized by the D / A converter 23 and written in the memory 24, and then converted into a bright pixel or a dark pixel by the binarization circuit 25.

  The counting circuit 41 sets each of the bright pixel areas where a plurality of bright pixels on the luminance image are connected as defect candidates. For each defect candidate, the counting circuit 41 counts the number of pixels up to the defect candidate in one line, that is, in the Y direction, as the adjacent distance pixel number Cp. When the adjacent distance pixel number Cp counted by the counting circuit 41 is equal to or less than the predetermined pixel number Cpth, the defect detection unit 22 identifies the defect candidate as a defect portion. The predetermined pixel number Cpth is a value smaller than the interval Cy between the boundary lines D2 on the luminance image.

  Note that a bright pixel region from one bright pixel can also be used as a defect candidate. However, in this example, such a bright pixel region does not have a problem even if it is not detected as noise or a practical defect, so there is no problem. Not.

  As shown in FIG. 10, the bright pixel region that is a defect candidate on the luminance image I includes each boundary line D2 in addition to the bright pixel region D8 that actually becomes a defect. As described above, the boundary line D2 is a defect candidate. However, in the normal part, the boundary lines D2 are separated from each other by a predetermined number of pixels Cpth in the Y direction (Cp> Cpth). It is never specified. On the other hand, when there is a defect in the first phase difference region 14 or the second phase difference region 15 or when the width of the boundary line is abnormally thick, the bright pixel region is different from the other bright pixel regions. It will be closer than the interval Cy between the boundary lines. Therefore, for such a bright pixel region, the adjacent distance pixel number Cp is equal to or less than the predetermined pixel number Cpth and is specified as a defective portion. Note that, when a defect candidate other than the boundary line D2 is in the vicinity of the boundary line D2, the boundary line D2 is also determined as a defect, but there is no actual damage in operation of the defect candidate inspection.

  This example uses the fact that each boundary line is separated at a constant interval (number of pixels), and defect extraction can be performed with extremely simple logic. The direction in which the number Cp of adjacent distance pixels is counted may be counted in one direction from the defect candidates as long as it is the Y direction, or may be counted in both directions. Further, similarly to the other embodiments, a removal circuit that removes, as noise, a bright pixel region in which the number of pixels has a predetermined value (for example, 2 pixels or less) may be provided.

[Fourth Embodiment]
In the fourth embodiment, a retardation compensation plate that cancels the retardation characteristics of the support of the retardation film is provided, and the retardation characteristics of the support are offset. In addition, it can be the same as that of 1st thru | or 3rd Embodiment except providing a phase difference compensating plate.

  In the fourth embodiment, as shown in FIG. 11, a retardation compensation plate 37 is disposed between the retardation film 12 and the second polarizing plate 19. The retardation compensation plate 37 cancels the retardation characteristics of the support 12 a of the retardation film 12 and is arranged in a posture parallel to the retardation film 12. The retardation compensator 37 is not limited as long as it cancels the retardation characteristics of the support 12a, but is simply the same as the support 12a of the retardation film 12 to be inspected. This is also the case in this example. In this case, the phase difference compensation plate 37 is arranged so that its optical axis is orthogonal to the optical axis (for example, the slow axis) of the support 12a.

  According to this example, it is possible to improve the contrast of the luminance image by bringing the luminance levels of the first and second phase difference regions 14 and 15 closer to the extinction state. Thereby, the defect of the retardation film 12 can be detected with higher accuracy. In this example, the retardation compensation plate 37 is disposed between the retardation film 12 and the second polarizing plate 19, but the retardation compensation plate 37 is disposed between the first polarizing plate 18 and the retardation film 12. May be arranged.

[Example 1]
In Example 1, the defect inspection of the retardation film 12 was actually performed using the defect detection apparatus 10 configured as in the first embodiment. In this defect inspection, the retardation film 12 using TAC as a support 12a was used as an inspection target.

  The width of the first and second retardation regions 14 and 15 formed on the retardation film 12 was 270 μm. Further, the pixel resolution in the Y direction in the luminance image photographed by the photographing device 17 was 10 μm / pixel. Since the boundary between the first and second phase difference regions 14 and 15 is a boundary line (bright pixel) having 2 to 3 pixels in the Y direction on the luminance image, it contracts by 3 pixels in the Y direction. After performing the contraction process once, further erasing the bright pixel area of one pixel as noise and performing the expansion process of expanding three pixels in the Y direction, the bright pixel area was specified as a defective portion. As a result, a defect having a diameter of about 100 μm could be detected.

[Example 2]
In Example 2, the retardation compensation plate 37 was provided as in the fourth embodiment to detect defects in the retardation film 12. The defect detection apparatus 10 used for this defect detection has the same configuration as that of the second embodiment except that the phase difference compensation plate 37 is provided, and detects a defect by creating a difference image. The retardation compensator 37 is a normal part of the same transparent film as the support 12a of the retardation film 12, and is arranged so that the optical axis of the retardation compensator 37 is orthogonal to the optical axis of the support 12a. In Example 2, the width of the first and second phase difference regions 14 and 15 is 270 μm, and the pixel resolution is 3 μm / pixel in order to detect the defect size with higher accuracy. Therefore, on the luminance image, the boundary line has a width of 5 to 7 pixels in the Y direction.

  By providing the phase difference compensation plate 37, as described above, the first and second phase difference regions 14 and 15 of the phase difference film 12 are closer to the extinction state, and the luminance level of the normal surface is lower than that of the first embodiment ( The image became dark and the contrast of the bright spot was improved. It is possible to set the threshold value for binarization as low as the contrast is improved, and convert a defective portion with a small amount of light transmitted through the second polarizing plate 19 into a bright pixel with high accuracy. Defects can now be detected.

  For each pixel of the luminance image, “y1 = 90” was set, and a difference image with the counterpart pixel separated by 90 pixels in the Y direction was generated. As a result, each pixel on the boundary line has a new value of almost “0” as a difference is obtained with a pixel on another boundary line adjacent to the boundary line as a counterpart pixel. As a result, the boundary line could be erased from the luminance image by binarization. After binarization, defects could be detected with sufficient accuracy from an image obtained by erasing a bright pixel area of 2 pixels or less as noise. It was possible to detect even a minute defect that was only as large as the width of the boundary line.

DESCRIPTION OF SYMBOLS 10 Defect detection apparatus 12 Patterned phase difference film 14, 15 Phase difference area 16 Light source part 17 Camera 18, 19 Polarizing plate 20 Detection unit 21 Image processing part 22 Defect detection part 25 Binarization circuit 26 Shrinkage processing circuit 27 Expansion circuit 31 Difference image generation circuit 41 Counter As1, As2 Slow axis P1, P2 Polarization transmission axis

Claims (19)

Patterning for detecting defects in a patterned retardation film in which a plurality of stripe-shaped first and second retardation regions, each of which functions as a λ / 4 wavelength plate and whose slow axes are substantially orthogonal to each other, are arranged alternately. In the phase difference film defect detection device,
First and second polarizing plates are crossed Nicols so as to sandwich the patterned retardation film,
A light source unit that irradiates the patterned retardation film with inspection light through the first polarizing plate; a photographing device that obtains a luminance image by photographing the patterned retardation film through the second polarizing plate;
A defect detection unit for detecting defects from the luminance image,
Said first and second polarizers, either in one polarization transmission axes condition to be substantially parallel to one of the optical axes of said first and second retardation region, a normal first and second Patterned retardation film defect characterized in that the direction of the one polarization transmission axis is adjusted so that each luminance when the phase difference region is photographed with a photographing device is the same level near the extinction state Detection device. In the patterned retardation film, the first and second retardation regions are formed by laminating a retardation layer on a support made of a transparent film,
A retardation compensator arranged between the first polarizing plate and the patterned retardation film or between the second polarizing plate and the patterned retardation film and cancels the retardation characteristic of the support. The defect detection apparatus of the patterned retardation film of Claim 1 characterized by the above-mentioned.   3. The defect detection apparatus for a patterned retardation film according to claim 2, wherein the retardation compensation plate is the same transparent film as the support. A binarizing circuit shadows have been the luminance image is binarized with a predetermined threshold value, for each pixel in the dark pixel is lower than or more bright pixels and the threshold threshold shooting, a plurality of light pixels an image processing unit which have a defect candidate extracting circuit for an area defect candidate regions linked is,
The defect detection unit includes a boundary line in the luminance image in which the number of pixels between the defect candidate area and another adjacent defect candidate area corresponds to a boundary between the first and second phase difference areas. 4. The patterned phase difference according to claim 1, wherein when the number of counted pixels is equal to or smaller than a predetermined value, the defect candidate area is defined as a defect. 5. Film defect detection device. An image processing unit that erases a boundary line in the luminance image corresponding to a boundary between the first and second phase difference regions;
4. The defect detection apparatus for a patterned retardation film according to claim 1, wherein the defect detection unit detects a defect based on a luminance image from which the boundary line is deleted. 5.   The image processing unit binarizes the captured luminance image with a predetermined threshold value, and makes each pixel a bright pixel that is equal to or higher than the threshold value and a dark pixel that is lower than the threshold value; A contraction processing circuit that performs contraction processing for contracting bright pixel regions in the direction in which the boundary lines are arranged on the binarized luminance image at a number of times corresponding to the width of the boundary lines, and for erasing the boundary lines; 6. The defect detection apparatus for a patterned retardation film according to claim 5, further comprising: The image processing unit performs an expansion process for expanding a bright pixel area in the same direction as the contraction process on the luminance image subjected to the contraction process, and sets the bright pixel area remaining in the contraction process to a size before the contraction process. An expansion processing circuit to return to
The defect detection device for a patterned retardation film according to claim 6, wherein the defect detection unit detects a defect based on the brightness image subjected to the expansion process. The image processing unit determines, based on the luminance of the pixel on the luminance image, a pixel separated from the pixel by at least the number of pixels corresponding to the boundary line interval in the direction in which the boundary line is arranged as a counterpart pixel. A difference image generation circuit that generates a difference image in which the boundary line is substantially eliminated by performing difference processing on each pixel on the luminance image by using a value obtained by subtracting the luminance of the partner pixel as a new value of the pixel; ,
6. The binarization circuit comprising: binarizing the difference image with a predetermined threshold value to make each pixel a bright pixel equal to or higher than the threshold value and a dark pixel lower than the threshold value. The defect detection apparatus of the patterned retardation film of description. The difference image generation circuit generates a difference image in each direction by performing a difference process between the other pixels in different directions on the luminance image,
The binarization circuit binarizes each of the plurality of difference images with a predetermined threshold value to generate a plurality of binarized images,
9. The defect detection apparatus for a patterned retardation film according to claim 8, wherein the defect detection unit performs defect detection on a plurality of the binarized images. Patterning for detecting defects in a patterned retardation film in which a plurality of stripe-shaped first and second retardation regions, each of which functions as a λ / 4 wavelength plate and whose slow axes are substantially orthogonal to each other, are arranged alternately. In the method of detecting a defect in the retardation film,
When the patterned retardation film is irradiated with inspection light through the first polarizing plate in a state where the patterned retardation film is arranged between the first and second polarizing plates arranged in a crossed Nicol arrangement. The polarized light of one of the first and second polarizing plates so that the luminances of the normal first and second phase difference regions photographed through the second polarizing plate are at the same level in the vicinity of the extinction state. An adjustment step of adjusting the direction of the transmission axis within a range substantially parallel to the optical axis of one of the first and second phase difference regions;
Disposing the patterned retardation film between the first and second polarizer direction is adjusted in polarization transmission axis, the inspection light on the patterned retardation film through said first polarizer An imaging step of irradiating and capturing a luminance image by imaging a retardation film through the second polarizing plate;
And a detection step of detecting a defect based on the acquired luminance image. A defect detection method for a patterned retardation film, comprising: In the patterned retardation film, the first and second retardation regions are formed by laminating a retardation layer on a support made of a transparent film. In the adjustment step and the photographing step, A retardation compensation plate that cancels the retardation characteristics of the support is disposed between the polarizing plate and the patterned retardation film, or between the second polarizing plate and the patterned retardation film. The defect detection method of the patterned retardation film of Claim 10 characterized by these.   The defect detection method for a patterned retardation film according to claim 11, wherein the retardation compensation plate is the same transparent film as the support. Binarizing the captured luminance image with a predetermined threshold value, and making each pixel a bright pixel equal to or higher than the threshold value and a dark pixel lower than the threshold value;
A defect candidate extraction step in which a region where a plurality of bright pixels are connected is used as a defect candidate region,
In the detection step, the number of pixels between the defect candidate area and another adjacent defect candidate area is determined based on the boundary line in the luminance image corresponding to the boundary between the first and second phase difference areas. 13. The patterned retardation film according to claim 10, wherein the defect candidate region is defined as a defect when the number of pixels counted in a line-up direction is equal to or less than a predetermined value. Defect detection method . An erasing step of erasing a boundary line in the luminance image corresponding to a boundary between the first and second phase difference regions;
The defect detection method for a patterned retardation film according to any one of claims 10 to 12, wherein the detection step detects a defect based on a luminance image from which the boundary line is erased.   The erasing step binarizes the captured luminance image with a predetermined threshold value, and makes each pixel a bright pixel equal to or higher than the threshold value and a dark pixel lower than the threshold value; and A shrinking step of shrinking a region of bright pixels in a direction in which the border lines are arranged, performing a shrinking process on the binarized luminance image at a number of times corresponding to the width of the border line, and erasing the border line. The defect detection method of the patterned retardation film of Claim 14 characterized by these. The erasing step performs an expansion process for expanding the bright pixel area in the same direction as the contraction process on the luminance image subjected to the contraction process, and returns the area of the bright pixel remaining in the contraction process to the size before the contraction process. Having an expansion process step;
The defect detection method for a patterned retardation film according to claim 15, wherein the detection step detects a defect based on the brightness image that has been subjected to the expansion process. In the erasing step, with respect to a pixel on the luminance image, a pixel separated from the pixel by at least the number of pixels corresponding to the boundary line interval in the direction in which the boundary line is arranged is used as a partner pixel from the luminance of the pixel. A difference image generation step for generating a difference image in which the boundary line is substantially eliminated by performing a difference process for each pixel on the luminance image by subtracting the luminance of the pixel as a new value of the pixel;
15. The binarizing step of binarizing the difference image with a predetermined threshold value to make each pixel a bright pixel that is equal to or higher than the threshold value and a dark pixel that is lower than the threshold value. The defect detection method of the patterned retardation film of description. The difference image generation step generates a difference image in each direction by performing a difference process between the other pixels in different directions on the luminance image,
The binarization step binarizes each of the plurality of difference images with a predetermined threshold value to generate a plurality of binarized images,
The defect detection method for a patterned retardation film according to claim 17, wherein the detection step performs defect detection on a plurality of the binarized images. A manufacturing step of manufacturing a patterned retardation film in which a plurality of stripe-shaped first and second retardation regions, each of which functions as a λ / 4 wavelength plate and whose slow axes are substantially orthogonal to each other, are arranged in an alternating manner;
Disposing the patterned retardation film between the first and second polarizer direction is adjusted in polarization transmission axis, irradiating an inspection light to the patterned retardation film through said first polarizer And acquiring the luminance image by photographing the retardation film through the second polarizing plate;
A detection step of detecting a defect based on the acquired luminance image;
Prior to the obtaining step, inspection light is applied to the patterned retardation film via the first polarizing plate in a state where the patterned retardation film is arranged between the first and second polarizing plates. Either of the first and second polarizing plates is such that when irradiated, the respective luminances of the normal first and second retardation regions photographed through the second polarizing plate are at the same level in the vicinity of the extinction state. A patterned retardation film comprising: an adjusting step for adjusting a direction of one polarization transmission axis within a range substantially parallel to one of the optical axes of the first and second retardation regions. Manufacturing method.

Images