JP5982245B2 - Image processing apparatus and image processing method - Google Patents

Description

  Embodiments relate to image processing.

  In general, luminance unevenness depending on lens characteristics, for example, occurs in an image taken by a camera. The luminance unevenness is desired to be suppressed because it impairs the visibility of the subject of interest in the captured image or degrades the inspection accuracy when performing the automatic inspection process on the subject of interest in the captured image. Shading correction is known as one effective method for suppressing luminance unevenness. The shading correction is realized by creating a reference image expressing luminance unevenness and performing an operation (for example, division or subtraction) using the reference image.

  The effect of suppressing luminance unevenness by shading correction depends on the quality of the reference image. Specifically, if an unnecessary component such as a noise component remains in the reference image, the unnecessary component is amplified through shading correction, or the original signal component of the subject of interest is impaired. Conventionally, a technique for creating a reference image by performing a smoothing process or an edge removal process on an image including a subject of interest is known. However, according to such a technique, the unnecessary component may be generated in the reference image through the smoothing process or the edge removal process.

JP 2007-13231 A

  An object of the embodiment is to create a high-quality reference image.

  According to the embodiment, the image processing apparatus includes a selection unit and a creation unit. The selection unit selects two or more images from a plurality of input images including one or more images including the target subject and a plurality of images not including the target subject, from a plurality of images not including the target subject. Thus, a plurality of images to be averaged are obtained. The creation unit creates a reference image for shading correction by averaging a plurality of images to be averaged.

1 is a block diagram illustrating an image processing apparatus according to a first embodiment. 2 is a flowchart illustrating the operation of the image processing apparatus in FIG. 1. The figure which illustrates GUI shown to a user by the user interface part of Drawing 1. 4 is a flowchart illustrating the behavior of the GUI in FIG. 3. Explanatory drawing of the specific example of several sheets of input images. Explanatory drawing of the specific example of a reference image. An explanatory view of a specific example of a plurality of output images. FIG. 6 is a graph illustrating density value distribution in an arbitrary VIII-VIII ′ cross section among a plurality of input images of FIG. 5. 7 is a graph illustrating a density value distribution in a IX-IX ′ cross section of the reference image in FIG. 6. 8 is a graph illustrating a density value distribution in an X-X ′ cross section of any one of the plurality of output images in FIG. 7. The figure which illustrates GUI shown to a user by the user interface part of Drawing 1. The figure which illustrates the external appearance inspection system which concerns on 2nd Embodiment. The figure which illustrates the X-ray inspection system which concerns on 3rd Embodiment.

  Hereinafter, embodiments will be described with reference to the drawings. Hereinafter, the same or similar elements as those already described are denoted by the same or similar reference numerals, and redundant description is basically omitted.

(First embodiment)
As illustrated in FIG. 1, the image processing apparatus 100 according to the first embodiment includes an image acquisition unit 101, an image selection unit 102, a reference image creation unit 103, an image correction unit 104, and a user interface unit. 105.

  The image acquisition unit 101 acquires a plurality of input images from the outside. The plurality of input images include one or more images including the target subject and a plurality of images not including the target subject. The image acquisition unit 101 outputs the input image to the image selection unit 102. Furthermore, the image acquisition unit 101 outputs part or all of the plurality of input images to the image correction unit 104 and the user interface unit 105 as necessary.

  The plurality of input images may be a series of frames in the moving image, or may be a plurality of still images obtained by continuous shooting. In the following description, for the sake of simplicity, it is assumed that a plurality of input images correspond to a series of frames in a moving image. The plurality of input images are preferably taken from a fixed viewpoint, but may be taken from a movable viewpoint such as a drive recorder image. The subject of interest may be a moving object or may be stationary.

  The image selection unit 102 inputs a plurality of input images from the image acquisition unit 101. The image selection unit 102 selects a plurality of images to be averaged by selecting two or more images from among a plurality of images that do not include the noted subject from the plurality of input images. As will be described later, in order to select two or more images from a plurality of input images that do not include the subject of interest described above, the image selection unit 102 performs user input via a GUI (Graphical User Interface). May be used, pattern matching may be used, or a predetermined rule may be used. The image selection unit 102 outputs the averaging target image to the reference image creation unit 103.

  In general, by averaging N images, the standard deviation of the amplitude of the random noise component included in each image becomes 1 / √N times. Therefore, the larger the number of images to be averaged, the more the random noise component is suppressed.

  The reference image creation unit 103 inputs a plurality of averaging target images from the image selection unit 102. The reference image creation unit 103 creates a reference image by averaging a plurality of images to be averaged in units of pixels. The reference image creation unit 103 outputs the reference image to the image correction unit 104. Note that the reference image creation unit 103 may present the reference image to the user through a GUI described later before outputting the reference image to the image correction unit 104, and allow the user to check the quality of the reference image. .

  The image correction unit 104 inputs at least a part (hereinafter referred to as a correction target image) of the plurality of input images from the image acquisition unit 101, and inputs a reference image from the reference image creation unit 103. The image correction unit 104 generates one or more output images by performing shading correction of one or more correction target images using the reference image. The image correction unit 104 may present the output image to the user through a GUI as necessary.

  The user interface unit 105 presents a GUI described later to the user. The user interface unit 105 notifies the user of information or accepts user input through the GUI. For example, the GUI may display an input image, an averaging target image, a reference image, a correction target image, an output image, and the like. Further, the GUI may accept, for example, user input for designating a plurality of images to be averaged, user input for starting creation of a reference image, user input for saving a reference image, and the like. .

  Hereinafter, image processing performed by the image processing apparatus of FIG. 1 will be described with reference to FIG. The image processing in FIG. 2 starts from step S201. In step S201, the image acquisition unit 101 acquires a plurality of input images.

  For example, as illustrated in FIG. 5, the image acquisition unit 101 acquires, as a plurality of input images, a moving image in which a car as a subject of interest crosses the screen from the left side to the right side. In FIG. 5, the movement of the automobile is expressed in order to make it easy to grasp the outline of the input image, but in reality, the position of the automobile may be different in each of the plurality of input images. In the example of FIG. 5, for example, a difference in brightness due to illumination conditions, lens characteristics, and the like is generated in the background of the input image.

  That is, the plurality of input images illustrated in FIG. 5 include, for example, an image (hereinafter also referred to as an A-type image) in which a car overlaps a bright background (for example, an area illuminated by a streetlight), An image (hereinafter also referred to as a B-type image) in which the automobile overlaps a boundary between a light background and a dark background (for example, a shadow area of a building), an image in which the automobile overlaps a dark background ( Hereinafter, it includes an image that does not include a car (that is, an image that does not include a subject of interest) (hereinafter also referred to as a D-type image). According to the example of FIG. 5, when the automobile starts to overlap a dark background, the visibility of the automobile starts to deteriorate.

  The image selection unit 102 obtains a plurality of images to be averaged by selecting two or more images not including the subject of interest from the plurality of input images acquired in step S201 (step S202). The reference image creation unit 103 creates a reference image by averaging the plurality of averaging target images obtained in step S202 (step S203).

  For example, as shown in FIG. 6, the reference image creation unit 103 averages a plurality of averaging target images (all of which are the above-described D type images) that do not include the car that is the subject of interest. Create a reference image. According to the example of FIG. 6, a plurality of images corresponding to the background image are averaged. Accordingly, the reference image in FIG. 6 has the same contrast as the background of the input image illustrated in FIG. In addition, when an input image including an automobile (that is, the above-described A type, B type, or C type image) is included as a part of a plurality of images to be averaged, the reference image includes the signal of the automobile. Ingredients will remain.

  As illustrated in FIGS. 8 and 9, (random) noise components included in the background of each of a plurality of input images are suppressed through averaging. Here, FIG. 8 illustrates the distribution of density values in the section VIII-VIII ′ of FIG. 5 (more precisely, any one of the plurality of input images illustrated in FIG. 5), and FIG. 6 illustrates a distribution of density values in a section IX-IX ′ of FIG. Therefore, according to the reference image created in step S203, high-quality shading correction can be performed.

The image correcting unit 104 generates one or more output images by performing shading correction on some or all of the plurality of input images using the reference image created in step S203 (step S204). Specifically, the image correction unit 104 may perform shading correction according to the following mathematical formula (1), for example.

  In Equation (1), I (i, j) represents the density value of the pixel (i, j) in any one of the plurality of input images, and O (i, j) represents the pixel in the corresponding output image. (I, j) represents the density value, R (i, j) represents the density value of the pixel (i, j) in the reference image, and N represents the total number of pixels in the reference image.

  FIG. 7 illustrates output images corresponding to FIGS. In FIG. 7, the movement of the car that is the subject of interest is expressed with the same purpose as in FIG. 5. In short, the output image illustrated in FIG. 7 is generated by performing shading correction on the input image illustrated in FIG. 5 using the reference image illustrated in FIG. Since the shading correction suppresses the light / dark difference in the input image, the visibility of the vehicle is greatly improved particularly in the output image corresponding to the input image where the vehicle is superimposed on a dark background.

  As illustrated in FIG. 10, the density value of the output image substantially matches the density value average of the reference image except for fluctuation due to noise components. Here, FIG. 10 illustrates the distribution of density values in the X-X ′ cross section of FIG. 7 (more precisely, any one of the plurality of output images illustrated in FIG. 7). However, when the noise component is larger than the signal component in the input image (for example, when the SNR (Signal-to-Noise Ratio) is bad), the noise component may be emphasized by the calculation of Expression (1). is there.

As described above, the image selection unit 102 can select two or more images not including the subject of interest from a plurality of input images using various techniques.
For example, the image selection unit 102 may use a predetermined rule regarding shooting (that is, generation) of a plurality of input images. For example, regarding the shooting of a plurality of input images, it is possible to define a rule that the subject of interest is not shot for a specified time (or a specified number of frames) from the beginning or the end. Based on such rules, the image selection unit 102 may automatically select images within a range that does not exceed the specified time from the beginning or end of a plurality of input images.

  The image selection unit 102 may use pattern matching. Specifically, the image selection unit 102 uses, as a reference image for pattern matching, one image that does not include the subject of interest among a plurality of input images. Note that the image selection unit 102 may select the reference image in accordance with a user input through the GUI, or may automatically select the reference image using the above-described predetermined rule.

  The image selection unit 102 calculates SAD (Sum of Absolute Differences) or SSD (Sum of Squared Differences) of density values in all pixels or a part of pixel regions between each of a plurality of input images and the reference image. To do. Furthermore, the image selection unit 102 compares the calculated SAD or SSD with a threshold value. An image for which the calculated SAD or SSD is less than the threshold value matches the reference image, so there is a high possibility that the image does not include the target subject. Therefore, the image selection unit 102 only needs to select two or more images in which the SAD or SSD of the pixel density value between the reference image and the reference image is less than the threshold value.

  The performance of image selection using pattern matching depends on the threshold value. That is, if the threshold is too small, the total number of images selected may be very small. On the other hand, if the threshold value is excessively large, an image including the subject of interest may be selected by mistake. The threshold value may be automatically set to an empirically derived value, or may be set according to user input through a GUI. When the user designates a threshold value through the GUI, the user interface unit 105 may display a reference image that is provisionally created under the designated threshold value on the GUI. If the temporarily created reference image is displayed, the user can check the quality of the reference image at any time, so that the threshold can be easily adjusted.

  Alternatively, the image selection unit 102 may use user input through a GUI. Usually, the user can easily determine visually an image that does not include the subject of interest, so that the image selection unit 102 can more reliably select two or more images that do not include the subject of interest by using the user input.

  Specifically, according to the GUI of FIG. 3, the user selects a start image and an end image, and the image selection unit 102 selects a plurality of continuous images positioned from the selected start image to the end image. . According to the GUI of FIG. 11, the user determines a selection state / non-selection state for each of a plurality of input images, and the image selection unit 102 selects a plurality of images in the selection state.

Hereinafter, a specific example of the GUI displayed by the user interface unit 105 will be described.
In FIG. 3, a reference image creation window 300 including various GUI parts is depicted. The reference image creation window 300 includes a start image display area 301, a slide bar 302, a slider 303, a box 304, an end image display area 305, a slide bar 306, a slider 307, a box 308, and a creation button 309. A reference image display area 310, a save button 311, and an end button 312.

  The GUI of FIG. 3 behaves as illustrated in FIG. First, for example, the GUI in FIG. 3 is activated based on a user input for activating a shading correction application (step S401). Specifically, the user interface unit 105 presents a reference image creation window 300 to the user. 3 is forcibly terminated when the end button 312 is pressed.

  The user interface unit 105 displays the first one (or other default image) among the plurality of input images in the start image display area 301 (step S402). Further, the user interface unit 105 displays the last one (or other default image) of the plurality of input images in the end image display area 305 (step S403). Steps S402 and S403 may be executed in reverse order or in parallel.

  The image displayed in the start image display area 301 is determined by the position of the slider 303 on the slide bar 302. In addition, information for allowing the user to recognize the image currently displayed in the start image display area 301 (for example, frame number, reproduction time, shooting time, etc.) is displayed in the box 304. Similarly, the image displayed in the end image display area 305 is also determined by the position of the slider 307 on the slide bar 306. In addition, information for allowing the user to recognize the image currently displayed in the end image display area 305 is displayed in the box 308.

  The user can change the image displayed in the start image display area 301 by moving the slider 303 on the slide bar 302. In other words, as the slider 303 moves, the information displayed in the start image display area 301 and the box 304 changes. Similarly, the user can change the image displayed in the end image display area 305 by moving the slider 307 on the slide bar 306. That is, as the slider 307 moves, information displayed in the end image display area 305 and the box 308 changes.

  The user selects the start image by moving the slider 303 to a desired position on the slide bar 302 (step S404). Similarly, the user selects an end image by moving the slider 307 to a desired position on the slide bar 306 (step S405). In steps S404 and S405, the user moves the slider 303 and the slider 307 while looking at the images displayed in the start image display area 301 and the end image display area 305, so that the start image and the end image not including the target subject can be easily displayed. Can be selected. Note that step S404 and step S405 may be executed in reverse order or in parallel. Furthermore, one or both of step S404 and step S405 may be omitted. When step S404 is omitted, the image displayed in the start image display area 301 in step S402 is treated as the start image. When step S405 is omitted, the image displayed in the end image display area 305 in step S403 is treated as the end image.

  After step S404 and step S405, the user presses the creation button 309 (step S406). When the creation button 309 is pressed, a reference image is created (step S407). Specifically, the image selection unit 102 selects a plurality of continuous images located from the start image to the end image, thereby obtaining a plurality of averaging target images. Further, the reference image creation unit 103 creates a reference image by averaging a plurality of images to be averaged.

  The user interface unit 105 displays the reference image created in step S407 in the reference image display area 310 (step S408). Here, according to the GUI of FIG. 3, the user only has to select the start image and the end image, so the burden of user input is small. On the other hand, since the user does not always look through all the images positioned between the start image and the end image, for example, a plurality of images including unnecessary components such as noise components and signal components of a moving subject of interest are present. There is a risk of unexpected mixing as part of the averaging target image. In such a case, since the unnecessary component remains in the reference image, the image quality of the output image may be deteriorated.

  Therefore, the GUI of FIG. 3 displays the reference image in the reference image display area 310 to allow the user to visually confirm the quality of the reference image. Here, the user can easily determine whether or not an unnecessary component remains in the reference image by comparing the reference image display area 310 with the start image display area 301 or the end image display area 305. When determining that an unnecessary component remains in the reference image, the user may re-select the start image or the end image and then press the creation button 309 again.

  Here, if the above-described pattern matching is used, whether or not unnecessary components remain in the reference image and one or more images to be averaged that generate unnecessary components in the reference image are automatically determined. Is also possible. The reference image for pattern matching may be a start image, an end image, another default image, or may be selected according to user input.

  If the user determines that no unnecessary component remains in the reference image, the user presses the save button 311 (step S409). When the save button 311 is pressed, the reference image created in step S407 is saved in a memory (not shown). The reference image stored in the memory is read as necessary by the image correction unit 104 and is used for shading correction on a part or all of the input image.

  As described above, according to the GUI of FIG. 3, two or more continuous images are selected from a plurality of input images. Therefore, according to this GUI, although the burden of user input can be reduced, an image including unnecessary components may be unexpectedly mixed as a part of a plurality of images to be averaged. Therefore, the user interface unit 105 may display a GUI exemplified in FIG. 11 so that the user can designate a selected state / non-selected state for each of a plurality of input images.

  FIG. 11 depicts a reference image creation window 500 that includes various GUI components. The reference image creation window 500 includes an input image display area 501, a slide bar 502, a slider 503, a box 504, a selection button 505, a non-selection button 506, a creation button 507, a reference image display area 508, And a save button 509.

  The GUI of FIG. 11 behaves as described below. First, the GUI in FIG. 11 is activated based on, for example, a user input for activating a shading correction application. Specifically, the user interface unit 105 presents a reference image creation window 500 to the user. Note that the GUI in FIG. 11 is forcibly terminated when the end button 510 is pressed.

  The user interface unit 105 displays the first one (or other default image) of the plurality of input images in the input image display area 501. Here, the image displayed in the input image display area 501 is determined by the position of the slider 503 on the slide bar 502. In addition, information for allowing the user to recognize the image currently displayed in the input image display area 501 is displayed in the box 504.

  The user can change the image displayed in the input image display area 501 by moving the slider 503 on the slide bar 502. That is, as the slider 503 moves, the information displayed in the input image display area 501 and the box 504 changes. The user can specify the selection state / non-selection state of the input image by pressing the selection button 505 or the non-selection button 506 with respect to the input image displayed in the input image display area 501.

  In order to reduce the burden of user input, a default state (selected state or non-selected state) may be set in advance for each of the plurality of input images. Here, the default state may be set based on a predetermined rule, may be set using pattern matching, or based on user input via a GUI similar to FIG. It may be set. Further, when the selection button 505 or the non-selection button 506 is pressed, the user interface unit 105 may automatically display the next input image in the input image display area 501.

  The user presses the creation button 507 after designating the selected / non-selected state for each of the plurality of input images. When the creation button 507 is pressed, a reference image is created. Specifically, the image selection unit 102 selects a plurality of images in a selected state, thereby obtaining a plurality of images to be averaged. Further, the reference image creation unit 103 creates a reference image by averaging a plurality of images to be averaged.

  The user interface unit 105 displays the created reference image in the reference image display area 508. Here, according to the GUI of FIG. 11, the user selects a selection state / non-selection state for each of a plurality of input images, and thus an image including unnecessary components is mixed in as a part of the plurality of averaging target images. The possibility is suppressed. However, since the user may miss an image containing unnecessary components, it is desirable to prepare a confirmation operation similar to the GUI of FIG.

  According to the GUI of FIG. 11, the reference image is displayed in the reference image display area 508, so that the user can visually confirm the quality of the reference image. Here, the user can easily determine whether or not unnecessary components remain in the reference image by comparing the reference image display area 508 and the input image display area 501. Note that the user interface unit 105 preferably automatically displays the selected image in the input image display area 501. When determining that an unnecessary component remains in the reference image, the user changes the selection state / non-selection state of each of the plurality of input images as necessary, and then presses the creation button 507 again. May be.

  Here, if the above-described pattern matching is used, whether or not unnecessary components remain in the reference image and one or more images to be averaged that generate unnecessary components in the reference image are automatically determined. Is also possible. The reference image for pattern matching may be any image in a selected state, another default image, or may be selected according to user input.

  When the user determines that no unnecessary component remains in the reference image, the user presses the save button 509. When the save button 509 is pressed, the created reference image is saved in a memory (not shown). The reference image stored in the memory is read as necessary by the image correction unit 104 and is used for shading correction on a part or all of the input image.

  As described above, the image processing apparatus according to the first embodiment obtains a plurality of images to be averaged by selecting two or more images not including the subject of interest from a plurality of input images, A reference image is generated by averaging the plurality of averaging target images. Therefore, this image processing apparatus averages noise components included in each of a plurality of averaging target images while preventing an unnecessary component such as a signal component of a moving subject of interest from being mixed into the reference image. Repress. Therefore, according to this image processing apparatus, since a high-quality reference image is created, the effect of shading correction can be improved.

(Second Embodiment)
The appearance inspection system according to the second embodiment includes an image processing apparatus 100, an image inspection apparatus 600, a camera 610, and a stage 620 as illustrated in FIG.
The camera 610 is a digital camera using a solid-state image sensor such as a charge-coupled device (CCD) or a complementary metal oxide semiconductor (CMOS). The camera 610 electronically photographs the upper surface of the stage 620 from above the stage 620 described later. The camera 610 generates image data formed by arranging the pixel information obtained by the image sensor in a matrix (lattice) and outputs the image data to the subsequent image processing apparatus 100.

  The stage 620 is a stage having a table-like plane on which an object to be measured OJ (for example, an electronic component such as a liquid crystal panel or an integrated circuit) to be subjected to appearance inspection is placed. The object to be measured OJ is placed in a range photographed by the camera 610 on the stage 620. Imaging with the camera 610 is performed in a state where the object to be measured OJ is stationary on the stage 620.

  The image processing apparatus 100 performs the same or similar image processing as that of the first embodiment on the image data obtained by the camera 610. That is, the image processing apparatus 100 takes in a plurality of pieces of image data captured by the camera 610 as a plurality of input images, and performs image processing to obtain one or more output images. The image processing apparatus 100 supplies the output image to the image inspection apparatus 600.

  The image inspection apparatus 600 inputs one or more output images from the image processing apparatus 100 as inspection images. The image inspection apparatus 600 performs image analysis of the inspection image. As a result of the image analysis, the image inspection apparatus 600 detects an abnormality of the object to be measured OJ (for example, a defect of the object to be measured OJ, a foreign matter attached to the object to be measured OJ, etc.).

  As described above, the appearance inspection system according to the second embodiment generates an inspection image by performing the same or similar image processing as the image processing apparatus according to the first embodiment. Therefore, according to the appearance inspection system according to the present embodiment, since an inspection image generated by good shading correction can be used, high-accuracy inspection is possible.

(Third embodiment)
The X-ray inspection system according to the third embodiment includes an image processing apparatus 100, an image inspection apparatus 700, an X-ray source 710, a carbon stage 720, and an X-ray image sensor 730, as illustrated in FIG.

  The carbon stage 720 is a stage having a table-like plane on which the object to be measured OJ to be subjected to X-ray inspection is placed. The object to be measured OJ is placed in a range where X-rays are irradiated from the X-ray source 710 on the carbon stage 720. X-ray imaging is performed by the X-ray source 710 and the X-ray image sensor 730 while the object to be measured OJ is stationary on the carbon stage 720.

  The X-ray source 710 irradiates the object to be measured OJ placed on the carbon stage 720 with X-rays. The X-ray image sensor 730 detects X-rays irradiated by the X-ray source 710 and transmitted through the object to be measured OJ, and generates image data. The X-ray image sensor 730 outputs the image data to the subsequent image processing apparatus 100.

  The image processing apparatus 100 performs the same or similar image processing as in the first embodiment on the image data obtained by the X-ray image sensor 730. In other words, the image processing apparatus 100 takes in a plurality of pieces of image data captured by the X-ray image sensor 730 as a plurality of input images, and performs image processing to obtain one or more output images. The image processing apparatus 100 supplies the output image to the image inspection apparatus 700.

  The image inspection apparatus 700 inputs one or more output images from the image processing apparatus 100 as inspection images. The image inspection apparatus 700 performs image analysis of the inspection image. The image inspection apparatus 700 detects an abnormality of the object to be measured OJ as a result of the image analysis.

  As described above, the X-ray inspection system according to the third embodiment generates an inspection image by performing the same or similar image processing as that of the image processing apparatus according to the first embodiment. Therefore, according to the X-ray inspection system according to the present embodiment, since an inspection image generated by good shading correction can be used, a highly accurate inspection can be performed.

  Note that it is also conceivable that the X-ray inspection system according to the present embodiment is applied for medical diagnosis and a human (for example, a doctor) analyzes the inspection image. In such a case, the image inspection apparatus 700 is not necessary. Even when a human analyzes an inspection image, it is obvious that a high-accuracy inspection can be performed because an inspection image generated by good shading correction can be used.

  The processing of each of the above embodiments can be realized by using a general-purpose computer as basic hardware. The program for realizing the processing of each of the above embodiments may be provided by being stored in a computer-readable storage medium. The program is stored in the storage medium as an installable file or an executable file. Examples of the storage medium include a magnetic disk, an optical disk (CD-ROM, CD-R, DVD, etc.), a magneto-optical disk (MO, etc.), and a semiconductor memory. The storage medium may be any as long as it can store the program and can be read by the computer. Further, the program for realizing the processing of each of the above embodiments may be stored on a computer (server) connected to a network such as the Internet and downloaded to the computer (client) via the network.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  DESCRIPTION OF SYMBOLS 100 ... Image processing apparatus, 101 ... Image acquisition part, 102 ... Image selection part, 103 ... Reference image creation part, 104 ... Image correction part, 105 ... User interface part, 300 500, reference image creation window, 301 ... start image display area, 302, 306, 502 ... slide bar, 303, 307, 503 ... slider, 304, 308, 504 ... box, 305 ... End image display area, 309, 507 ... Create button, 310, 508 ... Reference image display area, 311, 509 ... Save button, 312, 510 ... End button, 501 ... Input image display area, 505... Selection button, 506... Non-selection button, 600, 700 ... image inspection device, 610. Camera, 620 ... stage, 710 ... X-ray source, 720 ... carbon stage, 730 ... X-ray image sensor

Claims (9)

By selecting two or more of a plurality of images not including the target subject from a plurality of input images including one or more images including the target subject and a plurality of images not including the target subject, A selection unit for obtaining a plurality of images to be averaged;
An image processing apparatus comprising: a creation unit that creates a reference image for shading correction by averaging the plurality of images to be averaged. A user interface unit for exchanging information with the user;
The selection unit selects two or more of a plurality of images not including the target subject from the plurality of input images according to a user input via a GUI presented by the user interface unit.
The image processing apparatus according to claim 1. The user input selects any one of the plurality of input images as a start image, and selects any one of the plurality of input images as an end image,
The selection unit obtains the averaging target image by selecting a plurality of continuous images positioned from the start image to the end image among the plurality of input images.
The image processing apparatus according to claim 2. The image processing apparatus according to claim 3 , wherein the GUI includes at least one of a start image display area for displaying the start image and an end image display area for displaying the end image.   The image processing apparatus according to claim 2, wherein the GUI includes a reference image display area for displaying the reference image.   The selection unit performs pattern matching using any one of a plurality of images not including the target subject as a reference image, thereby performing a plurality of sheets not including the target subject from the plurality of input images. The image processing apparatus according to claim 1, wherein two or more images are selected. A user interface unit for exchanging information with the user;
The selection unit executes the pattern matching using a threshold value specified by a user input via a GUI presented by the user interface unit, thereby not including a plurality of target subjects from the plurality of input images. Select two or more of the images,
The image processing apparatus according to claim 6. At least one of the first plurality and the last plurality of the input images is generated so as not to include the subject of interest,
The selection unit obtains the plurality of images to be averaged by automatically selecting at least one of the first plurality and the last plurality.
The image processing apparatus according to claim 1. By selecting two or more of a plurality of images not including the target subject from a plurality of input images including one or more images including the target subject and a plurality of images not including the target subject, Obtaining multiple images to average,
An image processing method comprising: creating a reference image for shading correction by averaging the plurality of images to be averaged.

Images