JP2015176433A - Image processing device, image processing method, and image processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing device, an image processing method and an image processing program capable of improving the accuracy of template matching processing using a frequency component.SOLUTION: An image processing device 100 includes a window function processing unit 220 for performing window function processing of each of a template image and an input image, a collation unit 230 for performing frequency conversion of each of the template image and the input image after the window function processing to match the template image and the input image after frequency conversion, and a correction unit 210 for selectively performing negative/position inversion processing of the template image and the input image in independence on a feature amount included in at least one of the template image and the input image before window function processing.

Description

  The present invention relates to an image processing apparatus, an image processing method, and an image processing program that perform template matching.

  In recent years, an image processing technique called template matching for detecting a position corresponding to a template image registered in advance from search target images has been put into practical use. As a template matching technique, a phase-only correlation method (POC method) is often used. The POC method is a method of searching for corresponding points between images using phase difference information of frequency components included in the images. By using the phase difference information, it is possible to improve the robustness against illumination fluctuations.

  In addition, various methods have been proposed as a template matching technique using the frequency component of the image. For example, Patent Document 1 discloses a “corresponding point search technique capable of achieving both maintenance of accuracy and speed improvement in corresponding point search processing for a plurality of images”. In the corresponding point search technique, for each of the two images, the frequency component of the image region of the image is extracted, and the correlation value between the image regions is calculated using each extracted frequency component.

  Patent Document 2 discloses an “image integration device that integrates time-series images”. The image integration device extracts a stationary body region for each time-series image and calculates a correlation of frequency components in each stationary body region between the time-series images. Accordingly, the image integration device searches for a corresponding stationary body region between the time series images and integrates the time series images.

  Patent Document 3 discloses "a compound eye imaging apparatus that combines a plurality of images obtained by imaging a common subject using a plurality of imaging systems to obtain a high-definition image". The compound-eye imaging apparatus adaptively changes a calculation area for obtaining a correspondence relationship based on frequency components of image signals obtained by a plurality of imaging systems, and obtains a correspondence relationship between the image signals based on the obtained correspondence relationship. A plurality of image signals are synthesized.

JP 2011-170519 A International Publication No. 2010/113239 Japanese Patent Laid-Open No. 06-133207

  By the way, when template matching is performed using the frequency component of the image, the frequency component is extracted from the image by Fourier transform such as FFT (Fast Fourier Transform). The POC process described above also applies Fourier transform. When Fourier transform is performed, theoretically, infinitely long image data is required. Actually, since image data is finite, when performing Fourier transform on an image, it is assumed that the finite image data is periodic infinite image data. When image data is treated as data that is repeated periodically indefinitely, discontinuous points in which data (pixel values) change abruptly occur at the end points of the image. This discontinuous point appears as noise of a high frequency component in the frequency space.

  In order to reduce such noise due to the discontinuity of the image data, a window function is multiplied to the image before Fourier transform. The window function is a function having a value of 0 outside a certain finite region. When the image is multiplied by the window function, the pixel value becomes 0 in the image area outside the finite area, and the pixel value remains in the image area in the finite area. As a result, the image after the window function processing has a pixel value of 0 at the upper and lower ends of the image and the left and right ends of the image, so that the image signal is assumed to be a periodic infinite signal. Data is continuous. Therefore, high frequency component noise can be removed in the frequency space.

  However, since the pixel value remains in the image area within the finite area of the window function, noise due to the pixel value in the image area may appear in the frequency space. That is, the signal component of the image data may be deteriorated due to strong noise caused by the window function. Thereby, the precision of template matching will fall.

  This disclosure has been made to solve the above-described problems, and an object thereof is an image processing apparatus, an image processing method, and an image processing apparatus capable of improving the accuracy of template matching processing using frequency components. And providing an image processing program.

  According to one embodiment, the image processing device includes a window function processing unit for performing window function processing on each of the first image and the second image, and the first image and the second image after the window function processing. Frequency conversion of each of the images, a matching unit for matching the first image and the second image after frequency conversion, and a feature amount included in at least one of the first image and the second image Accordingly, a correction unit for selectively performing a negative / positive inversion process on the first image and the second image before the window function process is provided.

  Preferably, the image processing apparatus further includes a determination unit for determining whether the negative / positive inversion process in the correction unit is necessary, using at least one of the first image and the second image as the determination image. Prepare.

  Preferably, the determination image includes a non-object region in the determination image that does not include an object to be inspected. The determination unit uses the pixel value in the non-object region for feature amount extraction.

  Preferably, the feature amount includes a statistical value corresponding to the size of each pixel in the image region. The determination unit determines that the negative / positive inversion process is necessary when the statistical value in the non-object region is larger than a predetermined value.

  Preferably, the statistical value is an average value, a median value, a mode value of each pixel value in the non-object region, and a ratio of the number of pixels having a predetermined value or more to the number of pixels in the non-object region. And at least one of

  Preferably, the determination image further includes an object region in the determination image including an object to be inspected. The determination unit uses each pixel value in the object region and each pixel value in the previous non-object region for feature amount extraction.

  Preferably, the determination unit compares the determination image with an image obtained by performing window function processing on the determination image, and determines whether or not negative / positive inversion processing is necessary.

  Preferably, the feature amount includes a statistical value indicating similarity between the determination image and an image obtained by performing window function processing on the determination image. The determination unit determines that the negative / positive inversion process is necessary when the statistical value is smaller than a predetermined value.

  Preferably, the determination unit compares the image obtained by performing window function processing on the determination image with the image obtained by performing negative / positive inversion processing on the determination image and further performing window function processing, It is determined whether or not negative / positive inversion processing is necessary.

  Preferably, in addition to the first image and the second image, the determination image is selected from an image obtained by negative-positive reversal of the first image and an image obtained by negative-positive reversal of the second image.

  According to another embodiment, an image processing method includes a step of performing window function processing on each of a first image and a second image, and each of the first image and the second image after the window function processing. The frequency conversion of the first image and the second image after frequency conversion, and depending on the feature amount included in at least one of the first image and the second image, Selectively performing a negative / positive inversion process on the image and the second image before the window function process.

  According to yet another embodiment, an image processing program is provided. The image processing program causes the computer to perform window function processing on each of the first image and the second image, and frequency-convert each of the first image and the second image after the window function processing. , Matching the first image and the second image after frequency conversion, and depending on a feature amount included in at least one of the first image and the second image, the first image and the second image A step of selectively performing a negative / positive inversion process on the image before the window function process.

  According to the present invention, the accuracy of template matching processing using frequency components can be improved.

  The above and other objects, features, aspects and advantages of the present invention will become apparent from the following detailed description of the present invention taken in conjunction with the accompanying drawings.

1 is a schematic diagram showing an overall configuration of an image processing system including an image processing device according to a first embodiment. It is the figure which showed the image obtained by multiplying a certain image with a window function. It is the figure which showed the image obtained by performing negative positive inversion processing with respect to a certain image. It is a block diagram which shows an example of a function structure of the image processing apparatus according to 1st Embodiment. It is a block diagram which shows an example of a function structure of the collation part contained in the image processing apparatus according to 1st Embodiment. It is a block diagram which shows an example of the hardware constitutions of the image processing apparatus according to 1st Embodiment. It is a flowchart showing a part of process which the image processing apparatus according to 1st Embodiment performs. It is a block diagram which shows an example of a function structure of the image processing apparatus according to 2nd Embodiment. It is a flowchart showing a part of process which the image processing apparatus according to 2nd Embodiment performs. 10 is a flowchart illustrating an example of processing of a determination unit according to a specific example 1; It is a figure which shows a mode that the determination image is divided | segmented into a work area | region and a non-work area | region. It is a figure which shows a mode that the determination image is divided | segmented into several blocks. 12 is a flowchart illustrating an example of processing of a determination unit according to Specific Example 2. 12 is a flowchart illustrating an example of processing of a determination unit according to a specific example 3; It is a conceptual diagram showing the outline of the process of the determination part according to the specific example 3. 10 is a flowchart illustrating an example of processing of a determination unit according to a fourth specific example. It is a conceptual diagram showing the outline of the process of the determination part according to the specific example 4. 12 is a flowchart illustrating an example of processing of a determination unit according to a specific example 6; It is a conceptual diagram showing the outline of the process of the determination part according to the specific example 6. It is a block diagram which shows an example of a function structure of the image processing apparatus according to 2nd Embodiment. It is a block diagram which shows an example of a function structure of the image processing apparatus according to 2nd Embodiment. 10 is a flowchart illustrating an example of processing of a determination unit according to a specific example 8. 22 is a flowchart illustrating an example of processing of a determination unit according to Specific Example 9;

  Hereinafter, the present embodiment will be described with reference to the drawings. In the following description, the same parts and components are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated. Each embodiment described below and / or each modification may be selectively combined.

<First Embodiment>
[System configuration]
FIG. 1 is a schematic diagram showing an overall configuration of an image processing system 50 including an image processing apparatus 100 according to the present embodiment. Referring to FIG. 1, for example, the image processing system 50 inspects the position and orientation of an inspection object (hereinafter also referred to as “work”). More specifically, the image processing system 50 performs template matching with an image acquired by photographing the workpiece 116 using a template image incorporated in a production line or the like and registered in advance.

  More specifically, the image processing system 50 includes an image processing device 100, a camera 110, a manufacturing device 112, a transport mechanism 114, a display device 120, and a mouse 130.

  The manufacturing apparatus 112 is an apparatus for manufacturing the workpiece 116. For example, the workpiece 116 includes mechanical parts such as a circuit board. When the workpiece 116 is manufactured by the manufacturing apparatus 112, the workpiece 116 is transferred by a transfer mechanism 114 such as a belt conveyor. The conveyed workpiece 116 is photographed by the camera 110 at a predetermined timing. An image obtained by the camera 110 is transmitted to the image processing apparatus 100.

  The image processing apparatus 100 has a registration mode for performing template image registration processing and an inspection mode for executing inspection on the workpiece 116. The user can alternately switch between the inspection mode and the registration mode using an input device such as a mouse 118. In the registration mode, the image processing apparatus 100 registers an image obtained by photographing the reference target workpiece 116 as a template image. In the inspection mode, the image processing apparatus 100 performs template matching using an image (hereinafter, also referred to as “input image”) obtained by imaging an inspection target such as the workpiece 116 and a template image. Thereby, the image processing system 50 can automatically inspect for defects or the like of the workpiece 116.

[Overview]
An overview of image processing apparatus 100 according to the first embodiment will be described with reference to FIGS. 2 and 3. FIG. 2 is a diagram showing an image obtained by multiplying a certain image by a window function. FIG. 3 is a diagram showing an image obtained by performing negative / positive inversion processing on a certain image. 2 and 3, the closer the pixel value is to 0, the closer the color of the image is to black, and the larger the pixel value (typically, the closer the pixel value is to 255), the color of the image becomes white. Shown to be close.

  FIG. 2A shows a window function image 34A obtained by the image processing apparatus 100 multiplying the image 30A by the window function 32. FIG. As described above, the image processing apparatus 100 multiplies an image with a window function before Fourier transform in order to reduce noise in the frequency space. The window function 32 is a function having a value of 0 outside a certain finite region, and a function having a value larger than 0 within the finite region. For example, in the window function 32 shown in FIG. 2A, the finite region is indicated by a white or gray portion (central region).

  In the central region of the window function image 34A obtained by multiplying the window functions 32, the pixel value becomes 0 or more. In the peripheral image area of the window function image 34A, the pixel value is 0 (black). Here, depending on how the pixel values in the central region of the window function image 34A remain, there is a possibility that noise is generated. This noise occurs when the pixel value is large (white) in an area where the work 116 of the image 30A is not shown (hereinafter also referred to as “non-work area”). An example of the noise generated by this is shown in the window function image 34A of FIG.

  In order to reduce this noise, image processing apparatus 100 according to the present embodiment corrects image 30A so that the pixel value of the non-work area becomes small. For example, as illustrated in FIG. 3, when the pixel value of the non-work area is large, the image processing apparatus 100 reduces the pixel value of the non-work area by performing negative / positive inversion processing on the image 30 </ b> A. Here, the negative / positive inversion processing is processing for reversing black and white in an image. That is, the image processing apparatus 100 reverses black and white by subtracting each pixel value of the image 30A from 255 (pixel value).

  As described above, the image processing apparatus 100 performs negative / positive inversion processing on the image 30A to obtain the image 30B, thereby reducing the pixel value in the non-work area. 2B, even if the image 30B is multiplied by the window function 32, the obtained image 34B does not contain noise in the central region. Therefore, the image processing apparatus 100 can reduce noise caused by the window function 32 and can realize a highly accurate template matching process.

  2 and 3, the closer the pixel value is to 0, the closer the image color is to black, and the larger the pixel value (typically, the closer the pixel value is to 255), the image color is white. However, the relationship between the size of the pixel value and the color is not limited to this. For example, the closer the pixel value is to 0, the closer the image color is to white, and the larger the pixel value (typically the closer the pixel value is to 255), the closer the image color is to black. May be. The image is not necessarily a black and white image, and may be a color image. The same applies to the drawings shown below.

[Function configuration]
With reference to FIG. 4, the functional configuration of image processing apparatus 100 according to the first embodiment will be described. FIG. 4 is a block diagram illustrating an example of a functional configuration of the image processing apparatus 100. The image processing apparatus 100 includes a correction unit 210, a window function processing unit 220, and a collation unit 230.

  The correction unit 210 selectively performs negative / positive inversion processing on the input image and the template image before the window function processing, depending on the feature amount included in at least one of the input image and the template image. Typically, the correction unit 210 is configured to execute the negative / positive inversion process by the user of the image processing apparatus 100 when the determination unit 200 described below determines that the negative / positive inversion process is necessary. Perform negative / positive inversion processing.

  The window function processing unit 220 multiplies each of the input image and the template image by a window function. The window function processing unit 220 matches the sizes of the input image or template image and the window function 32 (see FIG. 2) by the enlargement / reduction process. As a result, the window function processing unit 220 can multiply the input image and the template image by the window function. Examples of window functions include rectangular windows, gauss windows, hamming windows, tukey windows, black man windows, kaiser windows, exponential windows, generalized hamming windows, natto windows, black man-harris windows, black man natto windows, flats. -Top windows, Parzan windows, Akaike windows, Welch windows, sign windows, Vorbis windows, Kaiser-Bessel derived windows, Lanzosh windows, etc.

  The collation unit 230 performs frequency conversion on each of the input image and the template image after the window function processing, and matches the input image and template image that have been subjected to frequency conversion. Template matching is an example of the matching method. Template matching includes, for example, phase-only correlation (POC), rotation invariant phase only correlation (RIPOC), matching processing using Fourier Merin transform, matching processing using other frequency components, and the like. .

(Details of collation unit 230)
With reference to FIG. 5, RIPOC, which is an example of template matching processing by collation unit 230, will be described in detail. FIG. 5 is a block diagram illustrating an example of a functional configuration of the collation unit 230. The RIPOC process calculates how much the input image is rotated with respect to the template image, corrects the rotation amount of the template (reference numeral 230-1), and the corrected template image with the corrected rotation amount. And a process (reference numeral 230-2) for calculating how much similarity and translation amount exist between the input image (or its extraction region).

  As shown in FIG. 5, the collation unit 230 includes Fourier transform units 231A and 231B, logarithmic processing units 232A and 232B, polar coordinate conversion units 233A and 233B, a POC processing unit 234, and a rotation amount correction unit 235. , And POC processing unit 236.

  The Fourier transform unit 231A calculates frequency information (amplitude component and phase component) included in the template image. The Fourier transform unit 231B sets a search range around a position where the detection target object is likely to exist in the input image, and cuts out a template portion corresponding to the search range from the input image. Typically, the size of the region to be cut out is the same size as the template image. Further, the Fourier transform unit 231B calculates frequency information (amplitude component and phase component) included in the cut out region. In the RIPOC process, the phase component is not necessarily required and may not be calculated.

  The logarithmic processing unit 232A and the polar coordinate conversion unit 233A logarithmize the amplitude component of the template image and convert it to polar coordinates. Similarly, the logarithmic processing unit 232B and the polar coordinate conversion unit 233B logarithmize the amplitude component of the input image and convert it to polar coordinates. By this conversion to polar coordinates, the rotation amount is expressed as a coordinate point on two-dimensional coordinates.

  The POC processing unit 234 calculates a similarity and a parallel movement amount (corresponding to a rotation amount) for the result of the polar coordinate conversion output from the polar coordinate conversion unit 233A and the polar coordinate conversion unit 233B, respectively. In the POC processing, the template image and the input image are sequentially shifted, and the correlation value is calculated between the components of the spatial frequency included in each, thereby obtaining the highest similarity (POC peak value). Explore. The POC processing unit 234 specifies the position where the similarity is the highest among the results of the polar coordinate conversion, and outputs the corresponding rotation amount to the rotation amount correction unit 235.

  The rotation amount correction unit 235 corrects the rotation of the template image according to the rotation amount calculated by the POC processing unit 234. That is, the POC processing unit 234 generates a correction template by rotationally correcting the template image.

  As a method of rotating and correcting the template image in the rotation amount correcting unit 235, a method of rotating the template image itself in real space can be employed. When the frequency space is handled in the POC processing unit 236, it is not necessary to rotate in the real space, and a method of rotating the data (amplitude information and phase information) after Fourier transform, which is an internal representation of the POC processing, is adopted. May be.

  The POC processing unit 236 calculates a similarity and a position (a parallel movement amount) between the correction template image and the input image. The position with the highest similarity indicates a region that matches the correction template image included in the input image. The POC processing unit 236 outputs position information including information such as a translation amount, a rotation amount, and a magnification.

[Hardware configuration]
FIG. 6 is a block diagram illustrating an example of a hardware configuration of the image processing apparatus 100. A hardware configuration of the image processing apparatus 100 will be described with reference to FIG.

  Referring to FIG. 6, image processing apparatus 100 is mainly implemented on a computer having a general-purpose architecture. The image processing apparatus 100 includes, as main components, a ROM (Read Only Memory) 1, a CPU (Central Processing Unit) 2, a RAM (Random Access Memory) 3, a camera interface (I / F) 4, and a memory card interface. (I / F) 5, network interface (I / F) 6, and storage device 20 are included.

  The ROM 1 stores an initial program (boot program) that is executed when the image processing apparatus 100 is started up. The CPU 2 controls the entire image processing apparatus 100 by executing various programs such as an operating system (OS) and an image processing program 22 stored in the ROM 1 and the storage device 20. The RAM 3 functions as a working memory for executing a program by the CPU 2 and temporarily stores various data necessary for executing the program.

  The camera I / F 4 mediates data communication between the CPU 2 and the camera 110. For example, the camera I / F 4 includes an image buffer and temporarily stores the input image 24 transmitted from the camera 110. When at least one input image data is accumulated, the camera I / F 4 transfers the accumulated data to the storage device 20 or the ROM 1. The camera I / F 4 gives an imaging command to the camera 110 in accordance with an internal command generated by the CPU 2. The camera 110 may be incorporated in the image processing apparatus 100.

  The memory card I / F 5 reads / writes data from / to various memory cards (nonvolatile storage media) 5A such as an SD (Secure Digital) card and a CF (Compact Flash (registered trademark)) card. Typically, the memory card I / F 5 is mounted with a memory card 5A storing an input image 24 and a template image 26 acquired by another device, and the input image 24 and the template read from the memory card 5A are loaded. The image 26 is stored in the storage device 20.

  The network I / F 6 exchanges data with other devices (such as server devices) via various communication media. More specifically, the network I / F 6 is connected via a wired line such as Ethernet (registered trademark) (LAN (Local Area Network) or WAN (Wide Area Network)) and / or a wireless line such as a wireless LAN. Data communication.

  The storage device 20 typically includes a large-capacity magnetic storage medium such as a hard disk. The storage device 20 stores an image processing program 22 for realizing various types according to the present embodiment, an input image 24, and a template image 26. Further, the storage device 20 may store a program such as an operating system.

  The camera 110 may be incorporated in the image processing apparatus 100 instead of being externally attached, and the image processing apparatus 100 main body may have a function of capturing an image of a subject. The image processing apparatus 100 acquires the input image 24 and the template image 26 using a mechanism similar to a camera, and inputs the acquired input image 24 and template image 26 to the image processing apparatus 100 by an arbitrary method. It may be configured. In this case, the input image 24 and the template image 26 are input to the image processing apparatus 100 via the memory card I / F 5 or the network I / F 6.

  The image processing program 22 stored in the storage device 20 is stored and distributed in a storage medium such as a CD-ROM (Compact Disk-Read Only Memory) or distributed from a server device or the like via a network. The image processing program 22 may call a required module among program modules provided as a part of the operating system executed by the image processing apparatus 100 at a predetermined timing and order to realize the processing. In this case, the image processing program 22 itself does not include a module provided by the operating system, and image processing is realized in cooperation with the operating system.

  Further, the image processing program 22 may be provided by being incorporated in a part of an arbitrary program instead of a single program. Even in such a case, the image processing program 22 itself does not include a module that is commonly used in an arbitrary program, and image processing is realized in cooperation with the arbitrary program. Even such an image processing program 22 that does not include some modules does not depart from the spirit of the image processing apparatus 100 according to the present embodiment. Furthermore, some or all of the functions provided by the image processing program 22 may be realized by dedicated hardware.

  Furthermore, the image processing apparatus 100 does not necessarily perform processing in real time. For example, the image processing apparatus 100 may be configured in a form such as a so-called cloud service in which at least one server apparatus realizes processing according to the present embodiment. In this case, the input image 24 and the template image 26 are transmitted to the server device (cloud side), and the server device performs image processing according to the present embodiment on the received input image 24 and template image 26. Further, it is not necessary for the server device side to perform all functions (processing), and the user side terminal and the server device may cooperate to realize the image processing according to the present embodiment.

[flowchart]
The control structure of the image processing apparatus 100 will be described with reference to FIG. FIG. 7 is a flowchart showing a part of processing executed by the image processing apparatus 100. The processing in FIG. 7 is realized by the CPU 2 executing a program. In other aspects, some or all of the processing may be performed by circuit elements or other hardware.

  In step S <b> 510, the CPU 2 acquires a template image from the camera 110. In step S <b> 512, the CPU 2 acquires an input image from the camera 110. In step S520, the CPU 2 determines, as the correction unit 210, whether or not the feature amount extracted from at least one of the input image and the template image satisfies a predetermined condition. Typically, the CPU 2 determines that a predetermined condition is satisfied when the determination unit 200 described later determines that negative / positive inversion processing is required for the input image and the template image. When CPU 2 determines that the predetermined condition is satisfied (YES in step S520), CPU 2 switches control to step S522. If not (NO in step S520), CPU 2 switches control to step S524.

  In step S522, the CPU 2 performs a negative / positive inversion process on the input image and the template image as the correction unit 210. In step S524, the CPU 2 performs window function processing on each of the input image and the template image as the window function processing unit 220. In step S526, the CPU 2 performs frequency transformation such as Fourier transformation for each of the input image and the template image after the window function processing as the collation unit 230.

  In step S528, the CPU 2 uses the frequency component extracted from each of the input image and the template image as the matching unit 230 to match the input image and the template image. Typically, the CPU 2 performs matching processing on the frequency-converted image using the POC method or the RIPOC method. In step S530, the CPU 2 calculates the similarity between the input image and the template image from the template matching result as the matching unit 230.

[advantage]
When performing template matching using frequency components, when using an image that contains many pixels with large pixel values in the non-work area, noise caused by the window function remains when window function processing is performed. May end up. If frequency conversion is performed in this state, the image processing apparatus 100 deteriorates the signal component of the template image or the input image, and the accuracy of template matching decreases. Focusing on the fact that pixels with small pixel values are not easily affected by the window function, the image processing apparatus 100 creates an image that is not affected by the window function by negative / positive inversion of the template image and the input image. Thereby, the image processing apparatus 100 can suppress signal deterioration that occurs when the window function process is executed, and can further improve the accuracy of the template matching process.

<Second Embodiment>
[Overview]
Hereinafter, an overview of image processing apparatus 100A according to the second embodiment will be described. Image processing apparatus 100A according to the present embodiment is different from image processing apparatus 100 in that it has a function of determining whether or not to perform correction processing (that is, negative / positive inversion processing) on an input image and a template image. Since the hardware configuration is the same as that of image processing apparatus 100 according to the first embodiment, description thereof will not be repeated.

  Image processing apparatus 100A according to the present embodiment uses at least one of the input image and the template image to determine whether correction processing (that is, negative / positive inversion processing) for the input image and the template image is necessary. When it is determined that correction is necessary, the image processing apparatus 100A performs negative / positive inversion processing on the input image and the template image. If it is determined that correction is unnecessary, the image processing apparatus 100A does not perform negative / positive inversion processing on the input image and the template image.

  The image processing apparatus 100A performs correction determination processing on an input image obtained by sequentially photographing the workpieces flowing through the line, thereby making it possible to perform correction in accordance with a change in the environment. Become. That is, the image processing apparatus 100A can execute a template matching process that is resistant to environmental changes.

[Function configuration]
With reference to FIG. 8, the functional configuration of image processing apparatus 100A according to the second embodiment will be described. FIG. 8 is a block diagram illustrating an example of a functional configuration of the image processing apparatus 100A. The image processing apparatus 100A includes a determination unit 200, a correction unit 210, a window function processing unit 220, and a collation unit 230. Since correction unit 210, window function processing unit 220, and collation unit 230 are the same as described above, description thereof will not be repeated.

  The determination unit 200 performs correction determination using the template image as a determination image. Note that the determination unit 200 may use any of an input image, an image obtained by inverting the input image with negative and positive images, and an image obtained by inverting the negative and positive images instead of the template image. For example, the determination unit 200 may perform correction determination only once when a template image is first input, and apply the result to the second and subsequent input images, or each time an input image is input. Correction determination may be performed. Various methods can be considered for the correction determination method in the determination unit 200. Details of these methods will be described later.

[flowchart]
A control structure of the image processing apparatus 100A will be described with reference to FIG. FIG. 9 is a flowchart showing a part of processing executed by image processing apparatus 100A. The processing in FIG. 9 is realized by the CPU 2 executing a program. In other aspects, some or all of the processing may be performed by circuit elements or other hardware. Since each step other than step S600 is as described above, description thereof will not be repeated.

  In step S600, the CPU 2 uses the determination image as the determination unit 200 to determine whether correction (that is, negative / positive inversion processing) for the input image and the template image is necessary using the determination image. Details of the correction determination method will be described later.

[Details of Correction Determination Method of Determination Unit 200]
(Specific example 1 of correction determination method)
With reference to FIGS. 10-12, an example of the process of the determination part 200 is demonstrated. FIG. 10 is a flowchart illustrating an example of processing of the determination unit 200 according to the first specific example. FIG. 11 is a diagram illustrating a state in which the determination image 36 is divided into a work area and a non-work area. FIG. 12 is a diagram illustrating a state in which the determination image 36 is divided into a plurality of blocks.

  As described above, the noise caused by the window function occurs when each image value in a non-work area that does not include a work in the image area is large (white). That is, the noise does not occur when the image value in the non-work area of the image is small (black). Focusing on this point, the determination unit 200 determines that correction processing is necessary when the pixel value in the non-work area of the determination image is large.

  As a more specific processing procedure, in step S602, the determination unit 200 acquires a determination image. The determination unit 200 selects, as the determination image 36, at least one of an input image, a template image, an image obtained by performing negative / positive inversion processing on the input image, and an image obtained by performing negative / positive reversal processing on the template image. Can be used.

  In step S604, the determination unit 200 specifies at least a non-work area from the acquired determination image. For example, as illustrated in FIG. 11, the determination unit 200 divides a work area and a non-work area. As a specific area dividing method, for example, the determination unit 200 identifies a pixel having the strongest edge in the determination image 36, sets a work area around the pixel, and sets other areas as non-work areas. Set. Thus, by automatically dividing an area using edge information, it is possible to adaptively set a work area and a non-work area each time an image is input.

  The determination unit 200 may set a predetermined area as a work area. For example, the determination unit 200 sets a central block as a work area among areas divided into nine blocks of 1/3 vertical and horizontal, and sets other blocks as non-work areas. In addition, as illustrated in FIG. 12, the determination unit 200, in an area obtained by dividing the input image into 25 blocks of 1/5 vertical and horizontal directions, is inclined vertically and horizontally with respect to the central block and the central block. The block located is set as a work area, and the other square-shaped area is set as a non-work area. Note that the number of blocks that the determination unit 200 divides is arbitrary.

  In step S606, the determination unit 200 performs correction determination using an image value in at least the non-work area of the work area and the non-work area. The determination unit 200 determines that correction processing is necessary when the pixel value in the non-work area is relatively large (white). That is, the determination unit 200 determines that the correction process is unnecessary when the pixel value in the non-work area is relatively small (black).

  More specifically, the determination unit 200 calculates a statistical value corresponding to the size of each pixel value in the non-work area, and when the statistical value is larger than a predetermined value, correction processing (that is, negative / positive inversion) Process). Further, the determination unit 200 may determine that the negative / positive inversion process is not performed when the statistical value in the non-work area is smaller than a predetermined value. Note that the statistical values here are, for example, the average value, median value, mode, and ratio of the number of pixels having a predetermined value or more with respect to the number of pixels in the non-work area. Including. The said ratio is shown by the following formula | equation (1), for example.

Statistical value (pixel ratio) = number of pixels having a pixel value greater than or equal to a predetermined value in the non-work area / number of pixels in the non-work area (1)
The determination unit 200 determines to perform the negative / positive inversion process when the input image or the template image is used as the determination image and the ratio of the pixels shown in the equation (1) satisfies the following equation (2). To do.

Statistical value (pixel ratio) ≧ threshold value (2)
For example, when the number of pixels in the non-work area is 500, the number of pixels having a pixel value of 128 or more among the pixels included in the non-work area is 400, and the threshold value is 0.5, the following formula ( It becomes like 3). In this case, since the following equation (3) is satisfied, the determination unit 200 determines that correction is necessary.

Statistical value (pixel ratio) = 400/500 = 0.8 ≧ 0.5 Expression (3)
In step S520, the determination unit 200 determines whether correction for the input image and the template image is necessary using the statistical value calculated from the non-work area. Typically, the determination unit 200 determines that correction is necessary when the statistical value satisfies the above equation (2). When determining unit 200 determines that correction is necessary (YES in step S520), control is switched to step S522. Otherwise (NO in step S520), determination unit 200 switches control to step S524.

  The determination unit 200 inverts the magnitude relationship between the expression (2) and other determination conditions when an image obtained by performing negative / positive reversal processing on the input image or an image obtained by performing negative / positive reversal processing on the template image is used as the determination image. To do. In FIG. 11, the work area and the non-work area are indicated by rectangles, but the shapes of the work area and the non-work area are arbitrary. For example, the work area and the non-work area may be a circle, an ellipse, a hexagon, an area surrounded to include a work, or other shapes. Further, the non-work area may be all or a part of the area other than the work area.

(Specific example 2 of processing of determination unit 200)
Hereinafter, another example of the process of the determination unit 200 will be described with reference to the flowchart of FIG. FIG. 13 is a flowchart illustrating an example of processing of the determination unit 200 according to the second specific example.

  The determination unit 200 may perform the correction determination using not only the pixel value of the non-work area but also the pixel value of the work area as in the specific example 1 described above. In this case, when the pixel value in the non-work area relative to the work area is relatively large, the determination unit 200 determines that correction processing needs to be performed.

  As a more specific processing procedure, in step S602, the determination unit 200 acquires a determination image. In step S604A, the determination unit 200 divides the acquired determination image into a work area and a non-work area. Since the region dividing method is as described above, description thereof will not be repeated.

  In step S606A, the determination unit 200 calculates a statistical value corresponding to the size of the pixel value in the identified work area and non-work area. The statistical value includes at least one of an average value, a median value, and a mode value of each pixel value in each of the work area and the non-work area. The determination unit 200 determines to perform the negative / positive inversion process when the statistical value in the non-work area is relatively larger than the statistical value in the work area. More specifically, the determination unit 200 determines to perform negative / positive inversion processing when the following expression (4) is satisfied.

(Statistical value in non-work area) − (Statistical value in work area) ≧ threshold value (4)
For example, when the average value of the pixel values in the work area is 20, the average value of the pixel values in the non-work area is 240, and the threshold value is 200, the expression (4) is expressed by the following expression (5): become that way. In this case, since the above equation (4) is satisfied, the determination unit 200 determines that correction is necessary.

240−20 = 220 ≧ 200 Formula (5)
In addition, when the determination unit 200 uses an image obtained by performing negative / positive inversion processing on the input image or an image obtained by performing negative / positive reversal processing on the template image as the determination image, there is a magnitude relationship between Expression (4) and other determination conditions. Invert.

  In step S520, the determination unit 200 determines whether correction for the input image and the template image is necessary using the statistical values calculated from the work area and the non-work area. Typically, the determination unit 200 determines that correction is necessary when the statistical value satisfies the above equation (4). When determining unit 200 determines that correction is necessary (YES in step S520), control is switched to step S522. Otherwise (NO in step S520), determination unit 200 switches control to step S524.

(Specific example 3 of processing of determination unit 200)
Below, with reference to FIG. 14 and FIG. 15, the further another example of the process of the determination part 200 is demonstrated. FIG. 14 is a flowchart illustrating an example of processing of the determination unit 200 according to the third specific example. FIG. 15 is a conceptual diagram illustrating an outline of processing of the determination unit 200 according to the third specific example.

  If the image after the window function is multiplied with the input image is not similar to the input image, it is considered that correction is necessary (see FIG. 2A). On the other hand, if the image after the window function is multiplied by the input image is similar to the input image, it is considered that no correction is required (see FIG. 2B).

  Focusing on this point, for the correction determination processing in the specific example 3, the determination unit 200 is a determination image and an image obtained by performing window function processing on the determination image (hereinafter also referred to as “window function image”). To make correction judgment. As described above, when the determination unit 200 performs correction determination using the window function image, the image processing apparatus 100A can more reliably determine whether noise due to the window function appears. Accordingly, the image processing apparatus 100A can further improve the accuracy of the template matching process performed by the collation unit 230.

  As a more specific processing procedure, in step S610, the determination unit 200 performs window function processing on the determination image 36 to obtain a window function image 38 (see FIG. 15). In step S <b> 612, the determination unit 200 uses the determination image 36 and the window function image 38 to calculate a statistical value corresponding to pixel value variation. The statistical value includes, for example, a difference between the maximum value and the minimum value of the signal strength in the frequency space for each of the determination image 36 and the window function image 38. Alternatively, the statistical value may be a variance value of pixel values for each of the determination image 36 and the window function image 38.

  In step S520, the determination unit 200 compares the statistical values calculated from each of the determination image 36 and the window function image 38. That is, the determination unit 200 determines whether or not the following expression (6) is satisfied.

(Statistical value in window function image) − (Statistical value in input image) ≧ threshold value (6)
For example, when the variance value (statistical value) calculated from the window function image is 5.2, the variance value (statistical value) calculated from the input image is 0.8, and the threshold value is 3.0 The following equation (7) is obtained. In this case, since the formula (6) is satisfied, the determination unit 200 determines that correction is necessary.

5.2-0.8 ≧ 3.0 Formula (7)
In step S520, the determination unit 200 determines whether correction for the input image and the template image is necessary using each statistical value calculated from each of the determination image 36 and the window function image 38. Typically, the determination unit 200 determines that correction is necessary when the statistical value satisfies the above equation (6). When determining unit 200 determines that correction is necessary (YES in step S520), control is switched to step S522. Otherwise (NO in step S520), determination unit 200 switches control to step S524.

(Specific example 4 of processing of determination unit 200)
Hereinafter, still another example of the process of the determination unit 200 will be described with reference to FIGS. 16 and 17. FIG. 16 is a flowchart illustrating an example of processing of the determination unit 200 according to the fourth specific example. FIG. 17 is a conceptual diagram illustrating an outline of processing of the determination unit 200 according to the fourth specific example.

  The determination unit 200 may use a difference image between the determination image and the window function image as another method for comparing the determination image of the specific example 3 and the window function image. By using the difference image, the determination unit 200 does not need to calculate statistical values for both the determination image and the window function image as in the specific example 3, and only needs to calculate the statistical values from the difference image. Time can be shortened.

  As a more specific processing procedure, in step S <b> 610, the determination unit 200 performs window function processing on the determination image 36 to obtain a window function image 38. In step S620, the determination unit 200 creates a difference image using the determination image 36 and the window function image 38, as shown in FIG.

  In step S622, the determination unit 200 calculates a statistical value using the difference image. As described above, when the correction process is not necessary, the determination image and the window function image are similar to each other, so that each pixel value in the difference image is almost zero. Focusing on this, the determination unit 200 calculates a statistical value indicating the similarity between the determination image and the window function image. More specifically, the determination unit 200 calculates a statistical value corresponding to the size of the pixel value of the difference image or a statistical value corresponding to the variation in the pixel value of the difference image as the statistical value indicating similarity. .

  For example, the statistical value includes an average value, a median value, a mode value, and the like of each pixel value of the difference image. In addition, the statistical value includes a ratio of the number of pixels having a predetermined value or more to the number of pixels in the difference image. In addition, the statistical value includes a difference between the maximum value and the minimum value of the signal intensity in the frequency space in the difference image. In addition, the statistical value includes a variance value calculated from each pixel value in the difference image.

  In step S520, the determination unit 200 determines whether the difference image is larger than a predetermined value. That is, the determination unit 200 determines whether or not the following formula (8) is satisfied.

(Statistical value in difference image) ≧ threshold value (8)
For example, when the average value (statistical value) calculated from each pixel value of the difference image is 200 and the threshold value is 128, the following equation (9) is obtained. In this case, since the formula (8) is satisfied, the determination unit 200 determines that correction is necessary.

200 ≧ 128 (9)
In step S520, the determination unit 200 determines whether correction for the input image and the template image is necessary using the statistical value calculated from the difference image. Typically, the determination unit 200 determines that correction is necessary when the statistical value satisfies the above equation (8). When determining unit 200 determines that correction is necessary (YES in step S520), control is switched to step S522. Otherwise (NO in step S520), determination unit 200 switches control to step S524.

(Specific example 5 of processing of determination unit 200)
In the specific example 3 and the specific example 4, the determination unit 200 performs the correction determination using both the determination image and the window function image. However, the determination unit 200 may perform the correction determination using only the window function image. Good. When the pixel value is large (white) in the non-work area in the window function image (see FIG. 2A), the determination unit 200 can determine that noise due to the window function has occurred. Focusing on this point, the determination unit 200 may determine that correction is performed when a statistical value corresponding to the pixel value of the non-work area in the window function image is larger than a predetermined value.

(Specific example 6 of processing of determination unit 200)
Below, with reference to FIG. 18 and FIG. 19, the further another example of the process of the determination part 200 is demonstrated. FIG. 18 is a flowchart illustrating an example of processing of the determination unit 200 according to the sixth specific example. FIG. 19 is a conceptual diagram illustrating an outline of the process of the determination unit 200 according to the sixth specific example.

  Regarding the correction determination process in the specific example 6, the determination unit 200 performs a correction process (that is, a negative / positive inversion process) on the image obtained by performing the window function process on the determination image, and the window function. The correction is determined by comparing the processed image. As described above, when the determination unit 200 performs correction determination using an image that has actually been subjected to correction processing, the image processing apparatus 100A can more reliably determine whether noise due to the window function appears. Is possible. Accordingly, the image processing apparatus 100A can further improve the accuracy of the template matching process performed by the collation unit 230.

  As a more specific processing procedure, as shown in FIG. 18, in step S <b> 630, the determination unit 200 performs a negative / positive reversal process on the determination image 36 to obtain a reversed image 40 (see FIG. 19). In step S632, the determination unit 200 performs window function processing on each of the determination image 36 and the inverted image 40, and creates a window function image 38 and an inverted window function image 42.

  In step S634, the determination unit 200 uses the window function image 38 and the inverted window function image 42 to calculate a statistical value corresponding to the variation in pixel values. The statistical value includes, for example, the difference between the maximum value and the minimum value of the signal strength in the frequency space for each of the window function image 38 and the inverted window function image 42. Alternatively, the statistical value may be a variance value of pixel values for each of the window function image 38 and the inverted window function image 42. Alternatively, the statistical value may be the number of pixels whose pixel values are within a predetermined range.

  In step S520, the determination unit 200 compares the statistical values calculated from each of the window function image 38 and the inverted window function image 42. That is, the determination unit 200 determines whether or not the following formula (10) is satisfied.

(Statistical value in window function image) − (Statistical value in inverted window function image) ≧ threshold value (10)
For example, the pixel number of each pixel of the window function image 38 is 150 within a predetermined range (100 to 156), and the pixel value of each pixel of the inverted window function image 42 is within a predetermined range (100 to 156). When the number of pixels in is 20 and the threshold is 50, the following equation (11) is obtained. In this case, since the formula (10) is satisfied, the determination unit 200 determines that correction is necessary.

150-20 ≧ 50 Formula (11)
In step S520, the determination unit 200 determines whether correction for the input image and the template image is necessary using each statistical value calculated from each of the window function image 38 and the inverted window function image 42. Typically, the determination unit 200 determines that correction is necessary when the statistical value satisfies the above equation (10). When determining unit 200 determines that correction is necessary (YES in step S520), control is switched to step S522. Otherwise (NO in step S520), determination unit 200 switches control to step S524.

(Specific example 7 of processing of determination unit 200)
Hereinafter, still another example of the process of the determination unit 200 will be described with reference to FIG. FIG. 20 is a block diagram illustrating an example of a functional configuration of the image processing apparatus 100A.

  The determination unit 200 can use not only a template image as shown in FIG. 8 but also an input image as a determination image as shown in FIG. The determination unit 200 can obtain the same effects as those of the first to sixth specific examples even when the input image is used as the determination image.

(Specific example 8 of processing of determination unit 200)
Hereinafter, still another example of the process of the determination unit 200 will be described with reference to FIGS. 21 and 22. FIG. 21 is a block diagram illustrating an example of a functional configuration of the image processing apparatus 100A. FIG. 22 is a flowchart illustrating an example of processing of the determination unit 200 according to the eighth specific example.

  As illustrated in FIG. 21, the determination unit 200 may select which of the input image and the template image is used as the determination image. When the determination unit 200 selects an image suitable for correction determination, the image processing apparatus 100A can more reliably determine whether noise due to the window function appears. Accordingly, the image processing apparatus 100A can further improve the accuracy of the template matching process performed by the collation unit 230.

  As a more specific processing procedure, as shown in FIG. 22, in step S600, the determination unit 200 uses the method of any one of the above specific examples 1 to 7 for each of the input image and the template image. Calculate the value. For example, as in the second specific example, the determination unit 200 calculates a value obtained by subtracting the average of the pixel values in the work area from the average of the pixel values in the non-work area for the input image. The determination unit 200 calculates a statistical value for the template image in the same manner.

  In step S640, the determination unit 200 compares the calculated statistical value in the input image with the statistical value in the template image. Typically, the determination unit 200 determines the input image and the template image having a larger statistical value as a determination image used for correction determination. For example, the difference value (statistical value) of the average value of the pixel values in each of the work area and the non-work area for the template image is 60, and the pixels in each of the work area and the non-work area for the input image When the difference value of the average value of the values is 190 (statistical value), the determination unit uses the input image as the determination image because the statistical value of the input image is larger than the statistical value of the template image.

(Specific example 9 of processing of determination unit 200)
Hereinafter, still another example of the process of the determination unit 200 will be described with reference to FIG. FIG. 23 is a flowchart illustrating an example of processing of the determination unit 200 according to the ninth specific example.

  As illustrated in FIG. 23, the determination unit 200 may perform correction determination using the correction determination result for both the input image and the template image. As a result, the image processing apparatus 100A can more reliably determine whether or not noise due to the window function appears. Accordingly, the image processing apparatus 100A can further improve the accuracy of the template matching process performed by the collation unit 230.

  As a more specific processing procedure, as shown in FIG. 23, in step S650, the determination unit 200 should use the method in any one of the specific examples 1 to 7 to correct an input image. A determination result indicating that is obtained. Similarly, the determination unit 200 obtains a determination result indicating whether or not the template image should be corrected.

  In step S652, if the determination unit 200 indicates that correction is necessary in any of the determination results for the input image and the template image, the correction unit 210 performs a correction process on the input image and the template image. Run. In addition, when the determination unit 200 indicates that correction is unnecessary in both the determination result for the input image and the template image, the correction unit 210 does not perform the correction process on the input image and the template image.

[advantage]
As described above, image processing apparatus 100A according to the present embodiment performs correction determination processing using an input image and a template image, thereby easily determining whether correction processing is necessary or not with relatively easy processing. Can be judged. In addition, when performing correction determination using an inverted image, the image processing apparatus 100A can compare the change width of the statistical value calculated from the pixel values in the images before and after the negative / positive inversion processing of the input image and the template image. Therefore, it can be easily determined whether or not correction is necessary.

  Furthermore, when dividing the determination image into a work area and a non-work area and performing correction determination, the image processing apparatus 100A performs a simple process of comparing the statistical values of pixels calculated in the work area and the non-work area. Correction determination can be made. As a result, the image processing apparatus 100A can improve the accuracy of the template matching process without significantly increasing the processing time.

  Furthermore, even when the image processing apparatus 100A performs correction determination on the determination image using one of the statistical values of the work area and the non-work area, due to the influence of the window function among the pixel values in the determination image. It is possible to calculate how many pixels in which the fluctuation range of the pixel value is large are included. This makes it possible to perform the determination process with a shorter processing time than when both the work area and the non-work area are used.

  Furthermore, the image processing apparatus 100A can confirm the change width of the statistical value calculated from the pixel values in the images before and after the window function processing by using the determination image and the window function image for the determination. Correction correction results can be obtained.

  Furthermore, the image processing apparatus 100A can confirm the change width of the statistical value calculated from the pixel value in the difference image before and after performing the window function processing by using the difference image between the determination image and the window function image. Therefore, a more reliable determination result can be obtained.

  Further, the image processing apparatus 100A uses the statistical value calculated using the window function image and the inverted window function image to select an image in which the influence of the window function does not occur among the window function image and the inverted window function image. can do. Thereby, the image processing apparatus 100A can obtain a more reliable determination result.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

  1 ROM, 2 CPU, 3 RAM, 4 camera I / F, 5 memory card I / F, 5A memory card, 6 network I / F, 6A antenna, 20 storage device, 22 image processing program, 24 input image, 26 template Image, 30A, 30B, 34A, 34B image, 32 window function, 36 judgment image, 38 window function image, 40 inverted image, 42 inverted window function image, 50 image processing system, 100, 100A image processing apparatus, 110 camera, 112 Manufacturing device, 114 transport mechanism, 116 workpieces, 118, 130 mouse, 120 display device, 200 determination unit, 210 correction unit, 220 window function processing unit, 230 collation unit, 231A, 231B Fourier transform unit, 232A, 232B logarithmic processing Part, 233A, 233B polar coordinate conversion part, 34,236 POC processor, 235 rotation amount correction section.

Claims (12)

A window function processing unit for performing window function processing on each of the first image and the second image;
A matching unit for frequency-converting each of the first image and the second image after the window function processing to match the first image and the second image after the frequency conversion;
Depending on a feature amount included in at least one of the first image and the second image, a negative / positive inversion process is performed on the first image and the second image before the window function process. An image processing apparatus comprising: a correction unit for performing selectively.   The image processing apparatus includes a determination unit for determining whether or not a negative / positive inversion process in the correction unit is necessary, using at least one of the first image and the second image as a determination image. The image processing apparatus according to claim 1, further comprising: The determination image does not include an object to be inspected, includes a non-object area in the determination image,
The image processing apparatus according to claim 2, wherein the determination unit uses a pixel value in the non-object region for extracting the feature amount. The feature amount includes a statistical value corresponding to the size of each pixel in the image region,
The image processing apparatus according to claim 3, wherein the determination unit determines that the negative / positive inversion process is necessary when the statistical value in the non-object region is larger than a predetermined value.   The statistical value is an average value, a median value, a mode value of each pixel value in the non-object region, and a ratio of the number of pixels having a predetermined value or more to the number of pixels in the non-object region. The image processing apparatus according to claim 4, comprising at least one of: The determination image further includes an object region in the determination image including an object to be inspected,
The image processing device according to claim 3, wherein the determination unit uses each pixel value in the object region and each pixel value in a previous non-object region for extraction of the feature amount. .   The determination unit compares the determination image with an image obtained by performing the window function process on the determination image, and determines whether the negative / positive inversion process is necessary. The image processing apparatus according to any one of the above. The feature amount includes a statistical value indicating similarity between the determination image and an image obtained by performing the window function processing on the determination image,
The image processing apparatus according to claim 7, wherein the determination unit determines that the negative / positive inversion process is necessary when the statistical value is smaller than a predetermined value.   The determination unit compares the image obtained by performing the window function processing on the determination image with the image obtained by performing the negative / positive inversion processing on the determination image and further performing the window function processing. The image processing apparatus according to claim 2, wherein it is determined whether the negative / positive inversion process is necessary.   In addition to the first image and the second image, the determination image is selected from an image obtained by negative-positive reversal of the first image and an image obtained by negative-positive reversal of the second image. The image processing apparatus according to claim 2. Performing window function processing on each of the first image and the second image;
Frequency-converting each of the first image and the second image after the window function processing, and matching the first image and the second image after the frequency conversion;
Depending on a feature amount included in at least one of the first image and the second image, a negative / positive inversion process is performed on the first image and the second image before the window function process. And an image processing method. An image processing program,
The image processing program is stored in a computer.
Performing window function processing on each of the first image and the second image;
Frequency-converting each of the first image and the second image after the window function processing, and matching the first image and the second image after the frequency conversion;
Depending on a feature amount included in at least one of the first image and the second image, a negative / positive inversion process is performed on the first image and the second image before the window function process. An image processing program for executing the step of selectively performing.