JP5266721B2 - Offset correction processing device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform offset correction of an output of an element of an imaging device while imaging a target when a usual scene is imaged. <P>SOLUTION: An offset corrective table is changed so as to eliminate the difference in output after the correction between different elements occurring when the same part of the target is imaged with the elements, thereby correcting the offset without stopping the imaging. <P>COPYRIGHT: (C)2009,JPO&amp;INPIT

Description

  The present invention relates to a technique for correcting a variation in images acquired in an infrared imaging device.

  Since each element included in the infrared imaging device has gain variation and offset variation, if the image is taken without correction, the output is not constant even if the uniform temperature surface is imaged. Therefore, it is necessary to correct variations in gain and offset so that the output of each element when imaging a uniform temperature surface is constant. At this time, correction is performed using data obtained by imaging two types of heat sources, a low temperature reference heat source and a high temperature reference heat source.

Specifically, the gain correction table (Gi) of each element and the offset correction table (Oi) of each element are calculated as follows.
Gi = (Have−Lave) / (Hi−Li)
Oi ＝ Lave−Li × Gi
here,
Hi: Output of each element during high temperature reference heat source imaging
Li: Output of each element during low-temperature reference heat source imaging
Have: Average output of all elements during high temperature reference heat source imaging
Lave: Output average of all elements during low-temperature reference heat source imaging.

Further, output correction is performed as follows.
Ci = Ii x Gi + Oi
here
Ii: Output of each element during scene imaging
Ci: Correction output of each element during scene imaging.

  However, after the data for correcting the gain and offset variation is created, if the gain variation and offset variation change further, the correction data is again used using two types of reference heat sources. Need to be recreated. Here, since the change in gain variation with time is small, a gain correction table used for gain correction can be created once immediately after power-on and used for subsequent correction. Alternatively, it is also possible to store a gain correction table generated in advance in a ROM or the like and read and use it. That is, it is possible to update the correction data only with respect to the offset variation in consideration of the change over time. In this case, it is possible to update the offset correction table by imaging only one type of reference heat source using an existing gain correction table. Similarly, it is possible to correct only the offset by imaging a background with a substantially uniform temperature, so offset correction can also be performed by blurring the focal point of the optical system and imaging the background etc. as a substantially uniform temperature surface. Has been done.

However, in the above-described conventional method, the target imaging must be interrupted to update the offset correction table.
Patent Document 1 describes a sensitivity correction device for an infrared imaging device in consideration of the above problems. That is, an apparatus is disclosed that acquires offset correction data by thinning out and integrating a plurality of frame data obtained by scanning an infrared detector so as to make a round with respect to the visual field.

However, in the apparatus described in Patent Document 1, since the visual axis is largely moved by the directing means, the area that is imaged in common is very small, and the target cannot be continuously imaged greatly. That is, only a very limited target can be captured, such as capturing a minute object in a simple background such as the sky.
Japanese Patent Laid-Open No. 5-264357

  As described above, among the methods for correcting the output offset of the element according to the prior art, in the method using the reference heat source and the method of performing the defocusing of the optical system, the imaging of the target is temporarily suspended to update the correction table. Need arises. Further, in the method described in Patent Document 1 in which the visual axis is moved greatly by the directing means, the target cannot be imaged greatly. Furthermore, since the method disclosed in Patent Document 1 images a scene in a pseudo-uniform manner, it can be used only when the background is a simple background such as the sky. That is, in the case where an image of a scene with a complicated background is captured or the target is not very small relative to the size of the background portion, it is necessary to substantially stop the image capturing of the scene in order to update the offset correction value.

  The object of the present invention is to solve this problem. That is, an object of the present invention is to make it possible to update an offset correction table while capturing an image of an arbitrary target.

  The offset correction processing apparatus of the present invention is an image acquisition means, a directing means for moving the visual axis of the image acquisition means, a correction means for correcting the gain and offset of the output of each element included in the image acquisition means, After the image acquisition means acquires an image, before the next image is acquired, the directing means repeatedly moves the visual axis of the image acquisition means to acquire a plurality of images, thereby obtaining a plurality of different elements. Then, an update amount is obtained by using the output difference between the elements of the plurality of elements after correction by the correction means for the same imaging target, and is used in the correction means for at least one of the plurality of elements based on the update amount. Data updating means for updating offset correction data.

  In the offset correction processing apparatus, the offset correction data is updated using data when the same part of the same imaging target is imaged by a plurality of elements included in the image acquisition means. When the same part of the same imaging target is imaged, the output from any element should be the same if the offset and gain of the element are sufficiently corrected. If there is a difference between elements in the output when the same part of the same imaging target is imaged, variation in the output between the elements can be reduced if correction is made to reduce the difference. Therefore, the update amount of the offset correction data is obtained using the output difference between the elements, and the update is performed based on the obtained update amount. Further, by moving the visual axis of the image acquisition means by the directing means, the same part of the same imaging object can be imaged with different elements.

  Such processing can be performed while imaging an imaging target. Therefore, according to the present invention, it is possible to reduce the offset variation between elements without stopping the imaging of the imaging target.

In the conventional technique, in order to update the offset correction data of the element output, it is necessary to stop the imaging of the target scene having a complicated background. The method described in Patent Document 1 can be used only when the target (imaging target) is small and the background is simple. In effect, the imaging of the target scene is stopped to update the offset correction data. There was a need to do. According to the present invention, the offset correction table can be updated using the data being imaged. For this reason, it is possible to obtain a sufficiently corrected image without stopping the imaging of the target scene.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
<Outline of offset correction processing apparatus>
FIG. 1 is a diagram showing an outline of an offset correction processing apparatus. An example of the offset correction processing apparatus according to the present invention includes an image acquisition unit 1, a directing unit 2, a correction unit 3, a correction table updateable area calculation unit 4, and a correction table update unit 5. With these parts, the offset correction value is updated while imaging is continued by calculating the initial correction processing table, acquiring the image, moving the visual axis by the directing unit 2, determining the correction table updatable area, and updating the correction table. .

(1) Calculation of initial correction processing table When power is supplied to the offset correction processing device, an initial correction processing table is first created using two types of heat sources, a low-temperature reference heat source and a high-temperature reference heat source, as in the conventional method. . The created initial correction processing table is stored in the memory.

  During a period when the data included in the created initial correction table is an effective value, the output of each element after correction shows the same value when a uniform temperature surface is imaged. Even if the target scene is the target scene, while the data included in the correction table is valid, the corrected output should be the same between the elements capturing the same part of the target scene. is there. That is, when there is a difference in output between elements, the gain variation hardly changes with time as described above, and therefore, the difference in output can be regarded as a shift in the offset correction value. Therefore, if the offset correction table is changed so as to eliminate the difference in the output after correction that occurs when the same part of the target is imaged with different elements, the offset can be corrected without stopping the imaging. It becomes.

(2) Image acquisition and movement of visual axis by directing unit In order to obtain data obtained by imaging the same part of the target with different elements, the image acquiring unit 1 and the directing unit 2 are used.

  First, when a target is imaged by the image acquisition unit 1, the obtained target image data is sent to the correction unit 3. The correction unit 3 obtains an output in which the gain and offset of the output of each element are corrected using the data of the initial correction processing table stored in advance in the memory. The corrected data is output as a corrected image and simultaneously sent to the correction table updateable area calculation unit 4 and the correction table update unit 5.

  Next, the directing unit 2 moves the visual axis of the image acquisition unit 1. After the visual axis is moved, the target is imaged by the image acquisition unit 1, and similarly, the correction unit 3 performs processing to obtain a corrected image. The corrected image acquired after the visual axis movement is also sent to the correction table updateable area calculation unit 4 and the correction table update unit 5. In this specification, “visual axis” and “optical axis” are synonymous.

(3) Determination of Correction Table Updatable Area When a plurality of correction images are sent to the correction table updatable area calculation unit 4, “data obtained by imaging the same portion of the target scene with different elements” is obtained for the plurality of correction images obtained. "Is determined.

This determination is made by the correction table updateable area calculation unit 4. In the correction table updatable area calculation unit 4, a part of each corrected image is cut out as a comparison area. Here, if there is no movement between the imaging device and the target in addition to the movement of the visual axis of the image acquisition part, the obtained frame is shifted by the movement of the visual axis to capture the same part of the target scene. Become. Therefore, the cut-out position of the comparison region is determined for each frame based on the movement of the visual axis performed by the directing unit 2 when each image is acquired with reference to the visual axis when the image is acquired. That is, by changing the cut-out position from the frame in accordance with the movement of the visual axis, the images of the parts considered to be capturing the same part are compared. A correlation value between the cut out comparison areas is obtained, and if the correlation value is equal to or greater than a threshold value, it is determined that there is a possibility that the same portion has been imaged, and the comparison area is set as a correction table updateable area. In this specification, the “correction table updatable area” may be described as an “updatable area”.

(4) Correction Table Update The determination result obtained by the correction table updateable area calculation unit 4 is sent to the correction table update unit 5. The correction table update unit 5 compares the outputs from a plurality of elements that have captured the same portion with respect to the portion that can be updated based on the determination result of the correction table updateable region calculation unit 4. Here, as described above, since the variation in gain of the output of the element has a small change with time, it can be considered that the difference in output between the elements is a deviation of the offset value. The correction table update unit 5 updates the correction table based on the output differences from a plurality of elements that image the same portion so that the difference in the output of each element becomes small. Note that the difference between the value after the update of the correction table and the value before the update may be described as “offset correction table update amount”, “correction table update amount”, and “update amount”, which are synonymous.

  By repeating the above operations, it is possible to perform offset correction while continuing to capture the target scene. The present invention is mainly realized as an infrared imaging device, but is not limited to an infrared imaging device. It can be used for any device that needs to update the offset correction value of the output of the element regularly or irregularly. This is particularly effective when the target scene is imaged in an environment where the offset variation of the element changes drastically because it is necessary to perform imaging in a state exposed to the outside air. It is also effective when it is necessary to continue capturing the target scene for a certain period, such as when capturing moving images.

An example of a specific embodiment of the configuration of the offset correction processing apparatus according to the present invention is shown in FIG.
The image acquisition unit 1 includes a lens unit 10, a photoelectric conversion unit 15, an amplification unit 16, and an A / D conversion unit 17. In the embodiment illustrated in FIG. 2, the directing unit 2 includes the micro scanner unit 11, and the correction processing unit 23 corresponds to the correction unit 3. The correction table updatable area calculation unit 4 includes a comparison area extraction unit 24 and a phase correlation processing unit 29. The offset correction table update unit 30 corresponds to the correction table update unit 5.

The overall operation of the offset correction processing apparatus will be described with reference to FIG. Here, a case where the offset correction table is updated based on four frames will be described.
(1) Immediately after turning on the power, the low temperature reference hot plate portion 13 and the high temperature reference hot plate portion 14 are used to image the reference hot plate in order to create an initial correction table. The optical path switching unit 12 switches the optical path to the low temperature reference hot plate 13 and converts the infrared light from the low temperature reference hot plate into an electrical signal by the photoelectric conversion unit 15. After the converted signal is amplified by the amplification unit 16, the A / D conversion unit 17 converts it into a digital image, and the initial correction table calculation unit 18 stores the obtained image in the low-temperature image storage memory 19. After the optical path switching unit 12 switches the optical path to the high temperature reference hot plate unit 14, similarly, the image acquisition unit 1 acquires an image of the high temperature reference hot plate and stores it in the high temperature image storage memory 20.

(2) The initial correction table calculation unit 18 calculates gain and offset correction tables using the images stored in the low-temperature image storage memory 19 and the high-temperature image storage memory 20. The obtained gain correction table and offset correction table are stored in the gain storage memory 21 and the offset storage memory 22 by the initial correction table calculation unit 18, respectively.

  (3) Next, the target scene is imaged. First, the optical path switching unit 12 switches the optical path to the micro scanner unit 11. At the time of imaging, the photoelectric conversion unit 15 converts the infrared light of the scene incident through the lens unit 10 into an electrical signal. After the amplification unit 16 amplifies the converted signal, the A / D conversion unit 17 converts the signal into a digital image, and the correction processing unit 23 uses the correction table stored in the gain storage memory 21 and the offset storage memory 22 to gain. Variations and offset variations are corrected, and a corrected image is output.

  (4) While capturing the target scene, the image acquisition unit 1 acquires a frame for updating the offset correction table. The corrected image is also sent to the comparison area extraction unit 24 and stored in one of the frame (1) storage memory 25 to the frame (4) storage memory 28 according to the direction of the optical axis adjusted by the micro scanner unit 11 at the time of imaging. The The micro scanner unit 11 scans the optical axis in a range corresponding to four pixels in order to update the offset correction table using four frames. Since the scanning of the optical axis is performed in a stepwise manner during a non-imaging time during imaging of the frame, the optical axis during one frame imaging period is constant.

  (5) After the frames (1) to (4) are stored in the memory, the offset correction table updatable area is calculated. The comparison area extracting unit 24 sets a correlation processing target point for each frame, and sends a small region such as 15 pixels × 15 pixels around the correlation processing target point to the phase correlation processing unit 29. The phase correlation processing unit 29 confirms by the phase correlation that the correlation value is equal to or greater than the threshold and that the peak is at the center. The comparison area is extracted by sequentially changing the attention point for correlation processing for all pixels. An area where the correlation value is equal to or greater than the threshold and the peak is in the center is set as the offset correction table updateable area.

  (6) The offset correction table is updated in the offset correction table updatable area. The offset correction table updating unit 30 performs offset correction on the pixels that have been made the offset correction table updatable area by the phase correlation processing unit 29 based on the pixel values in the frame (4) storage memory 28 from the frame (1) storage memory 25. The update amount of the table is calculated. The offset correction table update unit 30 adds the calculated update amount of the offset correction table to the average value calculation memory 31. When the addition of the update amount is completed for all the pixels in the updatable area using the four frames, the offset correction table update unit 30 adds the update amount to the correction amount of the pixel stored in the offset storage memory 22. As a result, the offset correction table is updated.

(7) The offset correction table is updated continuously throughout the imaging period.
<Initial correction table calculation>
Hereinafter, the operation of the offset correction processing apparatus will be described in detail.

  FIG. 3 shows a flowchart of initial table calculation for correcting the gain variation and offset variation of the element output. In the offset correction processing apparatus, as described above, the initial correction table calculation unit 18 calculates the initial correction table by a conventional method after the power is turned on.

First, the optical path switching unit 12 switches the optical path to the low temperature reference hot plate part 13 in order to obtain outputs from each element when the low temperature reference hot plate is imaged in step S1. Next, in step S2, the image acquisition unit 1 images the low temperature reference hot plate and acquires an image. The acquired image is stored in the low-temperature image storage memory 19 for storing the image data (low-temperature image) of the low-temperature reference hot plate by the initial correction table calculation unit 18. When the imaging of the low temperature reference hot plate is completed, the optical path switching unit 12 switches the optical path to the high temperature reference hot plate unit 14 and similarly images the high temperature reference hot plate (steps S3 and S4). After imaging, the image data (high temperature image) of the high temperature reference heat source is also stored in the high temperature image storage memory 20.

  High-temperature reference hot plate imaging from the output (Hi) of each element during high-temperature reference hot plate imaging and the output (Li) of each element during low-temperature reference hot plate imaging obtained by imaging two types of reference hot plates The average output (Have) of all elements at the time and the average output (Lave) of all elements at the time of low-temperature reference hot plate imaging are obtained (steps S5 and S6). Next, the initial correction table calculation unit 18 calculates the gain correction table (Gi) of each element and the offset correction table (Oi) of each element as follows.

Gi = (Have−Lave) / (Hi−Li)
Oi = Lave−Li × Gi
When Gi and Oi are calculated for all elements, the calculation of the initial correction table ends (step S9).

  The initial correction table calculation unit 18 stores the calculated gain correction table in the gain storage memory 21 and stores the offset correction table of each element in the offset storage memory 22. The calculated output average of all elements during high-temperature reference hot plate imaging, output average of all elements during low-temperature reference hot plate imaging, and the like are also stored in the memory. When the image acquisition unit 1 acquires an image, these values are read, and the correction unit 3 generates a corrected image.

  Here, the imaging of the high-temperature reference hot plate is performed after the imaging of the low-temperature reference hot plate, but the order of imaging may be changed. That is, the order of steps S1 and S2 and steps S3 and S4 may be switched. The order of steps S5 and S6 for calculating the output average of the elements at the time of imaging each reference hot plate is also arbitrary. Also, when storing the output average of all elements during high-temperature reference hot plate imaging, the output average of all elements during low-temperature reference heat source imaging, the gain correction table, and the offset correction table of each element are all the same. It may be stored in a memory area, or may be stored in a different memory area for each type of data. That is, the storage method in the memory area is arbitrary.

<Acquire corrected image>
Next, the operation of the offset correction processing apparatus when capturing an image of the target scene will be described. FIG. 4 shows a flowchart of the operation when imaging is performed after the initial correction table is calculated.

  First, each element of the photoelectric conversion unit 15 converts the infrared light that has entered the image acquisition unit 1 through the lens unit 10 of the apparatus into an electrical signal, and the amplification unit 16 amplifies (steps S11 and S12). The amplified signal is AD converted to a digital signal and sent to the correction unit 3 (step S13).

  When the gain variation and offset variation of the output (Ii) of each element at the time of scene imaging obtained are corrected according to the following equations, the corrected output (Ci) of each element at the time of scene imaging is obtained (step S14, S15).

Ci = Ii x Gi + Oi
The image acquired in step S16 is output as a corrected image. Also output to the correction table updateable area calculation unit 4 and the correction table update unit 5. Further, the corrected image is stored in any of the frame (1) storage memory 25 to the frame (4) storage memory 28 for use in updating the offset correction table as necessary.

<Visual axis movement by directional unit and imaging after visual axis movement>
After the image acquisition unit 1 images the target scene, the directing unit 2 moves the visual axis of the image acquisition unit 1 in any direction, up, down, left, and right. FIG. 5 shows an example of visual axis movement performed by the directing unit 2 when offset correction is performed using four pixels adjacent to each other. FIG. 6 shows a flowchart of the visual axis movement and subsequent imaging.

  In FIG.6 S21, the visual axis of the image acquisition part 1 is match | combined with the part of (1) of FIG. Next, the image acquisition unit 1 acquires an image of the target scene, acquires an image when the visual axis of the image acquisition unit 1 is at (1), outputs a corrected image by the correction unit 3, and a frame (1) storage memory 25. The corrected image is stored in (Step S22).

  When the acquisition of the image of the target scene is completed, the directing unit 2 moves the visual axis of the image acquisition unit 1. In the example of FIG. 5, target scene imaging after moving the visual axis is performed when the visual axis is moved to the right by a distance corresponding to one pixel and the visual axis is aligned with (2) (step S23 in FIG. 6). , S24).

  After that, similarly, when the directing unit 2 aligns the visual axis by one pixel below the visual axis (FIG. 5 (3)) (steps S25 and S26 in FIG. 6), the left by one pixel (FIG. 5 (4)). The same processing is performed when the visual axis is aligned with ()) (steps S27 and S28 in FIG. 6).

  Note that the directing unit 2 moves the visual axis during a non-imaging time period after capturing an image of a certain frame and before capturing an image of the next frame. Therefore, the visual axis during the period of capturing one frame image is kept constant.

  As described above, when the directional unit 2 moves the visual axis and the image acquisition unit 1 captures an image after moving the visual axis, the area in the image acquisition unit 1 is smaller by one pixel vertically and horizontally than the original number of pixels. In other words, the target scene is imaged in common with other images. FIG. 9 shows the range of the target scene that each frame captures with reference to the captured target. Here, it should be noted that FIG. 9 is not a diagram illustrating where an element that captures the same portion of the same target exists.

For example, it is assumed that a portion 40 of the target is imaged at a pixel 41 having a frame (1). Next, before the image acquisition unit 1 captures the frame (2), the directing unit 2 moves the visual axis to the right by one pixel as shown in FIG. At this time, in frame (2), 40 is imaged on the left pixel 42 by one pixel. That is, when the frame (2) is imaged, the image of 40 is captured by the element existing on the left side of the element that captured 40 in the frame (1). Similarly, when the image acquisition unit 1 captures the frame (3), the visual axis is moved down by one pixel, and the visual axis has moved to the lower right when considered from the time of imaging of the frame (1). Become. Therefore, 40 is imaged by the pixel 43 located at a position shifted to the upper left of the frame (1) in the frame (3). That is, at the time of imaging the frame (3), the element existing at the upper left of the element that images 40 in the frame (1) captures 40. When the frame (4) is imaged, the visual axis is shifted by one pixel below that when the frame (1) is imaged. Therefore, in the frame (4), 40 is imaged by the pixel 44 that exists one pixel above the frame (1). Has been. That is, 40 is imaged by an element that exists on the element that imaged 40 in frame (1). In summary, if the visual axis is moved to the right, bottom, and left one pixel at a time as shown in FIG.
In the frame (2), the visual axis moves to the right, and the element that images the same part moves to the left.
In the frame (3), the visual axis moves to the lower right, the element that images the same part moves to the upper left,
In the frame (4), the visual axis moves down, and the element that images the same part moves up.

<Calculation of correction table updatable area>
The area common to the frames shown in FIG. 9 can be treated as an “effective area” 50 that may update the correction table. That is, the effective area 50 corresponds to a common area of the frames (1) to (4).

  The correction table updateable area calculation unit 4 calculates a correction table updateable area that is an area in which “data obtained by imaging the same part of the target with different elements” is obtained between different frames in the effective area 50. To do. When the imaging device and the target are both stationary, the entire effective area 50 can be handled as an area having data obtained by imaging the same portion of the target with different elements.

  On the other hand, if any one of the imaging device and the target moves, a portion that cannot be handled as “data obtained by imaging the same portion of the target with different elements” occurs even in the effective area 50. Since there is a time difference between imaging, imaging time differs between frames. Therefore, when the target is moving, different data is imaged at the pixels on the movement locus of the target even if the pixels are in the effective area. Even if the target does not move, if the image pickup device moves greatly, the entire effective area moves uniformly, so that different parts are imaged in all pixels. In principle, the correction table cannot be updated for such a pixel, and it is necessary to extract a portion that is stationary between a plurality of frames.

  The correction table updateable area calculation unit 4 cuts out a part of the effective area 50 of each frame as a “comparison area” 51 in order to obtain a part imaged in a state where the imaging apparatus and the target are stationary. First, in order to cut out the comparison region 51, “correlation processing attention point” 52 is set. A region around the correlation processing attention point 52 is cut out as a comparison region 51. The size of the comparison area 51 is an arbitrary size, but it needs to be an appropriate size for judging the correlation between the comparison areas. For example, the size may be 3 vertical pixels × 3 horizontal pixels, and in the case of determining the correlation using the phase correlation, from the size of 8 vertical pixels × 8 horizontal pixels, 16 vertical pixels × 16 horizontal pixels. It may be cut out to a certain size. The number of pixels in the vertical and horizontal directions may be different.

  The coordinates of the image (frame) and the coordinates of the comparison area 51 to be cut out are as shown in FIGS. When M pixels are included in the horizontal direction and N pixels are included in the vertical direction, the coordinates of the image are as shown in FIG. FIG. 11 shows a case where the comparison area 51 including 2a + 1 pixels horizontally and 2b + 1 pixels vertically is cut out. Since each frame is imaged after the visual axis is shifted by the directing unit 2, a portion shifted vertically and horizontally is imaged. Therefore, the cut-out coordinates of the comparison area 51 are shifted so that the same place can be cut out when the imaging device and the target are stationary. The above-described frames (1) to (4) are taken as an example. As shown in FIG. 12, when the correlation processing target point 52a of the frame (1) is set to (x, y), the sighting unit 2 performs the visual axis until the imaging of the frame (2) after the imaging of the frame (1). Is moved to the right by one pixel, the correlation processing attention point 52b for the frame (2) is (x-1, y). Similarly, the frame (3) is set to (x-1, y-1), and the frame (4) is set to (x, y-1).

  After the comparison area 51 is cut out, the correction table updatable area calculation unit 4 calculates a correlation value between the comparison areas. When the correlation value is equal to or greater than the threshold value, it is determined that the target imaged in the comparison area 51 (51a to 51d) is substantially the same part of the target scene. The correction table updatable area calculation unit 4 also determines whether the imaging apparatus and the target are stationary from the correlation between the images in the comparison areas, and determines whether the correction table can be updated based on the determination. As a correlation value calculation method, for example, there are various methods such as normalized correlation, luminance correlation, and phase correlation, and any of them may be used.

  When the correlation value is calculated using phase correlation, the imaging device and the target are stationary when an image including a certain comparison region 51 is captured and when an image including the comparison region 51 to be compared is captured. There is an advantage that it is easy to judge whether or not. As a calculation result when the phase correlation is used, for example, a correlation image as shown in FIG. 13 is obtained. When the imaging device and the target are stationary, the central pixel ((x ′, y ′) = (0, 0)) as shown in FIG. 13 has a peak value, and this peak value becomes a correlation value. Here, as the threshold value of the correlation value, a numerical value determined in advance is used before imaging is performed, typically before the offset processing apparatus is shipped. In determining the threshold value, a plurality of images are captured for an arbitrary target scene that is not a uniform temperature surface by an offset processing device. The correlation value of the comparison area 51 is actually calculated for the image obtained by imaging at this time, and the threshold value is determined based on the fluctuation of the peak size due to noise.

  FIG. 7 shows an example of a flowchart in which the correction table updateable area calculation unit 4 determines whether the comparison area 51 is a correction table updateable area for the above-described frames (1) to (4). First, correlation processing attention points 52 (52a to 52d) are set (step S31), and comparison areas 51a to 51d are cut out from each frame (step S32). Here, the “correlation processing attention point” 52 refers to a pixel serving as a reference for setting a range to be cut out when the comparison area 51 is cut out. Next, a correlation value between the frame (1) and the comparison area 51 of another frame is calculated, and if the correlation value between the frame (1) and all other frames is equal to or greater than a threshold value, the comparison is performed. In area 51, it is determined that the correction table can be updated (steps S33 to S39). On the other hand, if any one of the frames having a correlation value with the frame (1) is equal to or less than the threshold value, it is determined that the comparison area 51 is an area where the correction table cannot be updated (step S40). ).

  The correction table updateable area calculation unit 4 includes an attention point update availability table that records whether or not the correction table can be updated for the comparison area 51 in advance. The correction table updatable area calculation unit 4 “updates” the attention area update availability table for the correlation processing attention point of the comparison area 51 for the comparison area 51 determined to be able to update the correction table. "Yes" is set (step S39). On the other hand, for the region where it is determined that the correction table cannot be updated, “not updateable” is set in the attention point update availability table (step S40). The attention point update availability table can take any format that can record the coordinates of the attention point and the availability of the update. As an example, it is conceivable to record a value of 1 for “updatable” and 0 for “not updatable” in association with the coordinates of the point of interest.

Until the above processing is performed for all the pixels in the effective region 50, the coordinates of the correlation processing target point 52 are changed (step S41).
<Update correction table>
The correction table update unit 5 compares outputs from a plurality of elements that have captured the same part of the target scene with respect to the part that can be updated by the determination result of the correction table updateable area calculation unit 4. Hereinafter, the case where the offset correction table is updated using four adjacent pixels using the frames (1) to (4) will be described in detail with reference to the flowchart (FIG. 8) of the operation of the correction table update unit 5. In the following. One set of four frames (1) to (4) may be used, or a plurality of sets may be used.

{Calculation of 4 pixel relative relationship}
First, a 4-pixel relative relationship calculation target point is set, and it is checked whether the position is set in the correction table updatable area. Here, the “4-pixel relative relationship” refers to a relative shift amount of values at four pixels obtained by imaging the same portion of the target scene. Specifically, it is the difference between the value of another pixel and the value of the reference pixel when the value at a certain pixel is the reference. In other words, if the “4-pixel relative relationship” is defined using the elements included in the image acquisition unit 1, the relationship between the elements generated in the element group that has the relationship of capturing the same part of the target scene (imaging target). It can be said that this is a difference in output (output difference between elements). Further, the “four-pixel relative relationship calculation attention point” is a point for calculating the relative relationship between the pixels in the frames (1) to (4), and is a point to be set on the correction table updatable area. . The attention point for calculating the 4-pixel relative relationship can be set at an arbitrary position in the correction table updatable area. In the following description, when a point such as a four-pixel relative relationship calculation attention point or attention point is set in the correction table updateable area, the pixel of each frame corresponding to the pixel that is the set point is expressed as “ May be described as “a pixel of interest”.

  When the pixel for which the attention point for calculating the four-pixel relative relationship is set does not exist in the correction table updateable area, the attention point for calculating the four-pixel relative relationship is reset until the correction table updateable area is reached (steps S51 and S52). ). When the coordinates of the pixels existing in the correction table updateable area are found, the difference between the pixel values of each frame corresponding to the set four-pixel relative relationship calculation target point is calculated (step S53).

  For example, a case where the four-pixel relative relationship is calculated with reference to the upper left pixel of the four pixels is shown. Consider the case where “attention point for calculating relative relationship of four pixels” is set at (x, y) = (A, B) in the correction table updatable area. Here, as an example, a case where the relative relationship is calculated with reference to the pixel at the upper left among the four pixels is shown. That is, among the four frames to be compared, a frame obtained by imaging the same portion of the same target with the element at the upper left is used as a reference, so that the pixel corresponding to (A, B) of frame (3) is used as a reference. An example of performing an operation is shown. When the attention point for calculating the four-pixel relative relationship is set, the offset correction table updating unit 30 recognizes the output of the element at the coordinate of the attention point for calculating the four-pixel relative relationship in the reference frame (3) from the corrected image data. To do. Next, the movement of the visual axis by the directional unit 2 at the time of frame (3) image capturing and other frame image capturing is performed at the position of the element that outputs the pixel data of the “four-pixel relative relationship calculation target point” in another frame. Get from quantity.

Here, it is assumed that the output values of the elements around (A, B) are as follows.
Output value of element (A, B) corresponding to target pixel of frame (3): A1
Output value of element (A + 1, B) corresponding to the target pixel of frame (4): A2
Output value of element (A + 1, B + 1) corresponding to the target pixel of frame (1): A3
Output value of element (A, B + 1) corresponding to the target pixel of frame (2): A4
Since these values are the result of outputting the same part in the same target scene with four different elements, they should originally be the same output value. It can be considered that the value is changed like A1 or A4 due to the change with time of offset variation. From the above values, the 4-pixel relative relationship is calculated based on the element (A, B) as follows.

Difference between output value of (A, B) and output value of (A + 1, B): A1-A2
Difference between output value of (A, B) and output value of (A + 1, B + 1): A1-A3
Difference between output value of (A, B) and output value of (A, B + 1): A1-A4
After calculating the relative relationship between the four pixels, it is checked whether the four-pixel relative relationship has been calculated for all the pixels in the updatable area (step S54). If all the pixels have not ended, the same calculation is performed by moving the attention point for calculating the relative relationship of the four pixels. For example, the attention point for calculating the four-pixel relative relationship is moved to the adjacent point on the right side, and the pixel in the updatable area captured by the element at (x, y) = (A + 1, B) in the frame (3) is “ Consider the case where “attention point for calculating relative relationship of four pixels” is set. In this case, as described above, the target pixel is (A + 2, B) in frame (4), (A + 2, B + 1) in frame (1), and (A + 1, B + 1) in frame (2). Here, it is assumed that the output value of the element is the following value for each pixel of interest in the frames (1) to (4).

Output value of element (A + 1, B) corresponding to target pixel of frame (3): A5
Output value of element (A + 2, B) corresponding to target pixel of frame (4): A6
Output value of element (A + 2, B + 1) corresponding to the target pixel of frame (1): A7
Output value of element (A + 1, B + 1) corresponding to the target pixel of frame (2): A8
Since the values of A5 to A8 are also the results of outputting the same part in the same target scene with four different elements, they should be the same output value. From the above values, the 4-pixel relative relationship is calculated based on the element (A + 1, B) as follows.

Difference between the output value of (A + 1, B) and the output value of (A + 2, B): A5-A6
Difference between the output value of (A + 1, B) and the output value of (A + 2, B + 1): A5-A7
Difference between the output value of (A + 1, B) and the output value of (A + 1, B + 1): A5-A8
As is clear from the above example, the values of the pixels output from the (A + 1, B) element and the (A + 1, B + 1) element are calculated by calculating the relative relationship at the first attention point for calculating the 4-pixel relative relationship. This is used for calculating the relative relationship at the attention point for calculating the 4-pixel relative relationship after the movement. However, the output from (A + 1, B) uses the value of the pixel in frame (4) when the first four-pixel relative relationship calculation point is set, but after the movement of the four-pixel relative relationship calculation point Uses the pixel values in frame (3). Therefore, even if the same pixel is imaged, the part (A + 1, B) imaged in frame (4) and the part (A + 1, B) imaged in frame (3) are different parts of the target scene. It should be noted that there are. For the sake of explanation, a part of the imaging target captured by the pixel output by (A, B) in the frame (3) is referred to as a “first part”. At the time of calculating the first four-pixel relative relationship, the output difference between elements when the first part is imaged is obtained for the element group that images the first part. On the other hand, when the attention point for calculating the 4-pixel relative relationship is moved as in the above example, the “second portion” of the imaging target captured by the pixel (A + 1, B) output in the frame (3) is displayed. For the imaged element group, the output difference between elements when the second portion is imaged is obtained. Therefore, the calculation of the four-pixel relative relationship means that, for each part to be imaged, the element group that images the part is recognized, and the output difference between elements in each element included in the element group is calculated. Equivalent to.

  In the above-described calculation example of the four-pixel relative relationship, for the sake of explanation, the calculation using the upper left as the reference (reference pixel) is shown, but the reference pixel may be placed anywhere. That is, the position of the reference pixel can be set to any element in the element group that images the same part of the same target. In this specification, an element that images a reference pixel may be referred to as a reference element.

  Furthermore, the calculation as 4 pixels is merely for the purpose of explanation, and the number of pixels used when obtaining the relative relationship is arbitrary, and the number varies depending on the number of frames used. That is, the “relative relationship” is an output difference between each element of an element group composed of an arbitrary number of elements obtained by imaging the same portion of the target scene.

{Origin setting and update amount calculation}
By simply calculating the 4-pixel relative relationship, an inter-element output difference for the first element group that images the first part is obtained, and an inter-element output difference for the second element group that images the second part. In other words, the output difference between the elements that image the same portion can only be obtained. However, in the correction of the offset correction table, the same object was imaged while eliminating the output difference between the elements obtained as a 4-pixel relative relationship for all of the plurality of element groups such as the first element group and the second element group. Sometimes it is necessary to prevent an output difference between element groups. Therefore, when the four-pixel relative relationship is calculated for all the pixels in the updatable area, the offset correction table update unit 30 sets the origin (step S55). The “origin” is a reference point for calculating the update amount. This “update amount” is the update amount of the offset correction table. That is, the “4-pixel relative relationship” is the difference in output value between only four elements, whereas the “update amount” is located at the origin and the output value of all elements existing in the updatable area. It is the difference in the output value of the element. Accordingly, since the origin serves as a reference for the update amount of all elements, the update amount of the element at the origin is always zero.

  When the origin is set, the renewal amount of the element for which the relative relationship is obtained with respect to the origin is calculated using the relative relationship with the origin as a reference. In the previous example, since the four-pixel relative relationship is calculated with reference to the upper left, the update amount is calculated for the four elements located at the right, lower right, and lower of the origin (step S56).

  As described above, it is assumed that the value of A1 is calculated as the output of the element (A, B) when the same part of the same target scene is imaged, and A2 is obtained as the output of the element (A + 1, B). . Since the output from the (A, B) element is A1 and the output from the (A + 1, B) element is A2, the offset value of the (A, B) element is essentially the same value. Is used as a reference, the offset value of the element (A + 1, B) is shifted by A1-A2, which is the value of the relative relationship calculated earlier. Therefore, the offset value of the element (A + 1, B) should be changed so that the deviation can be eliminated. Since other elements can be considered in the same manner, the following update amount is required.

(A + 1, B) offset correction table update amount: A1-A2
(A + 1, B + 1) offset correction table update amount: A1-A3
(A, B + 1) offset correction table update amount: A1-A4
As is clear in this example, when the reference point when the relative relationship is obtained matches the origin, the update amount for each element for which the relative relationship has been calculated always matches the value of the relative relationship.

When the offset correction table update amounts of the four elements around the origin are obtained, attention points are set in order to perform processing on the other elements. The attention point is an update amount already calculated point, and is set to a point adjacent to a pixel in the updatable area where the update amount is not calculated (step S57). Next, the update amount of the target point offset correction table is calculated using the update amount of the already calculated point and the relative relationship obtained using the target point as the reference pixel (step S58). Specifically, the update amount for the elements around the attention point is obtained as the update amount = the update amount of the attention point + the relative relationship between the attention point and the pixel for which the update amount is to be calculated.

  As is clear from this equation, the update amount of the point of interest is required to calculate the update amount. That is, attention points can be set only at points where the update amount has already been calculated. For example, when the point of interest is moved from (A, B), a new point of interest must be set at the point where the update amount is calculated with (A, B) as the point of interest. Accordingly, the adjacent points on the right, lower right, and lower of (A, B) are positions where the attention point can be moved.

  As an example, a case where a point of interest is set at (A + 1, B), which is an adjacent point of the origin (A, B) will be described. In this case, the point of interest exists at a position moved one pixel to the right with respect to the origin. As described above, relative to (A + 2, B), (A + 2, B + 1), (A + 1, B + 1) with (A + 1, B) as a reference, the relative relationships are A5-A6, A5-A7, A5-A8, respectively. It has been calculated. Further, the update amount of (A + 1, B) which is the attention point is A1-A2.

At this time, unlike the case where the update amount is obtained by placing the attention point at the origin, 4 (A + 1, B), (A + 1, B + 1), (A, B + 1) are already set with the (A, B) element as a reference. The amount of update between pixels is required. Therefore, (A + 1, B) and (A + 1, B + 1) out of (A + 1, B), (A + 2, B), (A + 2, B + 1), and (A + 1, B + 1), which are targets of calculation of the update amount this time. The amount of renewal has already been calculated for the element (). Therefore, (A
The update amount may be calculated for the two elements +2, B) and (A + 2, B + 1). The update amounts for the uncalculated points (A + 2, B) and (A + 2, B + 1) are as follows.

Update amount of (A + 2, B): (A1-A2) + (A5-A6)
Update amount of (A + 2, B + 1): (A1-A2) + (A5-A7)
While eliminating the output difference between the elements obtained as a four-pixel relative relationship for all of the plurality of element groups such as the first element group and the second element group by the above processing, It is possible to prevent an output difference between them. For example, let (A + 1, B) be the first element group, and (A + 1, B) be the elements for which the output difference between the elements is obtained with reference to (A + 1, B). The element (A + 1, B) is used as the first reference element. That is, the first element group is an element that has imaged the pixels used for calculating the four-pixel relative relationship using the pixel captured by the element (A + 1, B) as the reference pixel. Different from the first part of the imaging target being imaged by the first element group, the element group imaging the second part is the second element group, and the element group imaging the third part is the third element A group. Here, as an example, (A + 2, B) is a second reference element, and (A + 2, B + 1) is a third reference element. Further, the element obtained by calculating the output difference between the elements using the second reference element (A + 2, B) and the second reference element are set as the second element group. Similarly, the third element group includes an element for which an output difference between elements is obtained using (A + 2, B + 1) and a third reference element. Just calculating the 4-pixel relative relationship,
First element group: output difference between elements based on the first reference element (A + 1, B) Second element group: output difference between elements based on the second reference element (A + 2, B) Element group: a state in which only the output between elements based on the third reference element (A + 2, B + 1) is known.

  Therefore, the origin is set in step S55, and the update amount for the first reference element is obtained as described above with reference to the value of the origin (corresponding to step S56), and the update amount of the first reference element is used, The update amount was calculated for (A + 2, B) and (A + 2, B + 1) included in the first element group (steps S57 and S58). As described above, (A + 2, B) and (A + 2, B + 1) correspond to the second reference element and the third reference element, respectively. That is, the update amounts of the second reference element and the third reference element are obtained by obtaining the update amount for each element of the first element group. Since the update amount of the second reference element is known, the update amount of the element existing in the second element group is obtained by using the update amount of the second reference element and the output difference between the elements in the second element group. It is done. Further, the update amount for the second element group obtained by this processing is based on the origin, as is the update amount for the first element group. Therefore, after updating the offset correction table according to the update amount, even if the same part of the imaging target is imaged with the first element group and the second element group, the same output is provided as long as the offset does not deviate further. A value is obtained. The same applies to the third element group.

  From another viewpoint, the processes from step S55 to step S59 include the origin element (A, B), the first reference element (A + 1, B), the second reference element (A + 2, B), and The third reference element (A + 2, B + 1) calculates the output difference between the reference elements with reference to the output of the element at the origin when the same part to be imaged is imaged. I can say that.

  Thus, by calculating the offset correction table update amount of the uncalculated points based on the offset correction table update amount of the already calculated points, the relative variation of the entire screen can be corrected.

{Update offset correction table}
The same processing is performed by changing the attention point until there are no uncalculated points in the updatable area that are adjacent to the already calculated points (step S59). When there is no uncalculated updatable point adjacent to the point for which the update amount has been calculated, the offset correction table update unit 30 stores the update amount calculated for each element in the update amount average value calculation memory 31 (Step S1). S60). The specified number of times in step S61 corresponds to the number of sets of frames to be used. Therefore, the update amount is calculated for each element as many times as the number of sets to be used, and the obtained update amount is stored in the update amount average value calculation memory 31 (step S60). When the update amount calculation is completed for all sets, the average value of the update amounts is obtained and added to the offset correction table, and the update of the offset correction table is completed (steps S62 to S64).

For example, it is assumed that the offset correction table values for each element during frame imaging are as follows.
Offset correction table (A, B): O1
Offset correction table (A + 1, B): O2
Offset correction table (A + 1, B + 1): O3
Offset correction table (A, B + 1): O4
Offset correction table (A + 2, B): O5
Offset correction table (A + 2, B + 1): O6
The average value of the update amount for each calculated element is added to the value of the offset correction table. When the number of frames used is only one set shown in the specific example above, the average value of the update amounts is the calculated update amount itself. In this case, the following values are obtained as the updated offset correction table.

(A, B): O1
(A + 1, B): O2 + A1-A2
(A + 1, B + 1): O3 + A1-A3
(A, B + 1): O4 + A1-A4
(A + 2, B): O5 + (A1-A2) + (A5-A6)
(A + 2, B + 1): O6 + (A1-A2) + (A5-A7)
If the offset correction table after update is used, the offset shift between elements has been eliminated, so that the output when the same part of the target scene is imaged should be the same.

  Use of a plurality of sets has an advantage that the influence of noise generated in each frame is reduced. However, when the number of frames that can be used for offset correction table update is small, or when there is a limit to the memory area that can be used for arithmetic processing, it is possible to perform processing with only one set.

{Relationship between origin setting position and offset correction area}
In the method described above, the update amount is calculated by moving the attention point to the right side from the origin with the upper left as the reference in the calculation of the 4-pixel relative relationship. However, when the upper left is used as a reference in the calculation of the four-pixel relative relationship, the update amount is calculated not only for the right of the reference element but also for the lower and lower right elements. Accordingly, the attention point can be set not only on the right side but also on the lower right side or the lower side of the origin.

  As is apparent from the above description, in principle, in the calculation of the four-pixel relative relationship, when the upper left is used as a reference, there are already calculated points only at the right, lower right, and lower of the target point. That is, in principle, the update amount cannot be calculated for pixels on the left side or above the origin. Therefore, when the upper left is used as a reference in the calculation of the four-pixel relative relationship, the offset correction table can be updated for a wider range by selecting the upper left of the updatable area as much as possible.

However, if you can use a computer that does not hinder other processing even if you simply calculate the relative relationship of four pixels as a relative relationship based on another pixel, you can set where the origin is. It will be good. This will be specifically described using the above example. The relative relationship when the output value of (A, B) is used as a reference is (A + 1, B): A1-A2
(A + 1, B + 1): A1-A3
(A, B + 1): A1-A4
Met. Here, by subtracting A1-A3 which is the value of the relative relationship of (A + 1, B + 1) from the value of the relative relationship of each pixel, the relative relationship based on the output value of (A + 1, B + 1) can be calculated. it can.

(A, B): A3-A1
(A + 1, B): A3-A2
(A, B + 1): A3-A4
In such a case, even if the origin is set at (A + 1, B + 1), the relative value calculated based on (A, B) can be converted to a value based on (A + 1, B + 1). Also, the update amount can be calculated for the upper, upper left, and left elements. Similarly, the update amount can be calculated by converting the value of the 4-pixel relative relationship into a value based on any update amount already calculated point.

  In any method, the update amount is calculated for each pixel existing in one continuous updatable region because the relative offset correction value shift is calculated as the update amount with reference to one origin. . Therefore, it is desirable not to set the origin in an area that is surrounded by a non-updatable area even if it is an updatable area.

<Other embodiments>
The present invention is not limited to the above-described embodiment, and can be variously modified. Some examples are described below.

{Processing when imaging device and target move}
In the description so far described, it is assumed that only the area where the imaging device and the target are stationary is the offset correction table updateable area. However, there are cases where the offset correction table can be updated even when the imaging apparatus or the target moves. The correction table updatable area calculation unit 4 sets the comparison area 51 and checks the correlation of the comparison area 51 between different frames. If all the comparison areas are translated, the offset correction table updatable area There are cases where you can. In such a case, if the position of the pixel to be compared by the correction table update unit 5 is shifted by the amount of movement, it can be handled as an element that images the same portion of the target scene.

  In order to obtain the amount of movement, for example, a phase correlation method can be used. If the correlation of the comparison area 51 is obtained by the phase correlation method when all of the comparison areas are translated, a correlation image in which the correlation value exceeds the threshold value but the peak position is not at the center as shown in FIG. Is obtained. When the correlation value is obtained by the phase correlation method, it is the movement amount when the shift of the peak position from the center is moved in parallel. When the movement amount is obtained, when the correction table update unit 5 calculates the four-pixel relative relationship, the movement amount is used to calculate the relative relationship with the pixel at the position moved by the shift amount (step in FIG. 8). Equivalent to the processing in S53).

{Modification of visual axis movement distance}
In the above-described embodiment, the case where the visual axis is moved by the four pixels adjacent to each other in the portion of the visual axis movement by the directing unit 2 and the imaging after the visual axis movement is described only for easy understanding. Absent. That is, when the visual axis is moved and imaged in order to acquire offset correction data, the range in which the visual axis is moved does not have to be four pixels adjacent to each other. Further, the movement and imaging of the visual axis may be repeated over many pixels, and it is conceivable to perform correction based on the visual axis movement of two or three pixels. Further, the distance between the visual axis before the visual axis is moved by the directing unit 2 and the visual axis after the movement does not have to be one pixel, and the visual axis is viewed at a position several pixels or more away. Even if the axis is moved, the offset can be corrected by the same processing. With such an operation, for example, when about 10 frames are taken by moving one pixel at a time, it is possible to use a part of the frames by thinning out the frames. There is an advantage that becomes larger.

{Others}
If the offset fluctuation is not steep, the offset correction table may be updated intermittently once every fixed time. In the case of intermittently performing in this way, the directional unit 2 can be driven in a sub-pixel step and used for super resolution while the correction table is not updated.

  It is also possible to use all acquired frames in order to update the offset correction table. Even in this case, if the number of frames used for offset correction increases, there is an advantage that the influence of noise is reduced.

  In the embodiment described above, the offset correction table is updated based on the relative value obtained by comparing the output values after correction of the elements. Therefore, the offset correction table is updated based on the relative value of the output value after performing only the gain correction, not only when using the data after performing both the offset correction and the gain correction as shown in the embodiment. It is also possible to calculate a value.

  The directing unit 2 may direct the entire image acquisition unit or may move the lens by moving a lens or a mirror that is a part of the image acquisition unit 1. The directing unit 2 may be configured by a micro scanner or a gimbal.

(Appendix 1) Image acquisition means;
Directing means for moving the visual axis of the image acquisition means;
Correction means for correcting the gain and offset of the output of each element included in the image acquisition means;
After the image acquisition means acquires an image, before the next image is acquired, the directing means repeatedly moves the visual axis of the image acquisition means to acquire a plurality of images, thereby obtaining a plurality of different elements. Then, an update amount is obtained by using the output difference between the elements of the plurality of elements after correction by the correction means for the same imaging target, and is used in the correction means for at least one of the plurality of elements based on the update amount. Data updating means for updating offset correction data;
An offset correction processing apparatus characterized by comprising:

(Supplementary Note 2) The data update means calculates a first inter-element output difference when the first element group images the first part of the imaging target,
Among the plurality of images, one element of the plurality of elements is imaged with an image different from an image obtained by imaging the first part, and the one element is the second element to be imaged. When the second part is also picking up an image of the second part, the second element output difference that occurs when the second part is picked up with respect to the second element group picking up the second part is also obtained. Calculate based on the one element,
An update amount is obtained using the first output difference between the elements and the second output difference between the elements, and at least of the elements included in the first element group or the second element group based on the update amount. The offset correction processing apparatus according to appendix 1, wherein the offset correction data used by the correction unit for one element is updated.

(Supplementary Note 3) The data updating means may use one of the first element groups as a first reference element, and the other output of the first element group from the corrected output of the first reference element. Subtracting the corrected output of the element to calculate the output difference between the elements relative to the first reference element,
Using one of the second element groups as a second reference element, subtracting the corrected output of the other elements of the second element group from the corrected output of the second reference element Calculate the output difference between the elements for the two reference elements,
Further calculating an output difference between the first reference element and the second reference element between the reference elements that occurs when imaging the same portion of the imaging target,
From the calculated output difference between the reference elements and the update amount of the first reference element, obtain the update amount of the second reference element,
By adding the output difference between the elements with respect to the second reference element calculated by the data updating means, and the output difference between the reference elements, to the update amount of each element of the second element group, Update the offset correction data of at least one element of the second element group, the output difference between the elements when imaging the first part, and the element difference when imaging the second part The offset correction processing device according to appendix 2, wherein the output difference between the two is reduced.

(Supplementary Note 4) A comparison region for calculating a correlation value between images is set in each of the plurality of images, and the correlation between the comparison region of one image of the plurality of images and the comparison region of another image is set. And a correction table updatable area calculating means for determining that the comparison area of each image is an area where the correction table can be updated when a certain correlation or more is obtained,
The data updating means selects a plurality of elements that have imaged the area determined to be amendable by the correction table updateable area calculating means,
Based on the output after correction by the correction means of one of the selected elements, the update amount is determined using the difference in output from the other elements of the selected elements. The offset correction processing apparatus according to any one of appendices 1 to 3, wherein the offset correction data is updated according to the update amount.

(Supplementary Note 5) After the image is acquired by the image acquisition unit, the directing unit moves the visual axis of the image acquisition unit, and the image acquisition unit acquires the plurality of images by repeating the process of acquiring the image after the movement of the visual axis. To image the same imaging target with a plurality of different elements,
Using the difference in output from the plurality of elements when the gain and offset of the output of each element included in the image acquisition unit are corrected by a correction unit, an update amount is obtained,
An offset correction processing method, comprising: updating offset correction data used in the correction unit for at least one of the plurality of elements according to the update amount.

(Supplementary Note 6) By repeating the imaging by the image acquisition unit and the movement of the visual axis by the directing unit, a plurality of images are acquired,
In the plurality of images, specify a region in a relationship in which the same portion of the imaging target is imaged with different elements,
For the plurality of elements that have imaged the specified region, for each portion of the imaging target that has been imaged, when the gain and offset of the output of each element included in the image acquisition means are corrected by the correction means, Using the output difference between multiple elements, find the update amount,
6. The offset correction processing method according to appendix 5, wherein offset correction data used by the correction unit is updated according to the update amount.

(Appendix 7) A reference image is selected from the plurality of images,
Cutting out a first comparison region from the selected image;
The other image is clipped so that the position of the visual axis of the image acquisition means when each image is acquired and the amount of movement of the visual axis when the image acquisition means acquires the reference image are cancelled. Move to cut out the second comparison area,
Comparing the first comparison region cut from the reference image with the second comparison region cut from the other image;
In the compared images, when the same part of the imaging target is imaged at the same position in the certain area, it is determined that the same part of the imaging target is captured by different elements. The offset correction processing method according to appendix 6, wherein:

(Supplementary Note 8) As a result of imaging the same part of the imaging target with a plurality of different elements, when there are a plurality of parts imaged as the same part,
One of the plurality of elements is used as a reference element for updating offset correction data,
For each part imaged as the same part, determine the output difference between the elements that imaged the same part,
The output difference between the output of each element and the element used as the reference for updating, which occurs when each element of the plurality of elements and the reference element for updating image one of the same parts of the imaging target To calculate
The offset correction processing method according to any one of appendices 6 to 8, wherein the offset correction data used in the correction unit is updated with an update amount obtained from the calculated output difference.

FIG. 1 is a diagram illustrating an outline of an aspect of an offset correction processing apparatus. FIG. 2 is a diagram illustrating a specific example of the configuration of the offset correction processing apparatus. FIG. 3 is a flowchart for calculating the initial correction table. FIG. 4 is a flowchart for obtaining a corrected image. FIG. 5 is a diagram illustrating scanning by the directing unit. FIG. 6 is a flowchart of the movement of the visual axis by the directional unit and the imaging after the movement of the visual axis. FIG. 7 is a flowchart when the correction table updateable area is calculated. FIG. 8 is a flowchart for updating the offset correction table. FIG. 9 is a diagram for explaining the range of the target scene imaged in each frame. FIG. 10 is a diagram showing an image coordinate system. FIG. 11 is a diagram for explaining the cutting of the comparison area and the coordinate system of the comparison area. FIG. 12 is a diagram for explaining the cut-out coordinates of the comparison area. FIG. 13 is a diagram illustrating an example of a correlation image when the correlation value is calculated using the phase correlation. FIG. 14 is a diagram illustrating an example of a correlation image in the case where the movement of the imaging device or the like is detected by calculating a correlation value using phase correlation.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image acquisition part 2 Direction part 3 Correction | amendment part 4 Correction table updateable area calculation part 5 Correction table update part 10 Lens part 11 Micro scanner part 12 Optical path switching part 13 Low temperature reference | standard hot plate part 14 High temperature reference | standard hot plate part 15 Photoelectric conversion part DESCRIPTION OF SYMBOLS 16 Amplification part 17 A / D conversion part 18 Initial correction table calculation part 19 Low temperature image storage memory 20 High temperature image storage memory 21 Gain storage memory 22 Offset storage memory 23 Correction processing part 24 Comparison area extraction part 25 Frame (1) storage memory 26 Frame (2) storage memory 27 Frame (3) storage memory 28 Frame (4) storage memory 29 Phase correlation processing unit 30 Offset correction table update unit 31 Average value calculation memory 40 Part of imaging target 41 Frame of part (40) of imaging target 1) Pixel imaged at the time of imaging 42 Frame 40 ) Pixel imaged at the time of imaging 43 Frame of part 40 of the imaging target (3) Pixel imaged at the time of imaging 44 Frame (4) Pixel imaged of the part 40 of the imaging target at the time of imaging 50 Effective area 51, 51 a, 51 b, 51 c, 51 d Comparison area 52, 52a, 52b, 52c, 52d Points of interest for correlation processing

Claims (3)

Image acquisition means;
Directing means for moving the visual axis of the image acquisition means;
Correction means for correcting the gain and offset of the output of each element included in the image acquisition means;
After the image acquisition means acquires an image, before the next image is acquired, the directing means repeatedly moves the visual axis of the image acquisition means to acquire a plurality of images, thereby obtaining a plurality of different elements. Then, an update amount is obtained by using the output difference between the elements of the plurality of elements after correction by the correction means for the same imaging target, and is used in the correction means for at least one of the plurality of elements based on the update amount. Data updating means for updating offset correction data;
I have a,
The data updating means uses the first reference element as one of the first element groups when the first element group images the first part to be imaged. Subtracting the corrected output of the other elements of the first element group from the corrected output of the first element to calculate the output difference between the elements relative to the first reference element,
Among the plurality of images, one element of the plurality of elements is imaged with an image different from an image obtained by imaging the first part, and the one element is the second element to be imaged. In the case where the second part is also picked up by the element different from the second part, one of the second element groups is designated as the second reference element. And subtracting the corrected output of the other elements of the second element group from the corrected output of the second reference element to calculate the inter-element output difference with respect to the second reference element,
Further calculating an output difference between the first reference element and the second reference element between the reference elements that occurs when imaging the same portion of the imaging target,
From the calculated output difference between the reference elements and the update amount of the first reference element, obtain the update amount of the second reference element,
By adding the output difference between the elements with respect to the second reference element calculated by the data updating means, and the output difference between the reference elements, to the update amount of each element of the second element group, Update the offset correction data of at least one element of the second element group, the output difference between the elements when imaging the first part, and the element difference when imaging the second part To reduce the output difference between
An offset correction processing apparatus. A comparison region for calculating a correlation value between images is set for each of the plurality of images, and a correlation between a comparison region of one image of the plurality of images and a comparison region of another image is calculated. And a correction table updatable area calculating means for determining that the comparison area of each image is an area where the correction table can be updated when a certain correlation or more is obtained,
The data updating means selects a plurality of elements that have imaged the area determined to be amendable by the correction table updateable area calculating means,
Based on the output after correction by the correction means of one of the selected elements, the update amount is determined using the difference in output from the other elements of the selected elements. The offset correction processing apparatus according to claim 1 , wherein the offset correction data is determined and updated according to the update amount. In the offset correction processing method,
After the image acquisition by the image acquisition means, the directing means moves the visual axis of the image acquisition means, and after the movement of the visual axis, the image acquisition means repeats the process of acquiring images to acquire a plurality of images. An imaging step of imaging the same imaging target with a plurality of elements;
The difference update amount obtaining step asking you to update amount using the output from the plurality of elements when the gain and offset of the output of each element was corrected by the correction means included in the image acquisition unit,
An offset correction data update step for updating offset correction data used in the correction means for at least one of the plurality of elements according to the update amount ;
With
In the offset correction data update step, one of the first element groups when the first element group images the first part to be imaged is defined as the first reference element. Subtracting the corrected output of the other elements of the first element group from the corrected output of the reference element to calculate the inter-element output difference for the first reference element,
Among the plurality of images, one element of the plurality of elements is imaged with an image different from an image obtained by imaging the first part, and the one element is the second element to be imaged. In the case where the second part is also picked up by the element different from the second part, one of the second element groups is designated as the second reference element. And subtracting the corrected output of the other elements of the second element group from the corrected output of the second reference element to calculate the inter-element output difference with respect to the second reference element,
Further calculating an output difference between the first reference element and the second reference element between the reference elements that occurs when imaging the same portion of the imaging target,
From the calculated output difference between the reference elements and the update amount of the first reference element, obtain the update amount of the second reference element,
By adding the output difference between the elements with respect to the second reference element calculated by the data updating means, and the output difference between the reference elements, to the update amount of each element of the second element group, Update the offset correction data of at least one element of the second element group, the output difference between the elements when imaging the first part, and the element difference when imaging the second part To reduce the output difference between
An offset correction processing method characterized by the above.

Images