JP2009198452A - Device for measuring deformation with use of plural images - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a displacement measuring device that, even when an object to be measured is greatly deformed, can surely measuring the deformation. <P>SOLUTION: The displacement measuring device includes: a section for calculating a displacement in pixel between adjacent images by which section, for each pair of images among at least three images taken, that pixel in a post-deformation image which is associated with a target pixel b<SB>i</SB>(x, y) contained in that region of a pre-deformation image which is to be measured is determined as a reference pixel a<SB>j</SB>(x+m<SB>x</SB>, y+m<SB>y</SB>) and the amount of displacement in units of pixel (m<SB>x</SB>, m<SB>y</SB>) is computed for each pixel contained in the measured region; a section for calculating a displacement in subpixel between adjacent images by which section the amount of displacement in reference pixel in the order of subpixels (Δx<SB>k</SB>, Δy<SB>k</SB>) is calculated in accordance with image information on the reference pixel a<SB>j</SB>(x+m<SB>x</SB>, y+m<SB>y</SB>) and a pixel adjacent thereto; and a section for calculating the cumulative amount of displacement by which section the amount of displacement in reference pixel in units of pixels and the amount of displacement in the order of subpixels, which amounts have been calculated for each pair of images, are cumulatively added and the amount of displacement from the start of measurement to the end of measurement is calculated. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a deformation measuring apparatus that measures the amount of deformation of an object to be measured that is deformed over time. The present invention can be applied to deformation measurement in stress measurement, for example.

  Conventionally, in measuring the stress of an object to be measured, strain and displacement are measured by attaching a strain gauge to a measurement site. With strain gage measurement, highly reliable measurement data can be obtained, but only local measurement data at the location where it is attached can be obtained, and the overall deformation state of the object to be measured can be grasped. There wasn't.

  Therefore, as a method of measuring displacement over the entire measurement area set on the object to be measured, not only at the specific position where the strain gauge is attached, the amount of displacement is analyzed by analyzing the captured image taken with the camera device. A method of calculating is disclosed. That is, it is disclosed that two images before and after the deformation of the object to be measured are photographed, and the deformation that has occurred in the object to be measured is obtained from the correlation between the photographed images (see Patent Document 1).

In addition, when the amount of displacement is obtained from the correlation amount between images using an image captured by the camera device, the measurement accuracy is determined in pixel units (size of one pixel), but in order to further increase the measurement accuracy A method for measuring a so-called sub-pixel order displacement amount of one pixel or less by calculating a least-squares surface for the correlation between pixel information (for example, luminance information) between captured images is disclosed (see Patent Document 2). ).
JP-A-7-181075 JP 2006-343160 A

As described in Patent Document 1 and Patent Document 2, according to the method of calculating the amount of displacement by comparing the captured image before the deformation with the captured image after the deformation, the entire measurement region projected on the image The displacement amount can be measured for each part.
However, when the deformation or distortion of the object to be measured is large or accompanied by a large rotational movement, it may be difficult to compare the captured images before and after the deformation. That is, the correlation between pixel information (for example, luminance information) between captured images is reduced, and it is difficult to accurately associate the pixels that appear in the pre-deformation captured image with the pixels that appear in the post-deformation image.

  Therefore, the present invention provides a displacement measuring device that can reliably measure the amount of displacement in the entire measurement region even when the deformation or distortion of the object to be measured is large or when there is a large rotational movement. For the purpose.

In order to solve the above problems, the present invention is to capture at least one deformation state at an intermediate point in time in addition to the pre-deformation and post-deformation photographed images, and change the state with two images arranged in time series. Are obtained by measuring the amount of displacement and finally accumulating the amount of displacement.
That is, the deformation measurement apparatus of the present invention includes a captured image accumulation unit, a continuous image inter-pixel displacement calculation unit, a continuous inter-image sub-pixel displacement calculation unit, and a cumulative displacement amount calculation unit. The unit accumulates at least three captured images including a measurement start point, one or more intermediate points, and a measurement end point for the measurement target region. At this time, the number of photographed images at the intermediate point is increased as the deformation becomes larger so that the amount of displacement between the two photographed images arranged in time series is not increased.
The inter-continuous image pixel displacement calculation unit sequentially extracts two chronologically continuous captured images from the captured images stored in the captured image storage unit to form a pair of images. In each pair of images, the first captured image is the pre-deformation image, the later captured image is the post-deformation image, and for the pixel of interest b _i (x, y) included in the measurement target region of the pre-deformation image, Collation by searching for a pixel associated with the target pixel b _i (x, y) in the transformed image based on a correlation amount of image information (for example, luminance information) around the target pixel b _i (x, y). pixel _{a j (x + m x,} y + m y) is determined as. Further performs the target pixel _b i (x, y) and the collation pixel _{a j (x + m x,} y + m y) displacement in pixels of the _(m x, m _y) the calculation for obtaining the. Further, the pixel of interest is moved one after another to determine a collation pixel in the post-deformation image for each pixel in the measurement area of the pre-deformation image and calculate a displacement amount in pixel units. In this manner, for each pair of images, a matching pixel in the post-deformation image for each pixel (all pixels) in the measurement region of the pre-deformation image is determined, and a displacement amount in units of pixels is calculated.
Subpixel displacement calculation unit between successive images, the matching pixel _a j in the deformed image _{(x + m x, y +} m y), and, wherein matching pixel _{a j (x + m x,} y + m y) image information of a pixel adjacent to (e.g. Based on the correlation amount of (luminance information), the displacement amount (Δx _k , Δy _k ) of the pixel of interest b _i (x, y) in the sub-pixel order equal to or less than the pixel unit is _obtained . That is, the inter-continuous image pixel displacement calculation unit determines the matching pixel a _j (x + m _x , y + m _y ) associated with the target pixel b _i (x, y) in units of pixels, but further increases the measurement accuracy. Therefore, the calculation is performed including the correlation amount of the image information of the pixels adjacent to the matching pixel a _j (x + m _x , y + m _y ), and the target pixel b _i (x, y) in the sub-pixel order below the pixel unit. ) Is obtained (Δx _k , Δy _k ). At this time, for example, the least square approximation, the steepest descent method, or the like can be used as a method for capturing the correlation amount of the image information of adjacent pixels.
The cumulative displacement amount calculation unit cumulatively adds the displacement amount in pixels calculated for each pair of images and the displacement amount in the sub-pixel order, and calculates the displacement amount from the measurement start point to the measurement end point in the measurement target region. For each pixel.

According to the present invention, even when displacement or strain is large or rotational movement is applied, the displacement amount is divided and measured using the captured image at the intermediate time point, and finally each displacement amount is cumulatively added. By doing so, the displacement amount can be reliably measured.
Further, in the present invention, when obtaining the position of the matching pixel corresponding to the target pixel between adjacent images, the calculation is performed by performing only the calculation for each pixel and not including the displacement amount calculation in the sub-pixel order. When calculating the amount of displacement so that time is not enormous, calculate the amount of displacement in the sub-pixel order calculated for each pair of adjacent images, and accumulatively add together with the amount of displacement in pixels. Therefore, it is possible to increase the measurement accuracy and suppress the calculation time, and it is possible to perform measurement with a balance between the accuracy and the calculation time.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings, taking as an example the case where the object to be measured is a test piece attached to a tensile tester. The object to be measured is not particularly limited, and any object can be applied as long as the image before deformation and the image after deformation can be compared.

(Device configuration)
FIG. 1 is a diagram showing a configuration of a displacement measuring apparatus according to an embodiment of the present invention. The upper and lower ends of the test piece S that is the object to be measured are attached to the movable chuck 1 and the fixed chuck 2 of the tensile tester. By operating the movable chuck 1, a load is applied to the test piece S, and the test piece S is deformed in accordance with the load. In addition, a camera device 3 (CCD camera) is arranged beside the test piece S so that the measurement region K (width D, length L) set on the test piece S can be copied.
The camera device 3 captures an image before the load is applied (measurement start image) and an image after the load is applied (intermediate image, measurement end image), and transmits these images to the control unit 10. It is. If the camera apparatus is provided with a shooting interval adjustment mechanism that adjusts the time interval between shootings, the time interval between images can be adjusted more easily.

  The control unit 10 is configured by a computer device including a CPU and a memory, and further includes a display device 10a (for example, a liquid crystal panel) and an input device 10b (for example, a keyboard and a mouse), and displays a captured image on the display device 10a. Various inputs necessary for control can be performed via the input device 10b. For example, the measurement region K can be set as one of the input operations related to the present invention. The control unit 10 may be further provided with a data input / output device 10c (for example, a CD drive device or a DVD drive device). In this case, image data captured by an external camera device such as a digital camera is input as data. Input / output can be performed via the output device 10c.

  In order to describe the arithmetic processing performed by the control unit 10, the control unit 10 will be described separately for each functional block. The captured image storage unit 11, the inter-continuous image pixel displacement calculation unit 12, and the inter-continuous image sub-pixel displacement calculation. The unit 13 and the cumulative displacement amount calculation unit 14 are configured.

  The captured image storage unit 11 performs a process of storing three or more captured images including a measurement start point, an intermediate point, and a measurement end point for a region including the measurement target region. The captured image may be acquired from the camera device 3, or an image acquired by the external camera device may be acquired via the data input / output device 10c. The number of captured images to be accumulated varies depending on the number of captured images at the intermediate time point. The number of captured images required from the start of measurement to the end of measurement depends on the brightness distribution of the entire object to be measured, but it depends on the magnitude of deformation, distortion, or when rotational movement is involved. An appropriate number of images are accumulated so that the amount of displacement between images arranged in time series is not so large according to the magnitude of rotation.

  The inter-continuous image pixel displacement calculation unit 12 sequentially extracts two chronologically consecutive captured images from the captured images stored in the captured image storage unit 11 to form a pair of images. For example, FIG. 2 shows the case of three photographed images G1 (measurement start image), G2 (intermediate image), and G3 (measurement end image), but G1-G2 pairs and G2-G3 pairs are extracted. Become. When G4 and G5 are further increased, a G3-G4 pair and a G4-G5 pair are added.

Then, the inter-continuous image pixel displacement calculation unit 12 sets the first captured image as the pre-deformation image (denoted as “front” in FIG. 2) in each pair of images, and the subsequent image as the post-deformation image (see FIG. 2). and 2 referred to as "after" in), with respect to the target pixel b _i included in the measurement target region before deformation image _(x, y), is associated with the target pixel b _{i (x,} y) in the deformed image A pixel is determined as a matching pixel a _j (x + m _x , y + m _y ) by searching based on the correlation amount of image information (for example, luminance information) around the pixel of interest b _i (x, y).
Specifically, in the G1-G2 pair, it is considered that one pixel on the image G2 corresponding to the target pixel b ₁ (x ₁ , y ₁ ) on the image G1 (that is, the target pixel moves on the G2). _Is determined based on the luminance information around the pixel of interest b ₁ (x, y) and determined as a matching pixel a ₁ (x ₁ + m _x1 , y ₁ + m _y1 ). A specific method for searching for a matching pixel based on the correlation amount of luminance information will be described later.
When the matching pixel a ₁ (x ₁ + m _x1 , y ₁ + m _y1 ) on G2 is determined, the target pixel b ₁ (x ₁ , y ₁ ) and the matching pixel a ₁ (x ₁ + m _x1 , y ₁ + m _y1 ) A displacement amount (m _x1 , m _y1 ) in units of pixels between and is obtained. When the process for one pixel of interest is completed, the same calculation is repeated for a new pixel of interest after shifting by one pixel. The same process is repeated over the entire measurement area.

Subsequently, since the inter-continuous image pixel displacement calculation unit 12 performs the same search with the G2-G3 pair, the matching pixel a ₁ (x ₁ + m _1x , y ₁ + m _1y on G2 determined by the previous calculation is determined. ) Is set as the pixel of interest b ₂ (x ₂ , y ₂ ) on G2, and the image G2 is treated as a pre-change image in the next G2-G3 pair calculation. That is, the following substitution is performed.
x ₂ = x ₁ + m _1x (1)
y ₂ = y ₁ + m _1y (2)
Then, by performing the same search operation to determine a matching pixel _a 2 on the image G3 to be changed after the image _{(x 2 + m 2x, y} 2 + m 2y).
When the matching pixel a ₂ (x ₂ + m _x2 , y ₂ + m _y2 ) on G3 is determined, the pixel of interest b ₂ (x ₂ , y ₂ ) and the matching pixel a ₂ (x ₂ + m _x2 , y ₂ + m _y2 ) The displacement amount (m _2x , m _2y ) in units of pixels is obtained. The same process is repeated over the entire measurement area.
Thereafter, when there are still other image pairs, the same search operation and displacement amount calculation are repeated.

  By repeating the above calculation, a pixel on G2 (intermediate image) corresponding to each pixel in the measurement region on the image G1 (measurement start image), and further a G3 (measurement end) corresponding to each pixel in the measurement region on G2 Image) pixels are determined, and correspondence from pixels on the measurement start image to pixels on the measurement end image is made possible.

Here, a specific method by which the inter-continuous image pixel displacement calculation unit 12 searches for the matching pixel based on the correlation amount of the luminance information will be described.
FIG. 3 is a diagram for explaining image matching for obtaining a correspondence relationship between pixels of the pre-deformation image and the post-deformation image.
As shown in FIG. 3A, the target pixel to be associated with the post-deformation image is set in the pre-deformation image. Hereinafter, the pixel of interest in the pre-deformation image, measuring the target pixel _b i (x, y) that. Then, a small area of dx × dy with the measurement target pixel b _i (x, y) as the center is set. Hereinafter, this minute area is referred to as a reference image area portion B (b _i ). The reference image area B (b _i ) is an area where image matching is performed when searching for a pixel of the post-deformation image corresponding to the measurement target pixel b _i (x, y). If the number of pixels included in this region is increased, the accuracy of image collation increases, but the calculation time takes longer. Specifically, about 20 to 100 pixels are set as dx and dy.

On the other hand, as shown in FIG. 3B, the target pixel of interest is set in the post-deformation image in order to obtain a correspondence relationship with the measurement target pixel b _i (x, y). Hereinafter, the target pixel in the post-deformation image is referred to as a search target pixel a _j (x + Δx, y + Δy). Then, a small area of dx × dy (same size as the reference image area B (b _i )) with the search target pixel a _j (x + Δx, y + Δy) as the center is set. Hereinafter, this small area is referred to as a comparative image area portion A (a _j ). The comparison image area part A (a _j ) is an area where image comparison with the reference image area part B (b _i ) is performed. The search target pixel a _j (x + Δx, y + Δy) serving as the center is shifted by one pixel at the time of calculation, so that the position of the comparison image region portion A (a _j ) moves one after another.

Then, the image information of each comparison image region portion A (a _j ) when the position of the comparison image region portion A (a _j ) is successively shifted and the image information of the reference image region portion B (b _i ) Based on this, the correlation between each other is obtained. Here, luminance information is used as image information for obtaining the correlation. For color images, color information such as RGB information can be taken.

As the correlation value R of the luminance information used for image matching, the luminance value of each pixel included in the reference image region portion B (b _i ) and the luminance value of each pixel corresponding to the comparison image region portion A (a _j ) Or the luminance value residual norm R _ij (b _i , a _j ) (the square root of the luminance value residual square sum) represented by the following equation (3).
R _ij (b _i , a _j ) = || A (a _j ) −B (b _i ) || (3)

Then, since the comparison image area A (a _j ) when the residual norm R _ij (b _i , a _j ) becomes the minimum is the area having the highest correlation, this is obtained, and the search target pixel at that time _{a j (x + m x,} y + m y) obtained as a matching pixel position. Here m _x, m _y is the pixel of the [Delta] x, [Delta] y the residual norm is minimized, which is the displacement to the position after displacement from the position before displacement (displacement solution).

In the above example, the residual norm R _ij (b _i , a _j ) is used as a preferred example as the correlation amount of image information between pixels. However, when comparing pixels in a minute region, the similarities are used. Other correlations may be used as long as the degree of correlation can be quantified. For example, a weighted average value of luminance values of pixels in a minute area may be set as a correlation amount so as to be closer to the target pixel.

In this way, displacement m _x, m _y obtained by continuous image between pixels displacement calculation unit 12, a displacement amount in pixels. In order to increase the measurement accuracy, it is required to measure the amount of displacement in the sub-pixel order (pixel unit or less). Subsequently, the inter-image sub-pixel displacement calculation unit 13 performs arithmetic processing for increasing the accuracy. .

FIG. 4 is a diagram for explaining calculation processing for obtaining the displacement amount in the sub-pixel order based on the displacement amount (displacement solution) obtained in pixel units. Although this figure shows only the x direction, the same applies to the y direction.
Correct value than the measurement value in pixels as the displacement amount, matching pixel _{a j (x + m x,} y + m y) present in the central and the fluctuation width of one pixel of. Therefore, based on the position coordinates of the matching pixel a _j (x + m _x , y + m _y ) that is obtained by the inter-image pixel displacement calculation unit 12 and has the minimum luminance value residual norm R _ij (b _i , a _j ). , matching pixel _a j in the deformed image _{(x + m x, y +} m y), and, x, the correlation of pixel information of ± 1 pixel by the adjacent total 9 points plus the y direction (3 pixels × 3 pixels) Ask for. Specifically, a least square surface g (x, y) represented by the following equation (4) is created from the luminance correlation values of these nine points.
g (x, y) = ax ² + bx + cy ² + dy + exy + f (4)
Here, a to f are coefficients obtained by the method of least squares.

For example, as shown in Figure 4 for a one-dimensional direction of the x-coordinate, matching pixel _a j (x _{+ m} x) in the deformed image and the pixel _a j (x + _{m x} adjacent to the matching pixels _a j _(x + _m x) -1), a least square surface is created based on the correlation amount of the image information of the pixel a _j (x + m _x +1), and the pixel of interest b _i (sub pixel order) is calculated based on the extreme value obtained from the least square surface. The displacement amount (Δx _k ) of x) is obtained. In the two-dimensional coordinate system, matching pixel _a j in the deformed image _{(x + m x, y +} m y), and, matching pixel _{a j (x + m x ±} 1, y + m y ± 1) Total nine adjacent pixels A least-square surface is created based on the correlation amount of the image information, and the displacement amount (Δx _k , Δy _k ) of the pixel of interest b _i (x, y) in the sub-pixel order based on the extreme value obtained from the least-square surface. )

By executing the least square approximation based on this equation, each coefficient of the least square curved surface g (x, y) is determined, and a curved surface equation is obtained.
Then, by obtaining the change in coordinates (Δx _k , Δy _k ) when the obtained curved surface formula takes an extreme value, the displacement amounts Δx _k , Δy _k (displacement solution) are calculated in the subpixel order. Become.
Thereafter, the same processing is repeated for each pixel and each image pair in the measurement region.
Describing corresponding to the example of FIG. 2, the coordinates (x ₁ + m _1x + Δx _1k , y ₁ ) of the extreme value corresponding to the matching pixel a ₁ (x ₁ + m _1x , y ₁ + m _1y ) in the G1-G2 pair. + _{M 1y} + Δy _1k ) is obtained, and displacement amounts Δx _1k and Δy _1k (displacement solutions) are determined. Similarly, corresponding to matching pixel _{a 2 (x 2 + m x2} , y 2 + m y2) in G2-G3 pair, the coordinates of the extreme value _{(x 2 + m x2 + Δx} 2k, y 2 + m y2 + Δy 2k) is Motomari, Displacement amounts Δx _2k and Δy _2k (displacement solutions) are determined.

  In the above-described example, the least square surface approximation is used for obtaining the extreme value, but other approximation methods such as the steepest descent method may be used instead.

Next, the cumulative displacement calculation unit 14 will be described. The cumulative displacement amount calculation unit 14 cumulatively adds the displacement amount in pixel units calculated for each image pair and the displacement amount in the sub-pixel order for each pixel in the measurement target region, and from the measurement start time to the measurement end time. The amount of displacement up to is calculated.
In the specific example of FIG. 2, the sum (X, Y) of the displacement amount in pixel units and the displacement amount in the sub-pixel order from G1 (measurement start image) to G3 (intermediate image) to G3 (measurement end image). Ask for. That is,
X = m _x1 + m _x2 + m _x3 + Δx _1k + Δx _2k + Δx _3k
Y = m _x1 + m _x2 + m _x3 + Δx _1k + Δx _2k + Δx _3k
It becomes.
In this way, by accumulating the displacement amount with the intermediate image interposed, even when a large displacement occurs from the measurement start image to the measurement end image, a correspondence between the pixels can be obtained, and the displacement amount The sum can be measured.

(Measurement operation)
Next, the procedure of the measurement operation when measuring the deformation of the test piece by the displacement measuring device will be described. FIG. 5 is a flowchart showing the procedure of the measurement operation of the present invention.

  The measurement apparatus is activated, the measurement start image, the intermediate image, and the measurement end image are captured in time series, and the captured images are stored in the memory (S101). During this time, the movable chuck 1 of the tensile tester is operated to continuously increase the load so that a large deformation is applied to the test piece. The number of intermediate images to be taken is set to an appropriate number according to the size of the deformation.

Subsequently, the continuous image inter-pixel displacement calculation unit 12 extracts image pairs from the younger in time series, and applies the post-deformation to the target pixel b _i (x, y) included in the measurement target region of the pre-deformation image. target pixel b _{i (x,} y) in the image pixels associated with the, target pixel b _{i (x,} y) matching pixel a by searching based on a correlation of the image information around (e.g., luminance information) _j (x + m _x , y + m _y ) is determined, and the displacement amount (m _x , m _y ) in pixel units is calculated (S102).

Subsequently, the continuous image between the sub-pixel displacement calculation unit 13, the matching pixel _{a j (x + m x,} y + m y) , and, matching pixel _{a j (x + m x,} y + m y) the correlation of the luminance information of pixels adjacent to the Based on this, a least-square approximation is performed, and a displacement (Δx _k , Δy _k ) of the pixel of interest b _i (x, y) in the sub-pixel order is obtained to perform a calculation process to increase the accuracy (S103).

  It is determined whether the corresponding matching pixel has been determined for the pixels in the entire measurement region in the image pair being measured (S104). If there is an undetermined pixel, it is shifted by one pixel (S105), the process returns to S102, and the same processing is repeated. When the corresponding matching pixels are determined for all the pixels in the measurement region, the process proceeds to S106.

  When the collation pixel of the pixels in the entire measurement region has been determined for the image pair being measured, it is determined whether measurement has been completed for all image pairs from the measurement start image to the measurement end image (S106). If measurement for all image pairs has not been completed, the image pair is moved by one (S107), and the processing from S102 onward is repeated. If all image pairs have been completed, the process proceeds to S108.

  Subsequently, when the cumulative displacement amount calculation unit 14 obtains a correspondence relationship from the measurement start point to the measurement end point for each pixel in the measurement target region of the measurement start image, the intermediate image, and the measurement end image arranged in time series. At the same time, the displacement amount in units of pixels calculated for each image pair and the displacement amount in the sub-pixel order are cumulatively added to calculate the total displacement amount from the measurement start point to the measurement end point (S108).

  With the above processing procedure, the total displacement amount from the measurement start time point to the measurement end time point can be obtained regardless of the magnitude of the displacement amount.

  As an example of the present invention, the deformation of the test piece attached to the tensile tester has been described as an example, but as described above, the present invention can be used in other measurements as long as the image can be compared before and after the deformation. Can be applied.

  The present invention can be used in a displacement measuring device that measures the amount of displacement of an object to be measured that is deformed over time.

The block diagram which shows the structure of the displacement measuring device which is one Embodiment of this invention. The figure explaining the arithmetic processing performed for every image pair from a measurement start image to a measurement end image. The figure explaining the correspondence of the pixels of the image before a deformation | transformation, and the image after a deformation | transformation. The figure explaining the arithmetic processing which calculates | requires the displacement amount in a sub pixel order based on the displacement amount (displacement solution) calculated | required per pixel. The flowchart at the time of measuring the amount of displacement with the displacement measuring device which is one embodiment of the present invention.

Explanation of symbols

1: Movable chuck 2: Fixed chuck 3: Camera device (CCD camera)
10: Control unit 10a: Display device 10b: Input device 10c: Data input / output device 11: Captured image storage unit 12: Continuous image pixel displacement calculation unit 13: Continuous image subpixel displacement calculation unit 14: Cumulative displacement amount calculation unit

Claims (3)

A displacement measuring device that measures a displacement amount from a measurement start time point to a measurement end time point for a measurement target region of a deformed object to be measured,
A captured image storage unit that stores at least three captured images including a measurement start time, one or more intermediate time points, and a measurement end time for the measurement target region;
Two consecutive captured images in time series are sequentially extracted from the stored captured images, and in each pair of images, the first captured image is the pre-deformation image, the later captured image is the post-deformation image, undeformed target pixel included in the measurement target region of the image b _{i (x,} y) with respect to, the deformed image in the target pixel b _{i (x,} y) in the pixel associated with the, target pixel b _{i (x} , matching pixel _a j _(x + m _x by searching based on a correlation of the image information y) near, y + _{m y)} is determined as, the target pixel _b i (x, y) and the collation pixel _a j (x _{+ m} x, y + m _y ) is calculated to obtain the displacement amount (m _x , m _y ) in pixel units, and the pixel of interest is moved one after another, and each pixel in the measurement area of the pre-deformation image Determine the matching pixel and A continuous image between pixels displacement calculation unit for calculating a displacement amount of the cell unit,
Matching pixel _a j in said deformed image _{(x + m x, y +} m y), and, wherein matching pixel _{a j (x + m x,} y + m y) based on the correlation amount of the image information of the pixels adjacent to, the following pixels A sub-pixel displacement calculation unit between successive images for obtaining a displacement amount (Δx _k , Δy _k ) of the pixel of interest b _i (x, y) in the sub-pixel order;
Cumulative addition of the displacement amount in pixel units calculated for each pair of images and the displacement amount in the sub-pixel order for each pixel in the measurement target region, and calculating the displacement amount from the measurement start point to the measurement end point A displacement measuring device comprising a displacement amount calculating unit. The continuous image between pixels displacement calculation unit, the target pixel b _{i (x,} y) in the measurement target region before deformation image and the target pixel b _{i (x,} y) reference image composed of pixels included in the small region around An area B (b _i ) is set, and the comparison image area A (consisting of pixels included in a small pixel around the search pixel a _j (x + Δx, y + Δy) and the search pixel a _j (x + Δx, y + Δy) is set in the transformed image. set a _j), the reference image area part B (b _i) and the comparative image area part a (a _j) and the correlation amount of the image information between pixels in the comparison image with respect to the reference image area part B (b _i) By calculating the position of the area portion A (a _j ) while shifting one pixel at a time, one of the search pixels a _j (x + Δx, y + Δy) is converted into the matching pixel a _j (x + m _x , y + m _y ) The displacement measuring device according to claim 1 to be determined. The continuous image between the sub-pixel displacement calculation unit, matching pixel _a j in the deformed image _{(x + m x, y +} m y), and, wherein matching pixel _{a j (x + m x,} y + m y) of the image information of the pixels adjacent to the A least square surface is created based on the correlation amount, and a displacement amount (Δx _k , Δy _k ) of the pixel of interest b _i (x, y) in the subpixel order is obtained based on the extreme value obtained from the least square surface. The displacement measuring apparatus according to claim 1.