JP2017083364A - Calculation device for modulation transfer function and calculation program for modulation transfer function - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a modulation transfer function (MTF) calculation device for reducing variations in MTF evaluation.SOLUTION: An MTF calculation device 1 includes: a derivative area extraction part 31; a first calculation part 32; a determination part 33; and a second calculation part 34. The derivative area extraction part 31 acquires a small area being an image area extracted from a base image and an image area including a reference pixel, and causes the acquired small area to move to the base image in a range where the reference pixel is not projected out of the small area to extract a plurality of derivative areas being a plurality of image areas including the reference pixel from the base image. The first calculation part 32 calculates each edge profile of an image of the small area and an image of each of the plurality of derivative areas, and the determination part 33 determines whether or not each edge profile calculated by the first calculation part 32 is suitable for the calculation of an MFT. The second calculation part 34 calculates the MTF of the edge profile determined to be suitable by the determination part 33.SELECTED DRAWING: Figure 2

Description

  The present invention relates to an evaluation of an optical image acquired by an optical sensor of an optical satellite, and relates to an MTF calculation apparatus and an MTF calculation program for evaluating a modulation transfer function (hereinafter referred to as MTF) of an optical sensor.

  The operator of the ground system corresponding to the optical satellite handles the satellite image acquired from the optical satellite. The image quality of the satellite image is blurred due to the influence of atmospheric scattering due to the path radiance and the influence of defocusing of the optical sensor. The degree of blurring of the image acquired by the optical sensor is evaluated by measuring the MTF of the optical sensor. From the satellite image acquired by the optical sensor, it is possible to improve the blur of the image quality and create a highly visible satellite image by the MTF of the evaluated optical sensor.

JP 2008-39586 A JP 2003-274087 A

  However, in order to evaluate the MTF of the optical sensor, the operator needs knowledge about image processing of an image acquired by the optical sensor (for example, Patent Documents 1 and 2). In order to appropriately evaluate the MTF of the optical sensor, the operator needs to select a small region suitable for the MTF evaluation in the image. Note that the small region suitable for the MTF evaluation is an image region in which the luminance change of the pixel value at the edge is steep and there is little noise in the bright part and the dark part with the edge as a boundary. In order to appropriately evaluate the MTF of the optical sensor, an operator who evaluates needs to perform an operation for evaluating a plurality of scenes and regions, but the MTF of the optical sensor is evaluated by the operator who performs the evaluation. The result of evaluation varies.

  An object of the present invention is to provide an MTF calculation apparatus and an MTF calculation program that reduce variations in MTF evaluation.

An apparatus for calculating a modulation transfer function according to the present invention includes:
A designated area that is an image area extracted from a base image that is a base and that includes a reference pixel serving as a reference is acquired, and the acquired reference area is within a range in which the reference pixel does not go outside the specified area. A derivation area extraction unit that extracts a plurality of included image areas that are a plurality of image areas including the reference pixel from the base image by moving the base image with respect to the base image;
A first calculation unit for calculating each edge profile of the acquired image of the specified area and each of the extracted images of the plurality of included images;
A determination unit that determines whether each edge profile calculated by the first calculation unit is suitable for calculating a modulation transfer function;
A second calculation unit that calculates the modulation transfer function of the edge profile determined to be suitable for the determination unit.

  Since the modulation transfer function calculation apparatus according to the present invention includes the derivation region extraction unit, a derivation region that may be suitable for MTF evaluation can be picked up from the original small region. Therefore, variation in MTF evaluation can be reduced.

FIG. 3 is a diagram of the first embodiment and illustrates the image processing system. FIG. 3 is a diagram of the first embodiment and is a block diagram of an MTF calculation device. FIG. 3 is a flowchart of the operation of the MTF calculation apparatus in the first embodiment. FIG. 5 is a diagram for explaining small region extraction processing in the first embodiment. The figure of Embodiment 1 is a figure explaining derived area extraction processing S31. FIG. 3 is a diagram for explaining a linear direction of an edge in the first embodiment. FIG. 3 is a diagram for explaining the definition of the edge inclination angle in the diagram of the first embodiment. The figure for Embodiment 1 is a figure for demonstrating distance calculation process S32d to an edge. The figure of Embodiment 1 is a figure which shows the edge profile of the small area calculated | required by edge profile calculation process S32. The figure of Embodiment 1, The figure which shows the database 2a.

Embodiment 1 FIG.
<*** Explanation of configuration *****>
The image processing system 3 according to the first embodiment will be described with reference to FIGS.
FIG. 1 is a configuration diagram of an image processing system 3 according to the first embodiment. The image processing system 3 includes a modulation transfer function calculation device 1 (hereinafter referred to as an MTF calculation device 1) and a database device 2 having a database 2a. The MTF calculation apparatus 1 acquires a plurality of base images 70. Further, the MTF calculation device 1 acquires past MTF evaluation results from the database device 2. The MTF calculation apparatus 1 newly calculates the MTF evaluation result of a partial region of the base image using the past MTF evaluation result. The MTF calculation device 1 stores a new MTF evaluation result in the database device 2. This new MTF evaluation result is used to create the next new MTF evaluation result.

  FIG. 2 is a configuration diagram of the MTF calculation apparatus 1. As shown in FIG. 2, the MTF calculation device 1 includes a small region search unit 10, a small region information acquisition unit 20, an MTF evaluation result calculation unit 30, an imaging condition storage unit 40, a storage device 50, and an input / output interface device 60. .

  The MTF calculation apparatus 1 is a computer. As shown in FIG. 2, the MTF calculation device 1 includes hardware such as a processor 91, a storage device 50, and an input / output interface device 60. The storage device 50 stores programs that realize the functions of the small region search unit 10, the small region information acquisition unit 20, the MTF evaluation result calculation unit 30, and the imaging condition storage unit 40. Then, when the processor 91 executes the program, the operations of the small region search unit 10, the small region information acquisition unit 20, the MTF evaluation result calculation unit 30, and the imaging condition storage unit 40 are performed. FIG. 2 schematically illustrates a state in which the processor 91 is executing a program that realizes the functions of the small region search unit 10, the small region information acquisition unit 20, the MTF evaluation result calculation unit 30, and the imaging condition storage unit 40. ing.

(1) The small area search unit 10 executes a process for assisting the operator of the image processing system 3 to select an area suitable for MTF evaluation in an image.
(2) The small area information acquisition unit 20 receives a small area of an image used for MTF evaluation.
(3) The MTF evaluation result calculation unit 30 performs MTF calculation processing on the small region of the image selected by the small region information acquisition unit 20.
(4) The imaging condition storage unit 40 stores the calculation result of the MTF evaluation result calculation unit 30 in the database 2a of the database device 2 for each imaging condition of the image.

  The small area search unit 10 includes a reference image determination unit 10a and a reference image selection unit 10b. The MTF evaluation result calculation unit 30 includes a derived region extraction unit 31, a first calculation unit 32, a determination unit 33, a second calculation unit 34, and an evaluation result calculation unit 35.

<*** Explanation of operation *****>
FIG. 3 is a flowchart showing an MTF calculation method executed by the MTF calculation apparatus 1. The MTF calculation method is realized by an MTF calculation method program executed by the MTF calculation apparatus 1. An MTF calculation method executed by the MTF calculation apparatus 1 will be described with reference to FIG. In FIG. 3, a small region search unit 10, a small region information acquisition unit 20, an MTF evaluation result calculation unit 30, and an imaging condition storage unit 40, respectively, are a small region search process S 10, small region information acquisition process S 20, and MTF evaluation result calculation. Processing S30 and imaging condition storage processing S40 are executed.

  Hereinafter, each process constituting the small area search process S10 will be described. Specifically, the reference image determination unit 10a executes reference image determination processing S10a. The reference image determination process S10a includes a small area extraction process S11, a standard image acquisition process S12, and a reference image extraction process S13. The reference image selection unit 10b executes reference image selection processing S10b. The reference image selection process S10b includes a pixel value normalization process S14, a correlation calculation process S15, and an evaluation recommended area determination process S16.

<Small Region Extraction Processing S11>
In the small area extraction process S11, one base image 70 is acquired. The base image 70 is acquired via the input / output interface device 60 and stored in the storage device 50. In the small area extraction process S <b> 11, one base image 70 is read from the storage device 50.
FIG. 4 is a diagram for explaining the small area 71 extracted from the base image 70 by the small area extraction processing S11. In the small area extraction process S11, the small area 71 used for the MTF evaluation is randomly extracted from the base image 70. At that time, one pixel included in the small area 71 is extracted from the small area 71 as the reference pixel 71p. Any pixel in the small area 71 may be used as the reference pixel 71p. In this embodiment, a pixel located at the center of the small area 71 is set as a reference pixel 71p. Therefore, the reference pixel 71p is also a central pixel of the small area 71. In the present embodiment, the reference pixel 71p has an even number of M pixels arranged in both the vertical and horizontal directions as shown in FIG. 4 when the address of the pixel in the small area 71 is set to 0. In this case, the pixel at the position (0.5M−1, 0.5M−1) in the address may be the reference pixel 71p of the small region 71. FIG. 4 shows a case where addresses are 0 to 5 and M = 6 in both the X direction and the Y direction. When even M pixels are arranged in both the vertical Y direction and the horizontal X direction, assuming that the pixel at the position of (0.5M−1, 0.5M−1) is the reference pixel 71p, as shown in FIG. In addition, the pixel having the center point 71c on the lower right side is the reference pixel 71p.

<Reference Image Acquisition Process S12>
The reference image selection unit 10b uses, as the standard image 82, an image of the database device 2 that stores an image having an edge profile that has been determined to be suitable by the determination unit 33 in the past. That is, in the reference image acquisition process S12, the base image 70 that is an MTF calculation target and the imaging conditions (latitude, pointing angle) are the same, and the reference image 82 that is an image used for the past MTF evaluation is stored in the database device 2. From the database 2a. As shown in FIG. 10 described later, the reference image 82 is associated with imaging conditions (imaging date and time, latitude of the imaging point, and pointing angle at the time of imaging), and the imaging condition of the reference image 82 is referred to by the database device 2. Is possible.

<Reference image extraction processing S13>
In the reference image extraction process S13, the image of the small area 71 extracted in the small area extraction process S11 is cut out from the base image 70 that is the MTF calculation target. If the size of the small area 71 used for the MTF evaluation is Sx × Sy pixels (Sx, Sy are integers of 1 or more), the image of the small area 71 to be cut out has a pixel size of Sx × Sy, and the small area extraction process S11 Including the reference pixel 71p determined in (1). An image obtained by cutting out the small region 71 determined in the small region extraction processing S11 from the base image 70 is referred to as a reference image 72. A plurality of small areas 71 are extracted from one base image 70, but in the following description, it is assumed that N small areas have been extracted. Each small area 71 and the reference image 72 are described like a small area 71 _i and a reference image 72 _i by using a subscript i (i = 1, 2, 3... N). The reference image 72 _i corresponds to the small area 71 _i .

<Pixel value normalization processing S14>
The reference image selection unit 10b calculates a correlation value of each reference image by performing a correlation operation between a plurality of reference images having different image areas in the base image 70 and a standard image serving as a reference, and performs selection for selection. A reference image is selected based on the result of comparison between the selection threshold 94, which is a threshold, and the correlation value r _i of each reference image. In the pixel value normalization process S14, in order to adjust the contrast between the standard image 82 acquired in the standard image acquisition process S12 and the reference image 72 _i cut out in the reference image extraction process S13, the pixel value is 0 or more and 1 or less. Is converted to a numerical value of The pixel value of the extracted reference image 72 _i is
G _i (X, Y) (0 ≦ X ≦ S _x −1, 0 ≦ Y ≦ S _y −1)
And (X, Y) is the address position shown in FIG. 4, and S _x and S _y are the numbers of pixels in the X and Y directions.
In FIG. 4, S _x = S _y = 6. G _i (S _x −1, S _y −1) is a pixel value corresponding to the rightmost pixel in the bottom row. The maximum pixel value of the reference image 72 _i is G _max , and the pixel value after processing is G ′ _i (X, Y), and the reference image is normalized to a numerical value of 0 or more and 1 or less by the following expression 1.
G ′ _i (X, Y) = G _i (X, Y) / G _max (Equation 1)
Pixel value normalization step S14, for also the reference image 82, as in the reference image 72 _i, to implement the normalization of Formula 1. Let the normalized reference image 82 be B (X, Y), and the normalized reference image 72 _i be R _i (X, Y). (X, Y) is an address.

<Correlation calculation processing S15>
The correlation operation S15, calculates the N reference images 72 _i normalized by the pixel value normalization processing S14, the correlation value between one of the reference image 82. First, with respect to the standard image B (X, Y) and the reference image R _i (X, Y) (i = 1, 2,... N) normalized in the pixel value normalization process S14, the respective average pixels Value

Calculate
Next, a correlation value r _i (1 ≦ i ≦ N) between the base image B (X, Y) and each reference image R _i is calculated by the following equation 2.

<Small Region Determination Process S16>
In the small area determination process S16, the small area 71 _i of the reference image 72 _i in which the correlation value r _i calculated in the correlation calculation process S15 is equal to or greater than the threshold r _th that is the selection threshold 94 is used as an area recommended for MTF evaluation. Determine as. That is, among the _N small areas 71 _{i = 1 to} 71 _{i = N} , the K small areas 71 _{k = 1 to} 71 _{k = K} are selected by the reference image selection process S10b. Here, K is the total number of the small areas 71 determined to be _{equal to} or _larger than the threshold value r _th in the small area determining process S16. Therefore, K is N or less.

<Small region information acquisition process S20>
Next, in the small area information acquisition process S20, small area information used for the MTF evaluation is received.
(Example 1) When the small areas 71 _{k = 1 to} 71 _{k = K} are determined in the small area determination processing S16, the operator refers to the K small areas 71 _k and starts from the K small areas 71 _k. Select a small area to be used for MTF evaluation. The small area information acquisition process S20 accepts selection by the operator as small area information. Specifically, in the small area determination process S _< b> 16, images of the small areas 71 _{k = 1 to} 71 _{k = K} are displayed on a display device (not shown) included in the MTF calculation device 1. By selecting an image of the small area 71 _{k = 1 to} 71 _{k = K} displayed by the operator with the mouse, the small area information acquisition process S _< b> 20 receives the selected small area via the input / output interface device 60. .
In (Example 2) a small area information acquisition processing S20, operator is, the determined small region other than the small region 71 _k was, specified in the original base image 70 that subregion 71 _k has been determined, is this designated The small area information acquisition process S20 may accept the small area. Specifically, the small area determination process S16 displays the base image 70, and the operator designates a small area in the base image 70 with the mouse. The small area information acquisition process S <b> 20 receives the small area designated by the operator with the mouse via the input / output interface device 60.
(Example 3) Alternatively, the small region determination process S16 outputs information of the determined small regions 71 _{k = 1 to} 71 _{k = K} , and the small region information acquisition process S20 performs the small region 71 _{k = 1 to} 71 _{k = K.} The small area 71 _{k = 1 to} 71 _{k = K} may be received as the small area.

  Next, the MTF evaluation result calculation process S30 will be described. In the MTF evaluation result calculation process S30, the edge profile is calculated including the derived area shifted from the small area 71 received in the small area information acquisition process S20. Then, a small region suitable for MTF evaluation is extracted from the shape of the calculated edge profile of each small region, and the MTF is calculated.

  The MTF evaluation result calculation process S30 includes a derived area extraction process S31, an edge profile calculation process S32, an edge profile determination process S33, an MTF calculation process S34, an evaluation result calculation process S35, and a determination process S36. The determination process S36 is performed by the evaluation result calculation process S35. The derived area extraction unit 31 executes the derived area extraction process S31. The edge profile calculation process S32 is executed by the first calculation unit 32. The edge profile determination process S33 is executed by the determination unit 33. The second calculation unit 34 executes the MTF calculation process S34. The evaluation result calculation process S35 is executed by the evaluation result calculation unit 35.

  The edge profile calculation process S32 includes a line unit LSF calculation process S32a, a line unit LSF peak position calculation process S32b, an edge inclination angle calculation process S32c, an edge distance calculation process S32d, a distance order sort process S32e, and a resampling process S32f. And determination processing S32g. The edge profile determination process S33 includes a noise determination process S33a and an edge inclination determination process S33b. The MTF calculation process S34 includes an LSF calculation process S34a and an FFT process S34b.

The input data of the MTF evaluation result calculation process S30 is the small area received by the small area information acquisition process S20. In the above (Example 1) In the following, the operator is small region _{71 k = 1} among the small regions _{71 k = 1 ~71 k = K} , were selected two subregions _{71 k = 2.} Therefore, in the following example, it is assumed that the input data of the MTF evaluation result calculation process S30 includes two small areas 71 _{k = 1} and small areas 71 _{k = 2} .

<Derived area extraction processing S31>
The derived area extraction unit 31 acquires a designated area 92 that is an image area extracted from the base image 70 serving as a base and including the reference pixel 71p serving as a reference. In this example, there are two designated areas 92, which are small areas 71 _{k = 1} and small areas 71 _{k = 2} , which are input data. In this case, the derived area extraction unit 31 acquires, as the designation area 92, the small area 71 _{k = 1} and the small area 71 _{k = 2} that are the image areas of the reference image selected by the reference image selection unit 10b. The derived area extraction unit 31 moves the acquired designated area 92 with respect to the base image within a range in which the reference pixel 71p does not go outside the designated area 92, thereby including a plurality of image areas and the reference pixel 71p. A plurality of inclusion image areas 93 which are a plurality of image areas are extracted from the base image 70. Hereinafter, the derivation area extraction processing S31 by the derivation area extraction unit 31 will be described in detail.

FIG. 5 is a diagram illustrating the derivation area extraction process S31 performed by the derivation area extraction unit 31. The derived area extraction unit 31 shifts the designated area 92 pixel by pixel with respect to the reference pixel 71p, and extracts each image area determined by shifting the pixel by pixel as a plurality of included image areas. That is, in the derived area extraction process S31, a “derived area” of the small area 71 _{k = 1} is generated. The edge profile calculation process S32 is executed for the small area 71 _{k = 1} and its derived areas. This illustrative examples represented a small region _{71 k = 1} and the small region _{71 1 (0),} derived region of the small region _{71 1 (0),} derived region _{71 1 (1),} the derived region _{71 1 (2)} .. -Derived area 71 _{1 (m)} will be described. The derived area 71 _{1 (1)} to the derived area 71 _{1 (m)} are a plurality of included image areas 93 of the small area 71 _{k = 1} that is the designated area 92.

As shown in FIG. 5, in the derivation region extraction process S31, the derivation region 71 _{1 (1)} , the derivation region 71 _{1 (2)} ... The derivation region 71 _{1 is performed} by shifting the small region 71 _{1 (0)} pixel by pixel. _(M) is generated. The derived area always includes the reference pixel 71p.

As shown in FIG. 5, a small region _{71 1 (0),} as shown in FIG. 5 (b) (c), the base image 70, shifted small region _{71 1 (0)} by one pixel. In FIG. 5B, the derived area 71 _{1 (1)} is obtained by shifting the small area 71 _{1 (0)} leftward by one pixel, and in FIG. 5C, the small area 71 _{1 (0)} is further left. The derived region 71 _{1 (2)} is obtained by shifting one pixel in the direction.

  The first calculation unit 32 calculates each edge profile of the acquired image of the designated area 92 and each of the extracted images of the plurality of included image areas 93. Specifically, the edge profile calculation process S32 will be described in detail below.

<Line unit LSF calculation process S32a>
The line unit LSF calculation process S32a processes the image of the small area 71 _{1 (0)} . The line unit LSF calculation process S32a will be described with reference to FIG.
FIG. 6 is a diagram for explaining the line direction, that is, the linear direction of the edge. Here, the line is one line in which a plurality of pixels are arranged in the X direction. The following line unit LSF calculation processing S32a to resampling processing S32f are processing for the image of the small region 711 ₍₀₎ . In the line unit LSF calculation process S32a, as shown in FIG. 6, the image of the small region 711 ₍₀₎ in the base image is set to the line direction in the Y direction which is the linear direction of the edge, and the LSF in line units, that is, line units. Calculate (Line Spread Function). In the line unit LSF calculation process S32a, the image g (x, y) of the small area 71 _{1 (0} ) (0 ≦ x ≦ S _x −1, 0 ≦ y ≦ S _y −1).
On the other hand, the LSF _y (x) of the line y is calculated by the following expression 3 using the differential filter f (t) (−1 ≦ t ≦ 1). Here, S _x and S _y are the numbers of pixels in the X direction and the Y direction. The filter coefficient of the differential filter is (−1/2, 0, 1/2).

<Line unit LSF peak position calculation processing S32b>
In the LSF peak position calculation process S32b in line units, the peak position xmax (y) at which LSFy (x) is maximized is calculated for each line.

<Edge inclination angle calculation process S32c>
FIG. 7 is a diagram for explaining the definition of the edge inclination angle. In the edge inclination angle calculation process S32c, the edge inclination angle in the image of the small region 711 ₍₀₎ is calculated. A linear approximation function f (x) = ax + b is used for the line number y (0 ≦ y ≦ Sy−1) and the LSF peak position xmax (y) indicated by an asterisk in FIG. Lead. The calculated intercept b is a period of the edge inclination, and the edge inclination angle θ [rad] is calculated by the following equation 4 in the edge inclination angle calculation process S32c.
θ = arctan (1 / b) (Formula 4)

<Distance calculation processing S32d to edge>
FIG. 8 is a diagram for explaining the distance calculation processing S32d to the edge. In the distance calculation process S32d to the edge, as shown in FIG. 8, the distance δ between each pixel of the image of the small region 711 ₍₀₎ and the primary straight line obtained in the edge inclination angle calculation process S32c is calculated. To do. When the coordinates of the pixel of the image of the small area 71 _{1 (0)} are (x ₀ , y ₀ ) and the primary straight line calculated in the edge tilt angle calculation process S32c is ax + by + c = 0, the distance calculation process S32d to the edge Then, the distance δ (x ₀ , y ₀ ) between each pixel of the small region 71 _{1 (0)} and the primary line is calculated by the following equation (5).

<Distance Sorting Process S32e>
In the sort process S32e in order of distance, the distance δ (x ₀ , y ₀ ) of each pixel calculated in the distance calculation process S32d to the edge is sorted, and the pixel value data of the small area 71 _{1 (0)} is converted into the distance. The data are rearranged into a one-dimensional array in the order of δ (x ₀ , y ₀ ).

<Resampling process S32f>
In the resampling process S32f, the edge profile of the small area 71 _{1 (0)} is resampled by the distance to the edge of the pixels in the small area 71 _{1 (0)} rearranged in the sorting process S32e in the order of distance. Ask. Linear interpolation is used for the resampling process S32f.
FIG. 9 shows the edge profile P ₍₀₎ of the small area 71 _{1 (0)} obtained by the resampling process S32f.

The edge profile P _{(0) of the} small area 71 _{1 (0)} shown in FIG. 9 is calculated by the line unit LSF calculation process S32a to the resampling process S32f for the small area line 71 _{1 (0)} . In the determination process S32g, it is determined whether there is a next derived area. In this case, since there is a derivation area 71 _{1 (1)} , the process returns to the line unit LSF calculation process S32a, and the line unit LSF calculation process S32a to the resampling process S32f are executed for the derivation area 71 _{1 (1)} . Similar to the edge profile P _{(0) in} FIG. 9, the edge profile P ₍₁₎ of the derivation area 71 _{1 (1)} is calculated. Similarly, the edge profile _P of the derived region _{71 1 (2) ~71 1 (} m) (2) ~P (m) is calculated.

<Edge profile determination processing S33>
In the edge profile determination processing S33, an edge profile suitable for the MTF calculation processing S34 is targeted for the “m + 1” edge profiles of the edge profiles P _{(0) to} P _(m) calculated in the edge profile calculation processing S32. Determine if.
The determination unit 33 determines whether each edge profile calculated by the first calculation unit 32 is suitable for MTF calculation. The determination unit 33 acquires a calculation determination value that is a determination value calculated for each edge profile, and compares each acquired calculation determination value with a determination threshold that is a threshold for determination. Is suitable for the calculation of the MTF.
This will be described in detail below.

<Noise determination processing S33a>
The edge profile P ₍₀₎ will be described as an example. In the noise determination process S33a, a dark part and a bright part of the image are identified with respect to the edge profile P _{(0) of} the small region 71 _{1 (0)} . In the edge profile P _{(0) of} FIG. 9, a range 97 indicated by a broken line corresponds to the dark part of the image of the small region 71 _{1 (0)} , and a range 98 indicated by the broken line corresponds to the dark part of the image of the small region 71 _{1 (0).} Correspond. In the noise determination process S33a, it is determined whether or not there is little noise in the range 97 and the range 98. A specific noise determination method is a standard for a certain number of pixels (the number of data of the profile P ₍₀₎ × the ratio α) with respect to the range 97 as the head and the range 98 as the end of the edge profile P _(0). Calculate the deviation. The standard deviation of the range 97 and the range 98 is a calculation determination value. The ratio α is set in advance. In the noise determination process S33a, if both the standard deviation of the bright part and the dark part are smaller than the determination threshold, the noise is small and can be used for MTF evaluation, that is, it is determined that the process should proceed to the next process. It is determined that it cannot be used for evaluation, that is, does not proceed to the next processing. The above processing contents are similarly executed for the edge profiles P _{(1) to} P _(m) of the derivation area 71 _{1 (1)} to the derivation area 71 _{1 (m)} .

<Edge slope determination processing S33b>
In the edge inclination determination process S33b, if the period of the edge inclination angle of the edge profile P ₍₀₎ of the small region 71 _{1 (0)} calculated in the edge profile calculation process S32 is within the threshold value range, the edge profile P _{(0 )} Is determined to be suitable for MTF evaluation. The same applies to the edge profiles P _{1 to} P _m of the derived areas 71 _{1 (1)} to 71 _{1 (m)} . In this case, the period of the edge inclination angle is a calculated determination value, and a threshold value for the determination is a determination threshold value. Of course, the threshold value of the edge inclination angle period and the threshold value of the standard deviation are different.

In this example, it is determined that both the noise determination process S33a and the edge inclination determination process S33b are suitable for the MTF evaluation among the “m + 1” edge profiles P _{(0) to} P _(m) . Three edge profiles P ₍₀₎ , P ₍₁₎ , and P _{(2) are} assumed.

<MTF calculation process S34>
Next, the MTF calculation process S34 will be described. The second calculation unit 34 calculates the MTF of the edge profile that is determined to be suitable by the determination unit 33. In the MTF calculation process S34 by the second calculator 34, the MTF is calculated for the edge profile determined to be suitable for the MTF evaluation in both the noise determination process S33a and the edge inclination determination process S33b.

<LSF calculation process S34a>
Hereinafter, the edge profile P ₍₀₎ will be described as an example. The following P ₍₀₎ (x) is P ₍₀₎ . In the LSF calculation process S34a, the differential filter f (t) (-1 ≦ t ≦ 1) is used for the edge profile P ₍₀₎ (x) (0 ≦ x ≦ Np−1), and the LSF is expressed by the following equation (6). ₍₀₎ to calculate the _{L (0) (t) (} 0 ≦ t ≦ Np-1) of. Here, 0 ≦ x ≦ Np−1 means the number of data points of the edge profile P ₍₀₎ (x). The filter coefficient of the differential filter is (−1/2, 0, 1/2). Thus using Equation 6 to calculate the _{L (0) (t)} of the edge profile _{P (0),} obtaining the _{LSF (0)} of the edge profile _{P (0).}

ＬＳＦ計算処理Ｓ３４ａでは、エッジプロファイルＰ_（０）と同様に、エッジプロファイルＰ_（１）、Ｐ_（２）についても、ＬＳＦ_（１）及びＬＳＦ_（２）を求める。

In the LSF calculation process S34a, LSF ₍₁₎ and LSF ₍₂₎ are obtained for the edge profiles P ₍₁₎ and P ₍₂₎ as well as the edge profile P ₍₀₎ .

<FFT processing S34b>
The FFT processing S34b, each edge profile _{P (0), P (1} ), performs _{LSF (0) ~LSF (2)} against FFT (Fast Fourier Transform) processing of the _{P (2),} be normalized with the DC component The MTF is calculated. Hereinafter, the edge profile P ₍₀₎ will be described as an example. The data after the FFT processing is F ₍₀₎ (x) (0 ≦ x ≦ Nf−1, Nf is the number of data points after the FFT processing), and MTF ₍₀₎ (x) (0 ≦ x ≦ Nf−1) is calculated.
MTF ₍₀₎ (x) = F ₍₀₎ (x) / F ₍₀₎ (0) (Formula 7)
F ₍₁₎ (x), F ₍₂₎ (x) and MTF ₍₁₎ (x), MTF ₍₂₎ (x) are calculated for the edge profiles P ₍₁₎ and P ₍₂₎ , respectively. Is done.
Summarizing the above relationships,
Small region 71 _{1 (0)} → P ₍₀₎ → L ₍₀₎ → LSF ₍₀₎ → F ₍₀₎ (x) → MTF ₍₀₎ (x),
Derived region 71 _{1 (1)} → P ₍₁₎ → L ₍₁₎ → LSF ₍₁₎ → F ₍₁₎ (x) → MTF ₍₁₎ (x),
Derived region 71 _{1 (2)} → P ₍₂₎ → L ₍₂₎ → LSF ₍₂₎ → F ₍₂₎ (x) → MTF ₍₂₎ (x),
It becomes.

<Evaluation result calculation process S35>
In the evaluation result calculation process S35, the average of the MTFs calculated in the MTF calculation process S34 is calculated and calculated as the evaluation result. In this example,
MTF average = [MTF ₍₀₎ (x) + MTF ₍₁₎ (x) + MTF ₍₂₎ (x)] / 3 = Evaluation result.

<Determination process S36>
In the determination process S36, it is determined whether there is a next small area. In this example, there is another small area 71 _{k = 2} as input data of the MTF evaluation result calculation process S30. Therefore, the process returns to the derived area extraction process S31, and the edge profile calculation process S32, the edge profile determination process S33, the MTF calculation process S34, and the evaluation for the small area 71 _{k = 2} as in the case of the small area 71 _{k = 1.} The result calculation process S35 is executed. When the evaluation result calculation process S35 for the small area 71 _{k = 2} is completed, there is no next small area, and the process proceeds to the imaging condition storage process S40.

<Imaging Condition Saving Process S40>
In the imaging condition storage process S40 by the storage unit 40, an image having an edge profile determined by the determination unit 33 to be suitable for MTF calculation is stored in the database 2a of the database device 2. This will be described in detail below.
FIG. 10 is a diagram of the database 2 a included in the database device 2. In the imaging condition storage process S40, the MTF evaluation result is stored in the database 2a. At this time, together with the evaluation result 2a-1, the imaging condition 2a-2 corresponding to the evaluation result is extracted from the base image, and the imaging condition 2a- 2 and “image 2a-3 of the small area and derived area used for the evaluation result” are registered in the database 2a. The evaluation result is a unit of one record. In the database 2a, the MTF average value that is the evaluation result 2a-1 and the “small area used for evaluation, derivation” for each “imaging date and time, latitude of the shooting point, and pointing angle at the time of shooting” that is the imaging condition 2a-2 Region image 2a-3 "is associated and registered. For this example,
MTF average = [MTF ₍₀₎ (x) + MTF ₍₁₎ (x) + MTF ₍₂₎ (x)] / 3 = Records of evaluation results are as shown in (1) to (3) below.
(1) Since the base image is a base image in which the small area 71 _{1 (0)} , the derived area 71 _{1 (1)} , and the derived area 71 _{1 (2)} are extracted, the imaging condition 2a-2 is obtained from this base image. Extracted and saved.
(2) As the evaluation result 2a-1, the average value of the above MTF ₍₀₎ (x) to MTF ₍₂₎ (x) is stored.
(3) The “small region used for evaluation, derived region image 2a-3” corresponding to the evaluation result 2a-1 includes a small region 71 _{1 (0)} , a derived region 71 _{1 (1)} , and a derived region 71 _{1 ( 2)} Each image.
In the imaging condition storage process S40, these are registered as one record in the database 2a of FIG.

<*** Explanation of the effect of the embodiment *****>
(1) Since the MTF calculation apparatus according to the present embodiment includes the derived region extraction unit 31, a plurality of derived regions that are highly likely to be suitable for MTF evaluation can be extracted from an image.
(2) Since the MTF calculation apparatus according to the present embodiment includes the determination unit 33, an area that is highly likely to be suitable for MTF evaluation is selected from the acquired small area and the derived area extracted by the derived area extraction unit 31. Can be selected more reliably.
(3) Since the derivation region extraction unit 31 extracts the derivation region by shifting the small region by one pixel from the reference pixel, it is possible to select a region that is highly likely to be suitable for the MTF evaluation by simple processing.
(4) Since the reference image selection unit 10b determines a small region corresponding to the reference image by the correlation calculation process between the reference image and the plurality of reference images, the manual operation of the operator is not required, and the reference image selection unit 10b can be suitable for the MTF evaluation. A highly functional small area can be selected.
(5) Since the reference image selection unit 10b uses the image of the database apparatus 2 that stores an image having an edge profile determined in the past as being suitable by the determination unit 33 as the reference image, the MTF evaluation is performed. It is possible to select a small region that is highly likely to be suitable for
(6) The storage unit 40 stores, in the database device 2, an image having an edge profile that the determination unit 33 determines to be suitable for MTF calculation, and the reference image selection unit 10b stores the image as a reference image for correlation processing. Therefore, it is possible to select a small region that is highly likely to be suitable for MTF evaluation.
(7) In the MTF calculation apparatus of this embodiment, a region suitable for evaluation can be automatically extracted by the small region search processing S10 and the edge profile determination processing S33. Therefore, the operator who performs the evaluation of the MTF does not need to have knowledge regarding the image processing of the optical sensor, and can perform the evaluation. Moreover, since evaluation candidates are extracted by automatic processing, there is no variation among operators.
(8) In the MTF calculation apparatus according to the present embodiment, the determination of the small area, the extraction of the derived area, and the MTF measurement operation are automated, so that the operation of the operator is simplified.
(9) Since the result is stored in the database device 2b in the MTF calculation device of the present embodiment, the evaluated MTF can be managed in time series. For this reason, it is possible to confirm the secular change of the MTF of the sensor.
(10) In the MTF calculation device 1, the imaging condition storage unit 40 extracts the imaging condition 2a-2 from the base image, and the imaging condition 2a-2 is obtained from the evaluation result 2a-1 and the “small region used for evaluation, derived region”. Since it is stored in the database device 2 in association with the image 2a-3 ", management based on the imaging condition 2a-2 becomes possible.
(11) Further, “the image 2a-3 of the small region and the derived region used for the evaluation” can be used as the standard image of the reference image selection process S10b. That is, since a reference image having the same imaging conditions as the reference image acquired in the reference image determination process S10a can be acquired from the database 2a, it is possible to select a reference image that is highly likely to be determined to be suitable in the edge profile determination process S33. Will increase.

  Although the image processing system 3 includes the database device 2, the storage device 50 of the MTF calculation device 1 may be configured to perform the function of the database device 2.

  The MTF calculation apparatus and the image processing system according to the present embodiment are assumed to evaluate an optical image acquired by an optical sensor of an optical satellite. However, it is apparent from the above description that the present invention can be applied not only to an optical sensor of an optical satellite but also to an optical image acquired by another optical sensor.

<*** Additional explanation of hardware configuration *****>
Finally, a supplementary description of the hardware configuration of the MTF calculation apparatus 1 will be given. The processor 91 shown in FIG. 2 is an IC (Integrated Circuit) that performs processing. The processor 91 is a CPU (Central Processing Unit), a DSP (Digital Signal Processor), or the like. The storage device 50 shown in FIG. 2 is a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an HDD (Hard Disk Drive), or the like.

  The storage device 50 also stores an OS (Operating System). At least a part of the OS is executed by the processor 91. The processor 91 executes a program that realizes the functions of the small region search unit 10, the small region information acquisition unit 20, the MTF evaluation result calculation unit 30, and the imaging condition storage unit 40 while executing at least a part of the OS. Although one processor is illustrated in FIG. 2, the MTF calculation apparatus 1 may include a plurality of processors. In addition, information, data, signal values, and variable values indicating processing results of the small region search unit 10, the small region information acquisition unit 20, the MTF evaluation result calculation unit 30, and the imaging condition storage unit 40 are stored in the storage device 50 or the processor. 91 is stored in a register or cache memory. The programs for realizing the functions of the small region search unit 10, small region information acquisition unit 20, MTF evaluation result calculation unit 30, and imaging condition storage unit 40 are magnetic disk, flexible disk, optical disk, compact disk, Blu-ray (registered trademark). ) It may be stored in a portable storage medium such as a disk or DVD.

  In addition, each “part” of the small area search unit 10, the small area information acquisition unit 20, the MTF evaluation result calculation unit 30, and the imaging condition storage unit 40 is changed to “circuit” or “process” or “procedure” or “processing”. May be read as The “circuit” is a concept including not only the processor 91 but also other types of processing circuits such as a logic IC or GA (Gate Array), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array). .

  S10 small region search processing, S10a reference image determination processing, S11 small region extraction processing, S12 reference image acquisition processing, S13 reference image extraction processing, S10b reference image selection processing, S14 pixel value normalization processing, S15 correlation calculation processing, S16 small Area determination process, S20 small area information acquisition process, S30 MTF evaluation result calculation process, S31 derived area extraction process, S32 edge profile calculation process, S32a line unit LSF calculation process, S32b line unit LSF peak position calculation process, S32c edge slope Corner calculation processing, S32d Distance calculation processing to edge, S32e Sort processing in order of distance, S32f Resampling processing, S32g determination processing, S33 Edge profile determination processing, S33a Noise determination processing, S33b Edge inclination determination processing, S34 TF calculation unit, S34a LSF calculation process, S34b FFT process, S35 evaluation result calculation process, S36 determination process, S40 imaging condition storage process, 1 MTF calculation device, 2 database device, 3 image processing system, 4 base image, 10 small region Search unit, 10a reference image determination unit, 10b reference image selection unit, 20 small region information acquisition unit, 30 MTF evaluation result calculation unit, 32 first calculation unit, 33 determination unit, 34 second calculation unit, 35 evaluation result calculation unit , 40 storage unit, 50 storage device, 60 input / output interface device, 70 base image, 71 small area, 71p reference pixel, 71c center point, 72 reference image, 82 reference image, 91 processor, 92 designated area, 93 inclusion image area 94 selection threshold, 95 calculation determination value, 96 determination threshold, 97, 98 range.

Claims (7)

A designated area that is an image area extracted from a base image that is a base and that includes a reference pixel serving as a reference is acquired, and the acquired reference area is within a range in which the reference pixel does not go outside the specified area. A derivation area extraction unit that extracts a plurality of included image areas that are a plurality of image areas including the reference pixel from the base image by moving the base image with respect to the base image;
A first calculation unit for calculating each edge profile of the acquired image of the specified area and each of the extracted images of the plurality of included images;
A determination unit that determines whether each edge profile calculated by the first calculation unit is suitable for calculating a modulation transfer function;
A modulation transfer function calculation apparatus comprising: a second calculation unit that calculates the modulation transfer function of the edge profile determined to be suitable for the determination unit. The determination unit
By obtaining a calculation determination value that is a determination value calculated for each edge profile, and comparing each acquired calculation determination value with a determination threshold that is a threshold for determination, each edge profile is converted into the modulation transfer function. The modulation transfer function calculation apparatus according to claim 1, wherein the modulation transfer function is determined to be suitable for calculation. The derived area extraction unit
3. The modulation transfer function calculation apparatus according to claim 2, wherein the designated area is shifted pixel by pixel with respect to the reference pixel, and each image area determined by shifting the specified area is extracted as the plurality of inclusion image areas. The modulation transfer function calculation device further comprises:
A correlation value of each reference image is calculated by performing a correlation operation between a plurality of reference images having different image areas in the base image and a standard image serving as a reference, and a selection threshold that is a threshold for selection and each A reference image selection unit that selects the reference image according to a result of comparison with the correlation value of the reference image;
The derived area extraction unit
The modulation transfer function calculation device according to claim 1, wherein an image region of the reference image selected by the reference image selection unit is acquired as the designated region. The reference image selection unit
5. The modulation transfer function calculation device according to claim 4, wherein the image of a database device storing an image having the edge profile determined in the past as being suitable by the determination unit is used as the reference image. The modulation transfer function calculation device further comprises:
The modulation transfer function calculation device according to claim 5, further comprising: a storage unit that stores, in the database device, an image having the edge profile determined by the determination unit to be suitable for calculating the modulation transfer function. On the computer,
A designated area that is an image area extracted from a base image that is a base and that includes a reference pixel serving as a reference is acquired, and the acquired reference area is within a range in which the reference pixel does not go outside the specified area. An extraction process for extracting a plurality of included image areas, which are a plurality of image areas including the reference pixel, from the base image by moving the base image with respect to the base image.
A first calculation process for calculating each edge profile between the image of the specified area acquired in the extraction process and each of the extracted images of the plurality of included images;
A determination process for determining whether each edge profile calculated in the first calculation process is suitable for calculating a modulation transfer function;
A second calculation process for calculating the modulation transfer function of the edge profile determined to be suitable in the determination process;
Modulation transfer function calculation program to execute

Images