MTF measuring device and its program

Abstract

To provide a MTF measurement device that measures highly accurate MTF.SOLUTION: A MTF measurement device 1 comprises ROI image extraction means 13 that extracts a ROI image from a chart image that an imaging system 2 images a chart CH, edge inclination detection means 14 that detects the inclination of an edge from the ROI image, bin width calculation means 150 that detects a bin width on a project axis to which a pixel of the ROI image is projected, projection position mapping means 151 that maps a position of a pixel in the ROI image with a position of a bin in the projection axis perpendicular to the edge, edge profile generation means 17 that projects a pixel value for each pixel position of the ROI image to a corresponding bin, to generate an edge profile, and frequency characteristic calculation means 18 that calculates a MTF from the edge profile.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an MTF measuring device and a program for measuring an MTF (Modulation Transfer Function) that indicates the spatial frequency characteristics of an imaging system. [Background technology]

[0002] Conventionally, a method of measuring MTF by analyzing edge response (edge ​​method) has been widely used to evaluate the resolution characteristics of an imaging system such as a digital camera. The edge method calculates the MTF by analyzing the edge gradient of a one-dimensional edge spread function (ESF) super-resolved from a region of interest (ROI) in an image where two-tone edges are acquired, for sampling positions of a horizontal and vertical square grid of the captured image to be measured.

[0003] MTF is an index that indicates the fineness of an image, and is the modulation depth for a sine wave expressed as a function of spatial frequency. The unit of spatial frequency is cycles / pixel, where 0.5 cycles / pixel represents the Nyquist frequency and 1 cycles / pixel represents the sampling frequency. Note that the "pixel" in this unit is not the physical pixel pitch of the actual image sensor, but the sampling interval of the horizontal and vertical square grid defined by the image format of the output signal.

[0004] ISO12233 specifies a method for measuring the resolution characteristics of a digital camera (see Non-Patent Document 1). Hereinafter, the ISO12233 method will be referred to as the ISO method. The ISO method captures a slightly tilted edge close to the vertical direction or close to the horizontal direction, and selects a region of interest (ROI) that includes the edge of the part to measure MTF. If the edge is close to the vertical direction, the ISO method projects (bins) the pixels of the ROI in a direction parallel to the edge onto a bin array arranged on the horizontal projection axis. The bin width of this projection axis is an oversampled width for one pixel, and is a subpixel width such as 1 / 4 or 1 / 8 of the width of one pixel of the ROI.

[0005] The ISO method calculates an edge spread function (ESF) by averaging the pixel values ​​projected onto the bins on the projection axis for each bin, and calculates a line spread function (LSF) by differentiating the edge spread function. The ISO method then performs a discrete Fourier transform on the edge spread function to take the absolute value, and normalizes it to direct current (spatial frequency "0") to calculate the MTF. In the ISO method, if the edge is close to a horizontal edge, the ROI is rotated by 90 degrees and the MTF is calculated using the same algorithm as for edges close to a vertical edge.

[0006] In this ISO method, when the edge gradient becomes large, the edge gradient becomes shallower in the ESF obtained with the original bin width of 1 / 4, 1 / 8, etc., sub-pixel width, so the MTF is underestimated. Therefore, the ISO method uses a small angle (5° as the standard) as the edge gradient. To solve this problem, a method of scaling the spatial frequency after calculating the MTF (Non-Patent Document 2) and a method of finding the spatial frequency in a direction perpendicular to the edge (Patent Document 1) have been proposed.

[0007] Here, the relationship between the ROI and the projection axis in conventional MTF measurement will be described with reference to FIG. 9. In this case, in order to make the edge inclination θ easier to understand, it is set to 10° instead of the standard 5°, and in order to make the pixels easier to understand, the number of pixels is reduced and the ROI is configured with 4 pixels horizontally and 6 pixels vertically. In addition, here, the oversampling ratio (n bin ) is set to 4, and the bin width of the projection axis is set to 1 / 4 pixel.

[0008] Fig. 9(a) shows the relationship between the ROI and the projection axis in the ISO method. As shown in Fig. 9(a), in the ISO method, pixels are projected in a direction parallel to the edge, tilted toward the projection axis. Fig. 9(b) shows the relationship between the ROI and the projection axis, which is equivalent to the method of scaling the spatial frequency after MTF calculation described in Non-Patent Document 2. In the method described in Non-Patent Document 2, pixels are projected in a direction parallel to the edge onto a projection axis that is perpendicular to the edge direction and has a bin width of cosθ / 4 pixels. This is equivalent to dividing the calculated MTF spatial frequency value by cosθ. Hereinafter, the method described in Non-Patent Document 2 will be referred to as the ISO-cosine method. Fig. 9(c) shows the relationship between the ROI and the projection axis in the method of calculating the spatial frequency in the direction perpendicular to the edge described in Patent Document 1. In the method described in Patent Document 1, pixels are projected in a direction parallel to the edge onto a projection axis perpendicular to the edge direction and with a bin width of 1 / 4 pixel. Hereinafter, the method described in Patent Document 1 will be referred to as the OMNI method. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Patent Publication No. 2021-96260 [Non-patent literature]

[0010] [Non-Patent Document 1] ISO 12233:2017,“Photography - Electronic still picture imaging - Resolution and spatial frequency responses” [Non-Patent Document 2] JKM Roland, “A study of slanted-edge MTF stability and repeatability,” Proc. SPIE, vol. 9396, pp. 93960L, Feb. 2015, 10.1117 / 12.2077755. Summary of the Invention [Problem to be solved by the invention]

[0011] The conventional OMNI method uses a bin width of 1 / n bin Since the edge slope θ is fixed, there is no need to scale the spatial frequency according to the edge slope θ. However, a periodic shift occurs in the phase projected onto the projection axis. Furthermore, in the conventional ISO-cosine method, a periodic shift occurs in the phase projected onto the projection axis when the edge inclination θ is within a certain angle.

[0012] For example, as shown in FIG. 10(a), in the ISO-cosine method, when tan θ=1 / 4 for an edge gradient θ, pixels are projected onto a projection axis with a bin width of cos θ / 4 pixels without any periodic phase shift. However, as shown in FIG. 10(b), in the ISO-cosine method, when tan θ=1 / 3 for an edge gradient θ, a periodic shift occurs in the phase of pixels projected onto a projection axis with a bin width of cos θ / 4 pixels.

[0013] On the other hand, as shown in FIG. 10(c), in the OMNI method, when tan θ=1 / 4 for an edge gradient θ, pixels are projected onto a projection axis with a bin width of 1 / 4 pixel, but a periodic phase shift occurs. As described above, in the conventional method, a periodic phase shift occurs when projecting pixels onto the projection axis, which can cause large errors in the MTF results. The OMNI method improves accuracy by projecting pixels onto multiple projection axes with different phases and taking the MTF that is closest to the average as the measurement result, but a method with even higher accuracy was desired.

[0014] In view of the problems with the conventional methods and the demands for the conventional methods, an object of the present invention is to provide an MTF measurement device and a program therefor that are capable of measuring MTF with high accuracy and precision by eliminating periodic phase shifts when projecting pixels onto a projection axis. [Means for solving the problem]

[0015] In order to solve the above-mentioned problems, the MTF measurement device of the present invention is an MTF measurement device that measures the MTF, which represents the spatial frequency characteristics of an imaging system, by using a chart with different contrast at the boundaries, and is configured to include an ROI image extraction means, an edge gradient detection means, a bin width calculation means, a projection position matching means, an edge profile generation means, and a frequency characteristic calculation means.

[0016] In such a configuration, the MTF measurement device uses the ROI image extraction means to extract an ROI image, which is an image including a boundary, from a chart image captured by the imaging system. Then, the MTF measurement device detects edges from the ROI image using a Sobel filter or the like, and calculates the edge gradient using a Hough transform or the like, by the edge gradient detection means. Then, the MTF measurement device uses the bin width calculation means to calculate an integer multiple of 1 / sinθ or 1 / integer fraction of 1 / sinθ as a variable oversampling ratio when the gradient of the edge is θ, and calculates the reciprocal of the variable oversampling ratio as the bin width of the projection axis onto which the pixels of the ROI image are projected. For example, the MTF measurement device calculates a predetermined oversampling ratio by using a bin width calculation means. bin, the slope of the edge is θ, round(x) is the integer closest to x, and log2(v bin )=log2(n bin )-log2(sinθ tri )+round(log2(sinθ tri ), where θ tri Variable oversampling ratio v = arccos(cos(4θ)) / 4 bin The inverse of is calculated as the bin width of the projection axis onto which the pixels of the ROI image are projected. This makes it possible to eliminate periodic phase deviations since the projection interval of pixels in any row of the ROI image coincides with the projection interval of pixels in any column. Then, the MTF measurement apparatus uses the projection position matching means to match pixel positions of the ROI image with bin positions on the projection axis perpendicular to the edge.

[0017] Then, the MTF measurement device generates an edge profile by averaging each pixel value of the ROI image for each bin associated with the pixel position by the projection position association means, using the edge profile generation means. In this way, by projecting the pixel values ​​of the ROI image onto oversampled bins that are smaller than the pixel width, an oversampled edge spread function can be obtained. The MTF measuring device then uses the frequency characteristic calculation means to differentiate the edge profile to obtain a line spread function, and performs a discrete Fourier transform on the line spread function to calculate the MTF.

[0018] The MTF measurement device can be operated by an MTF measurement program that causes a computer to function as each of the above-mentioned means. Effect of the Invention

[0019] The present invention provides the following excellent effects. According to the present invention, it is possible to eliminate periodic phase deviations that occur when projecting pixels of an ROI image onto a projection axis. This allows the present invention to measure MTF with high accuracy and precision. [Brief description of the drawings]

[0020] [Figure 1] 1 is a block diagram showing the configuration of an MTF measurement device according to an embodiment of the present invention. [Diagram 2] FIG. 1 is an explanatory diagram for explaining ROIs in a captured image, where (a) is a vertical ROI (vertical edge image) and (b) is a horizontal ROI (horizontal edge image). [Diagram 3] FIG. 11 is an explanatory diagram for explaining an edge inclination. [Figure 4] FIG. 13 is an explanatory diagram for explaining the bin width of the projection axis using variable oversampling. [Diagram 5] FIG. 11 is an explanatory diagram for explaining the bin width of the projection axis according to the edge gradient, in which (a) shows the projection interval of an arbitrary column and the projection interval of an arbitrary row of an ROI image, (b) shows the projection interval when tan θ=1 / 4 at edge gradient θ, and (c) shows the projection interval when tan θ=1 / 3 at edge gradient θ. [Figure 6] FIG. 11 is an explanatory diagram for explaining a method for calculating a variable oversampling ratio, where (a) shows the relationship between the edge gradient θ and the angle θtri expressed by a triangular wave function, and (b) shows the relationship between the angle θtri and the variable oversampling ratio. [Figure 7] FIG. 11 is a graph showing the measurement results of an MTF measurement device according to an embodiment of the present invention (OMNI-sine method) and a conventional method (ISO-cosine method), in which (a) shows the measurement results when tan θ=1 / 4 for edge gradient θ, and (b) shows the measurement results when tan θ=1 / 3 for edge gradient θ. [Figure 8] 4 is a flowchart showing the operation of the MTF measurement device according to the embodiment of the present invention. [Figure 9] FIG. 1 is an explanatory diagram for explaining conventional methods, where (a) is a diagram for explaining the ISO 12233 method, (b) is a diagram for explaining a method for scaling spatial frequencies, and (c) is a diagram for explaining a method for calculating spatial frequencies in a direction perpendicular to an edge. [Figure 10] 1A and 1B are diagrams for explaining problems with conventional methods, where (a) and (b) are diagrams for explaining problems with a method for scaling spatial frequencies, and (c) is a diagram for explaining problems with a method for calculating spatial frequencies perpendicular to edges. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. [Configuration of MTF measurement device] First, with reference to FIG. 1, the configuration of an MTF measurement device 1 according to an embodiment of the present invention will be described.

[0022] The MTF measurement device 1 uses a chart CH with different contrast at the boundaries to measure the MTF that represents the spatial frequency characteristic of an imaging system 2. When performing measurements using the MTF measurement device 1, the imaging system 2 and a display device 3 are connected and used.

[0023] The imaging system 2 is a video camera or still camera whose MTF is to be measured, a camera including a lens whose MTF is to be measured, etc. The object to be measured may also be an image whose MTF changes depending on the DETAIL control of the camera. The imaging system 2 captures an image of the chart CH and outputs the image to the MTF measurement device 1.

[0024] The chart CH is a chart with different contrast at the boundary, and is a chart used in the slanted-edge method having a black and white pattern in which the boundary is a straight line tilted at a predetermined angle from the vertical or horizontal direction with respect to the imaging element (not shown) of the imaging system 2. The imaging system 2 captures an image with the black and white boundary of the chart CH tilted at a predetermined angle (about several degrees).

[0025] Here, when measuring frequency characteristics in the horizontal direction, the imaging system 2 images the chart CH with the boundary line oriented diagonally vertically as shown in Fig. 2(a). When measuring frequency characteristics in the vertical direction, the imaging system 2 images the chart CH with the boundary line oriented diagonally horizontally as shown in Fig. 2(b). The boundary line does not have to be located in the center of the chart CH, but may be located in the imaging area where the MTF of the imaging system 2 is to be measured, for example, in the center right (left, top, bottom) part of the chart CH, the upper right (left) diagonal part, the lower right (left) diagonal part, etc.

[0026] The display device 3 provides a user interface for operating the MTF measurement device 1, and displays the chart image captured by the imaging system 2, graphs representing the measurement results, etc. For example, the display device 3 is a liquid crystal display, an organic EL display, etc. The display device 3 may be provided separately from a display device that displays the chart image captured by the imaging system 2 and a display device that displays a graph or the like that is the measurement result.

[0027] Hereinafter, a detailed description will be given of the configuration of the MTF measurement device 1 that measures the MTF of the imaging system 2 using an image captured by the imaging system 2. As shown in FIG. 1, the MTF measurement device 1 includes a chart image storage means 10, an ROI setting means 11, a measurement instruction means 12, an ROI image extraction means 13, an edge gradient detection means 14, a projection information generation means 15, a projection information storage means 16, an edge profile generation means 17, a frequency characteristic calculation means 18, and a measurement result output means 19.

[0028] The chart image storage means 10 stores an image (chart image) of the chart CH for MTF measurement captured by the imaging system 2. The chart image storage means 10 receives the chart image from the imaging system 2 via a video input means (not shown) and stores the input chart image. When a new chart image is input from the imaging system 2, the chart image storage means 10 overwrites the already stored chart image with the new chart image. The chart image storage means 10 is, for example, a general storage device such as a hard disk or a memory. The chart image stored in the chart image storage means 10 is output to the display device 3 via a video output means (not shown), and is also read out by the ROI image extraction means 13.

[0029] The ROI setting means 11 sets a region of interest (ROI) including a boundary (edge) within a chart image captured by the imaging system 2 . When a closed region of any shape is specified by the measurer within the chart image displayed on the display device 3, the ROI setting means 11 sets the closed region as the ROI.

[0030] When measuring frequency characteristics in the horizontal direction, the ROI setting means 11 sets the ROI by having the operator specify a vertically long rectangular area from a chart image captured from a chart CH, as shown in Fig. 2(a). In this case, the ROI specifies an ROI image (vertical edge image) R whose brightness differs in the horizontal direction. When measuring the frequency characteristics in the vertical direction, the ROI setting means 11 sets the ROI by having the operator specify a horizontally long rectangular area from the chart image captured by the chart CH as shown in Fig. 2(b). In this case, the ROI specifies an ROI image (horizontal edge image) R whose brightness differs in the vertical direction.

[0031] If there is no need to change the ROI in the chart images that are input sequentially, the ROI setting by the ROI setting means 11 may be set once. Also, the shape of the ROI may be any shape, such as a quadrangle (rectangle, trapezoid, diamond, etc.), triangle, circle, ellipse, etc., as long as it is set to include only one edge within the region. The ROI setting means 11 outputs ROI information indicating the area (position, size, etc.) of the set ROI to the ROI image extraction means 13.

[0032] The measurement instruction means 12 accepts measurement instructions to start and end MTF measurement. For example, the measurement instruction means 12 may accept instructions using a switch (not shown), or may display an instruction screen on the display device 3 and accept instructions by operating a mouse or the like. The measurement instruction means 12 accepts an operation mode together with an instruction to start MTF measurement.

[0033] The operation mode specifies the operation of the MTF measurement device 1 when performing MTF measurement. Here, the MTF measurement device 1 operates in a normal operation mode and a high-speed operation mode. The normal operation mode is a mode in which detection of edge gradients in an ROI image (hereinafter referred to as ROI image), generation of projection information, generation of an edge profile, and calculation of frequency characteristics (MTF) are performed in sequence. The high-speed operation mode is a mode in which the detection of edge inclination and the generation of projection information are omitted by referring to projection information previously generated in the normal operation mode. The projection information is information that indicates the correspondence between each pixel position of the ROI image and the bin position of the projection axis in the edge method.

[0034] For example, if the imaging system 2 is a camera that inputs still images, such as a still camera, the MTF measuring device 1 is specified in normal operation mode and measures the MTF for one frame image for the image (still image) input from the imaging system 2. Also, for example, if the imaging system 2 is a camera that inputs moving images, such as a video camera, the MTF measuring device 1 is specified in normal operation mode and sequentially measures the MTF for each frame image of the images (moving images) input from the imaging system 2. When inputting a moving image in the normal operation mode, depending on the CPU power, it may not be possible to measure the MTF for all input frame images. In this case, the MTF measurement device 1 operates mainly for the purpose of generating projection information.

[0035] Furthermore, when the high-speed operation mode is specified, the MTF measurement device 1 sequentially measures the MTF for each frame image for the image (moving image) input from the imaging system 2. At this time, the MTF measurement device 1 measures the MTF at high speed by referring to projection information generated in advance in the normal operation mode. However, the high-speed operation mode is premised on the fact that the edge positions are the same as those in the normal operation mode that is performed in advance, which allows the MTF measurement device 1 to omit the detection of edge inclinations and the generation of projection information in the high-speed operation mode.

[0036] It should be noted that, in the normal operation mode, MTF measurement device 1 does not need to have separate operation modes as long as it has CPU power capable of detecting edges and generating projection information for each frame image and measuring the MTF. The measurement instruction means 12 outputs measurement start and end instructions to the ROI image extraction means 13. The measurement instruction means 12 also outputs a specified operation mode to the ROI image extraction means 13 and the edge profile generation means 17.

[0037] The ROI image extraction means 13 extracts an ROI image, which is an image including a boundary, from a chart image captured by the imaging system 2 of the chart CH. Here, the ROI image extraction means 13 extracts, as the ROI image, an image of the ROI area indicated by the ROI information set by the ROI setting means 11 from the chart image stored in the chart image storage means 10. When the start of measurement is instructed by the measurement instruction means 12, the ROI image extraction means 13 sequentially extracts ROI images from the chart images stored in sequence in the chart image storage means 10. When the normal operation mode is specified as the operation mode, the ROI image extraction means 13 outputs the extracted ROI image to the edge gradient detection means 14 and the edge profile generation means 17. When the high-speed operation mode is specified as the operation mode, the ROI image extraction means 13 outputs the extracted ROI image to the edge profile generation means 17.

[0038] The edge gradient detection means 14 detects the gradient of an edge (edge ​​gradient) which is a boundary of contrast from the ROI image extracted by the ROI image extraction means 13. Here, the edge gradient detection means 14 includes an edge extraction means 140 and a gradient calculation means 141.

[0039] The edge extraction means 140 extracts edges from the ROI image extracted by the ROI image extraction means 13. A known method can be used as the edge extraction method in this edge extraction means 140. For example, the edge extraction means 140 can apply a sharpening filter that emphasizes changes in luminance values ​​in an image, such as a Sobel filter, a Laplacian filter, or a Canny filter, to the ROI image and generate an edge-extracted image in which edges are extracted by binarizing the image.

[0040] In addition, when the ROI image has an arbitrary shape and the ROI boundary is extracted as an edge, the edge extraction means 140 deletes the edge of the ROI boundary by masking. For example, the edge extraction means 140 excludes edges included in the region where the outline of the ROI is thickened based on the ROI information that specifies the shape of the ROI. The edge extraction means 140 outputs an edge-extracted image in which the edges are extracted to the inclination calculation means 141 .

[0041] The gradient calculation means 141 calculates the gradient of an edge from an edge-extracted image in which edges are extracted by the edge extraction means 140, regarding the edges as straight lines. 3, the gradient calculation means 141 calculates the gradient of the edge e (edge ​​gradient θ) with respect to one predetermined coordinate axis of the orthogonal coordinates (here, the y-axis of the xy coordinates) in the edge extraction image EG. A known method can be used for the edge gradient calculation in the gradient calculation means 141. For example, the edge gradient calculation means 141 can calculate a rough edge gradient by a Hough transform, and then perform fitting by a normal cumulative density function or the like in the ROI image to obtain a more accurate edge gradient (for example, see the following references). (References) K. Masaoka, T. Yamashita, Y. Nishida, and M. Sugawara, “Modified slanted-edge method and multidirectional modulation transfer function estimation,” Opt. Express 22(5), 6040-6046 (March 2014). The edge gradient detection means 14 outputs the detected edge gradient (edge ​​gradient) and the ROI image to the projection information generation means 15.

[0042] The projection information generating means 15 generates projection information based on the edge gradient of the ROI image detected by the edge gradient detecting means 14. Here, the projection information generating means 15 includes a bin width calculating means 150 and a projection position matching means 151. The bin width calculation means 150 calculates the bin width of the projection axis perpendicular to the edge of the ROI image. The bin width calculation means 150 calculates a variable oversampling ratio v according to the edge gradient θ as shown in FIG. bin Using this, the bin width of the projection axis perpendicular to the edge of the ROI image R is set to 1 / v bin It is calculated as follows. The bin width calculation means 150 outputs the calculated bin width to the projection position association means 151 .

[0043] The bin width calculation means 150 calculates the bin width by 1 / sinθ, where θ is the gradient of the edge. tri Integer multiple of or 1 / sinθ tri The variable oversampling ratio is set to an integer multiple of 1, and the reciprocal of the variable oversampling ratio is calculated as the bin width of the projection axis onto which the pixels of the ROI image are projected. For example, the bin width calculation means 150 may set a pre-specified oversampling ratio as n bin , the edge gradient is θ, and round(x) is the integer closest to x. The variable oversampling ratio v satisfies the following formula (1). bin The reciprocal of (1 / v bin ) is calculated as the bin width.

[0044]

number

[0045] In addition, the oversampling ratio n bin is the fixed oversampling ratio of 4 specified by the ISO method.

[0046] Here, the technical meaning of formulas (1) and (2) will be explained with reference to FIG. 5 and FIG. In order to prevent periodic deviation from occurring in the projection axis for an arbitrary pixel matrix of an ROI image at an arbitrary edge gradient, it is necessary to adjust the bin width of the projection axis according to the edge gradient θ.

[0047] As shown in Fig. 5(a), for pixels in any row of the ROI image R, periodic phase shifts can be eliminated by multiplying the bin width of one pixel by cosθ. This corresponds to the ISO-cosine method in the conventional technology. In addition, for pixels in any column of the ROI image R, periodic phase shifts can be eliminated by multiplying the bin width of one pixel by tan θ.

[0048] Therefore, the method of the present invention (hereinafter referred to as the OMNI-sine method) multiplies a bin width of one pixel by both cos θ and tan θ. That is, the OMNI-sine method multiplies a bin width of one pixel by sin θ. As a result, whether tan θ=1 / 4 as shown in Figure 5(b) or tan θ=1 / 3 as shown in Figure 5(c), any row and any column are projected onto the projection axis at the same intervals, so no periodic shift occurs.

[0049] However, in this case, when the edge gradient θ becomes small, the bin width becomes too small, i.e., the oversampling ratio becomes too high, which may result in high computational costs and empty bins to which no pixels are projected. Therefore, in the OMNI-sine method, the bin width is scaled.

[0050] Here, as shown in Fig. 6(a), by utilizing geometric symmetry, the edge inclination θ from -90° to 90° is expressed as a triangular wave function that linearly increases and decreases in the range from 0° to 45° with a period of 90°. tri This corresponds to the above formula (2). The OMNI-sine method uses a variable oversampling ratio v bin is the specified oversampling ratio n bin The variable oversampling ratio v bin , n bin / √2, n bin This is limited to √2. This corresponds to the above formula (1). FIG. 6(b) shows the θ tri and variable oversampling ratio v bin This shows the relationship between:

[0051] As shown in Fig. 6(b), the edge slope θ (θ tri ) is in the range of 10.182° to 20.705°, 1 / sinθ is changed to the variable oversampling ratio v bin (No scaling). Also, the edge gradient θ(θ tri) is smaller than 10.182°, the variable oversampling ratio v is an integer multiple of 1 / sinθ. bin Then, the edge gradient θ(θ tri ) is greater than 20.705°, the variable oversampling ratio v is an integer multiple of 1 / sinθ (specifically, 2x). bin Let us assume that. In this example, the OMNI-sine method uses a variable oversampling ratio v bin can be limited to 1 / √2 to √2 times the width of one pixel. Of course, the variable oversampling ratio does not necessarily have to be limited to this range if the increased computational cost is not an issue and no empty bins are generated. Returning to FIG. 1, the configuration of the MTF measurement apparatus 1 will be further described.

[0052] The projection position matching means 151 matches the pixel position of the ROI image with the bin position of the projection axis perpendicular to the edge. The bin width of the projection axis is the bin width calculated by the bin width calculation means 150. For example, when the projection axis is the horizontal axis, the projection position matching means 151 rotates the ROI image on the x and y coordinates (both horizontal and vertical are one pixel width) so that the edge is perpendicular to the projection axis. Then, the projection position matching means 151 matches the pixel position (x and y coordinate position) of the ROI image before rotation with the bin obtained by projecting the x coordinate after rotation onto the projection axis of the bin width calculated by the bin width calculation means 150.

[0053] The projection position association means 151 generates projection information T that associates the pixel position of each pixel in the ROI image before rotation with the bin position of the projection axis after rotation. The projection position correlating means 151 stores the generated projection information T in the projection information storage means 16 . Furthermore, at the stage where the projection information generating means 15 stores the projection information T in the projection information storage means 16, it notifies the edge profile generating means 17 of the completion of generation of the projection information.

[0054] The projection information storage means 16 stores the projection information T generated by the projection information generating means 15, and is a general storage device such as a memory. The projection information T stored in the projection information storage means 16 is referred to by the edge profile generating means 17 .

[0055] The edge profile generating means 17 averages each pixel value of the ROI image extracted by the ROI image extracting means 13 for each bin associated with a pixel position by the projection position associating means 151, to generate an edge profile indicating the pixel distribution shape of the edge. Here, the edge profile generating means 17 includes a projection means 170 and an averaging means 171.

[0056] The projection means 170 projects the pixel value for each pixel position of the ROI image extracted by the ROI image extraction means 13 onto the corresponding bin on the projection axis. The projection means 170 refers to the projection information T stored in the projection information storage means 16, projects the pixel value of each pixel of the ROI image onto a bin on the projection axis, and generates a set for each bin. The projection means 170 outputs the set of pixel values ​​for each bin to the averaging means 171 .

[0057] The averaging means 171 averages, for each bin, a set of pixel values ​​for each bin on the projection axis generated by the projection means 170. This allows the averaging means 171 to generate an edge profile that indicates the pixel distribution shape of the edge in which the projection axis and pixel value of the oversampled sub-pixel unit are associated with each other. The averaging means 171 outputs the generated edge profile to the frequency characteristic calculation means 18.

[0058] In addition, when the edge profile generating means 17 is instructed by the measurement instruction means 12 to operate in the normal operation mode, it generates an edge profile from the ROI image extracted by the ROI image extraction means 13 when it is notified by the projection information generating means 15 that the generation of the projection information is complete. In addition, when the edge profile generating means 17 is instructed by the measurement instruction means 12 to operate in a high-speed operating mode, the edge profile generating means 17 generates an edge profile from the ROI image extracted by the ROI image extraction means 13 without waiting for the completion of generation of the projection information from the projection information generating means 15.

[0059] The frequency characteristic calculation means 18 calculates the MTF from the edge profile generated by the edge profile generation means 17 . This frequency characteristic calculation means 18 obtains a line spread function (LSF) by differentiating the edge profile generated by the edge profile generation means 17, performs a discrete Fourier transform on the LSF to obtain the absolute value, and normalizes it by DC (spatial frequency "0"). In this way, the frequency characteristic calculation means 18 can obtain the MTF. The frequency characteristic calculation means 18 outputs the calculated MTF to the measurement result output means 19 .

[0060] The measurement result output means 19 displays the measurement result of the MTF calculated by the frequency characteristic calculation means 18. The measurement result output means 19 graphs the MTF calculated by the frequency characteristic calculation means 18. For example, the measurement result output means 19 plots corresponding data on a graph with the horizontal axis representing spatial frequency (cycles / pixel) and the vertical axis representing MTF, as shown in FIG. The measurement result output means 19 outputs and displays this graph on the display device 3. This allows the person measuring to recognize the MTF of the imaging system 2 being measured.

[0061] Now, with reference to FIG. 7, the MTF measured by the MTF measurement device 1 (OMNI-sine method) will be compared with the MTF measured by the conventional ISO-cosine method. Figure 7 shows the camera MTF |sinc(ξx,ξy)| 3Assuming that the above is the case, the graph analytically combines the measurement results by the OMNI-sine method and the measurement results by the ISO-cosine method in an ROI image (200 x 200 pixels) with an edge gradient of θ. Note that Fig. 7(a) shows the case where tan θ=1 / 4 (see Fig. 5(b)), and Fig. 7(b) shows the case where tan θ=1 / 3 (see Fig. 5(c)). In the graph, the solid line shows the measurement results using the OMNI-sine method, and the dashed line shows the measurement results using the ISO-cosine method. The dotted line shows the true camera MTF |sinc(ξx,ξy)| 3 Shows.

[0062] As shown in FIG. 7(a), when tan θ=1 / 4, there is no phase shift in both the OMNI-sine method and the ISO-cosine method, and therefore the results approximate the true camera MTF.

[0063] As shown in Figure 7(b), when tanθ=1 / 3, the OMNI-sine method has no phase shift and is therefore close to the true camera MTF. However, the ISO-cosine method has a phase shift and is therefore unable to measure the MTF accurately. It should be noted that this graph is merely an example, and the OMNI-sine method can measure the MTF with high precision without any phase shift for any edge gradient θ.

[0064] With the above-described configuration, the MTF measurement device 1 can measure the MTF with high accuracy and precision by adjusting the bin width of the projection axis in accordance with the edge inclination. Furthermore, by calculating the variable oversampling ratio using equation (1), the MTF measurement device 1 can limit the oversampling ratio to a range from 1 / √2 to √2 times the desired value. This makes it possible to vary the bin width and suppress the increase in calculation cost that occurs with small edge gradients and the occurrence of empty bins to which no pixels are projected. The MTF measurement device 1 can be operated by an MTF measurement program that causes a computer (not shown) to function as each of the above-mentioned means.

[0065] [Operation of MTF measuring device] Next, the operation of the MTF measurement device 1 according to the embodiment of the present invention will be described with reference to FIG. 8 (and for the configuration, refer to FIG. 1 as appropriate). Here, the MTF measurement device 1 sequentially stores chart images obtained by capturing an image of the chart CH, which are input from the imaging system 2, in the chart image storage means 10, and displays the chart images on the display device 3. The operation of the MTF measurement device 1 to measure the MTF from the chart images will be described in detail below. It is assumed that the MTF measurement device 1 is set to the normal operation mode by the measurement instruction means 12.

[0066] In step S1, the ROI setting means 11 sets the position, size, etc. of the ROI. For example, the ROI setting means 11 sets the ROI when the measurer designates a closed region of any shape within the chart image displayed on the display device 3. In step S2, the ROI image extraction means 13 extracts, from the chart image stored in the chart image storage means 10, an image of the ROI area set in step S1 as an ROI image.

[0067] In step S3, the edge extraction means 140 of the edge gradient detection means 14 extracts edges from the ROI image extracted in step S2. In step S4, the gradient calculation means 141 of the edge gradient detection means 14 calculates the gradient of the edge extracted in step S3 (edge ​​gradient).

[0068] In step S5, the bin width calculation means 150 of the projection information generation means 15 calculates a bin width according to the edge gradient of the projection axis perpendicular to the edge of the ROI image. Here, the bin width calculation means 150 calculates a pre-specified oversampling ratio as n bin When the edge gradient calculated in step S4 is θ, the variable oversampling ratio v bin Calculate 1 / v bin Calculate the bin width. In step S6, the projection position matching means 151 of the projection information generating means 15 generates projection information T that matches a pixel position of the ROI image with a bin of the projection axis having the bin width calculated in step S5, the pixel position being vertically projected along the edge gradient. The projection position matching means 151 stores the projection information T in the projection information storage means 16 (not shown as a step).

[0069] In step S7, the projection means 170 of the edge profile generation means 17 refers to the projection information T to project the pixel values ​​of each pixel of the ROI image extracted in step S2 onto the bins of the projection axis, and generates a set for each bin. In step S8, the averaging means 171 of the edge profile generating means 17 averages the set of pixel values ​​on the projection axis generated in step S7 in units of bins. This enables the edge profile generating means 17 to generate an edge profile that indicates the pixel distribution shape of the edge of the ROI image.

[0070] In step S9, the MTF measurement device 1 uses the frequency characteristic calculation means 18 to differentiate the edge profile generated in step S8 to obtain a line spread function (LSF), and then calculates the MTF by performing a discrete Fourier transform on the LSF. In step S10, the MTF measurement device 1 displays the MTF measurement results calculated in step S9 by the measurement result output means 19. For example, as shown in Fig. 7, the measurement result output means 19 plots corresponding data on a graph with the horizontal axis representing frequency (cycles / pixel) and the vertical axis representing MTF, and displays the graph on the display device 3.

[0071] Through the above operations, the MTF measurement device 1 can measure the MTF with high accuracy and precision by adjusting the bin width of the projection axis in accordance with the edge gradient. Here, the operation when the normal operation mode is set as the operation mode has been described. When the high-speed operation mode is set as the operation mode, the MTF measurement device 1 may repeat the operations of steps S7 to S10 for the chart images stored in the chart image storage means 10 one by one. However, the high-speed operation mode is premised on the fact that steps S1 to S6 have been performed in advance in the normal operation mode, and the projection information has been stored in the projection information storage means 16. The configuration and operation of the MTF measurement device 1 according to an embodiment of the present invention have been described above, but the present invention is not limited to this embodiment and also includes design modifications and the like that do not deviate from the gist of the present invention. [Explanation of symbols]

[0072] 1. MTF measurement device 10 Chart image storage means 11 ROI setting means 12 Measurement instruction means 13 ROI image extraction method 14 Edge gradient detection means 140 Edge Extraction Method 141 Tilt calculation method 15 Projection information generation means 150 Bin width calculation method 151 Projection position matching means 16 Projection information storage means 17 Edge profile generation method 170 Projection means 171 Averaging means 18 Frequency characteristic calculation means 19 Measurement result output means 2 Imaging system 3 Display device CH ChartCH

Claims

1. 1. An MTF measurement device for measuring MTF representing a spatial frequency characteristic of an imaging system using a chart having different contrast at its boundaries, comprising: an ROI image extracting means for extracting an ROI image, which is an image including the boundary, from a chart image captured by the imaging system of the chart; an edge gradient detection means for detecting a gradient of an edge that is the boundary from the ROI image; a bin width calculation means for calculating a bin width of a projection axis along which pixels of the ROI image are projected, the bin width being a reciprocal of a variable oversampling ratio that is an integer multiple of 1 / sinθ or a multiple of an integer multiple of 1 / sinθ, where θ is a gradient of the edge; a projection position matching means for matching pixel positions of the ROI image with bin positions of the projection axis perpendicular to the edge; an edge profile generating means for averaging each pixel value of the ROI image for each bin associated with a pixel position by the projection position corresponding means to generate an edge profile; A frequency characteristic calculation means for calculating an MTF from the edge profile; An MTF measuring device comprising:

2. 1. An MTF measurement device for measuring MTF representing a spatial frequency characteristic of an imaging system using a chart having different contrast at its boundaries, comprising: an ROI image extracting means for extracting an ROI image, which is an image including the boundary, from a chart image captured by the imaging system of the chart; an edge gradient detection means for detecting a gradient of an edge that is the boundary from the ROI image; The pre-specified oversampling ratio is n ｂｉｎ , the gradient of the edge is θ, and round(x) is the integer closest to x, [0010] A variable oversampling ratio v that satisfies ｂｉｎ a bin width calculation means for calculating the inverse of the above as a bin width of a projection axis onto which pixels of the ROI image are projected; a projection position matching means for matching pixel positions of the ROI image with bin positions of the projection axis perpendicular to the edge; an edge profile generating means for averaging each pixel value of the ROI image for each bin associated with a pixel position by the projection position corresponding means to generate an edge profile; A frequency characteristic calculation means for calculating an MTF from the edge profile; An MTF measuring device comprising:

3. The edge profile generating means a projection means for projecting a pixel value for each pixel position of the ROI image onto the bin position associated by the projection position association means; an averaging means for averaging pixel values ​​projected onto the bins to generate the edge profile; 3. The MTF measuring device according to claim 1, further comprising:

4. 4. The MTF measuring device according to claim 1, further comprising a measurement result output unit that outputs the MTF calculated by the frequency characteristic calculation unit in the form of a graph.

5. 5. An MTF measurement program for causing a computer to function as the MTF measurement device according to claim 1.

Images