JPH0981742A - Method and device for measuring resolution - Google Patents

Abstract

PROBLEM TO BE SOLVED: To measure resolution without using a conventional test chart. SOLUTION: When one black image part is formed on a white background part in a window, the density histogram of the window is found (S2) and a high-density most-frequent value Ri and a low-density most-frequent value Rb are found (S3 and S4); when the part between those two density values Ri and Rb is denoted as 100%, the frequency Si included in a 50% density area (w) is found while the range (w) is moved along its density (S5), the range (w) wherein the smallest frequency Fm among frequencies Fi found by ranges (w) is obtained is detected to find edge width We, and the border is binarized by using density ReW having the center frequency in the range (w) wherein the smallest frequency Fm is obtained as a threshold to find the standard deviation σ of deviation from a longitudinal primary regression straight line B0, thereby finding the resolution R=1(Ew+σ) (S6-S11).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a resolution measuring method and apparatus for evaluating the image quality of a printed matter such as a copy obtained by a copying machine, a printer, a digital copier, a facsimile machine, a scanner, etc. The present invention relates to a method and apparatus for evaluating the edge portion of a binary image such as a character and estimating the resolution of the image.

[0002]

2. Description of the Related Art Conventionally, in order to measure resolution, a test chart in which a large number of adjacent straight lines are drawn is used, and this test chart is copied by using, for example, a copying machine whose resolution is to be measured, The mutual separation and contact of the black image portions formed on the white background portion of the copy image thus obtained are determined.

[0003]

In this prior art, it is only possible to evaluate the resolution of the image drawn on the test chart, and it is impossible to evaluate the resolution of any image.

It is an object of the present invention to provide a method and a device which make it possible to measure the resolution for any image.

[0005]

According to the present invention, a window is composed of a plurality of pixels each having a plurality of gradations.
In this window, 2 extending in a predetermined first direction
The frequency for each density of each image in this window including the boundary of the image to be digitized is calculated, and the first density Ri of the mode having the maximum frequency on the high density side and the frequency on the low density side are determined. The maximum modal value of the second density Rb and the maximum density of the second density Rb are obtained, and a predetermined constant density difference range w between the first and second densities Ri and Rb is moved and included in each range w. Frequency F
i is obtained, the range w in which the minimum frequency Fm is obtained is detected from among the frequencies Fi obtained in each range w, and the first range w in the range w in which the minimum frequency Fm is obtained is detected. The width ew extending in the second direction orthogonal to the one direction is obtained as the edge width Ew, and the third density Rew having the central frequency in the range w in which the minimum frequency Fm is obtained is set as a threshold value, and the third density Rew is calculated. The boundary as a threshold value is binarized, and the second with respect to the regression line B0 along the first direction of the boundary.
The resolution measuring method is characterized in that the standard deviation σ of the deviation in the direction is obtained, and the resolutions R and R = 1 / (Ew + σ) are obtained. The predetermined concentration range w is defined by w = r · (Rb−Ri), and r is a value of 1 / 1.5 to 1 / 2.5. Further, according to the present invention, an original having a linear image extending linearly is picked up by an image pickup device, and the image pickup device is arranged in a matrix in a first direction and a second direction perpendicular to the first direction. It has a plurality of light receiving elements that configure each pixel for detecting each gradation, stores the output of the light receiving elements in a memory, reads the stored contents of the memory, obtains the frequency for each density, and The first mode with the maximum frequency on the side
The density Ri and the second density Rb of the mode having the maximum frequency on the other density side are obtained, and the first and second densities Ri, Rb are obtained.
Range w (0 <r <1), w = r · (Rb−Ri) between 1 and a plurality of predetermined gradations are shifted and included in each range w. Of the frequency Fi obtained for each range w, the range w in which the minimum frequency Fm is obtained is detected, and the longitudinal direction of the line-shaped image is virtually coincident with the first direction. In this state, the width ew extending in the second direction of the range w in which the minimum frequency Fm is obtained in the window is obtained, the edge widths Ew and Ew = ew / r are obtained, and the minimum frequency Fm is obtained. The third density Rew having the central frequency in the obtained range w is used as a threshold value, and the boundary is binarized using the third density Rew as a threshold value, and linear regression along the first direction of the boundary is performed. The standard deviation σ of the deviation in the second direction with respect to the straight line B0 is obtained, and the resolution R, = A resolution measuring method characterized by determining 1 / a (Ew + σ). r is, for example, a value of 1 / 1.5 to 1 / 2.5. Further, in the present invention, when the unevenness correction coefficient is k1, the product σ · k1 of the standard deviation σ and the unevenness correction coefficient k1 is 50.
The unevenness correction coefficient k1 is determined so that
It is characterized in that k1 is used for calculation of the resolution R. Further, according to the present invention, the image pickup means for picking up an image has a means for optically enlarging the image, determines an edge width correction coefficient k2 corresponding to the enlargement magnification, and sets the edge width Ew to the edge width correction coefficient k2. It is characterized in that only the difference is subtracted and used for the calculation of the resolution R. Further, according to the present invention, an original having a linear image extending linearly is picked up by an image pickup device, and the image pickup device is arranged in a matrix in a first direction and a second direction perpendicular to the first direction. It has a plurality of light receiving elements that configure each pixel for detecting each gradation, stores the output of the light receiving elements in a memory, reads the stored contents of the memory, obtains the frequency for each density, and The first mode with the maximum frequency on the side
The density Ri and the second density Rb of the mode having the maximum frequency on the other density side are obtained, and the first and second densities Ri, Rb are obtained.
Range w (0 <r <1), w = r · (Rb−Ri) between 1 and a plurality of predetermined gradations are shifted and included in each range w. Of the frequency Fi obtained for each range w, the range w in which the minimum frequency Fm is obtained is detected, and the edge width Ew is the length L of the window in the first direction, the pixel A2 in the second direction
Then, Ew = a2 · Fm / L is calculated, which is a method for measuring an edge width of a line image. Further, according to the present invention, a document having an image on a line is imaged by an image pickup means, and the image pickup means is a first direction which is a vertical scanning direction parallel to the longitudinal direction of the image on the line and a horizontal direction perpendicular to the first direction. It has a plurality of light receiving elements forming pixels each having a plurality of gradations arranged in a matrix in the second direction which is the scanning direction, and stores the output of the light receiving element in a memory. Is read out to find the frequency for each density, and the first density Ri of the mode having the maximum frequency on the high density side and the second density Rb of the mode having the maximum frequency on the low density side are calculated. Then, while moving the range w, w = (Rb-Ri) / 2 between the first and second densities Ri, Rb by shifting by one gradation, the frequency Fi included in each range w is moved. Of the frequencies Fi calculated for each range w, the minimum frequency Fm is The line-shaped image in which the minimum frequency Fm is detected and the third density Rew having the central frequency in the range w in which the minimum frequency Fm is obtained is set as a threshold value and is equal to or more than or less than the third density Rew in the window. When the number B of pixels is calculated and the length a2 of the pixel in the second direction is calculated, the line widths Lew and Lew = a2 · B / L are calculated, and the line width of the line image is measured. Further, in the present invention, the image pickup means for picking up the image has a means for optically enlarging the image, obtains a line width correction coefficient k3 corresponding to the enlargement magnification, and calculates only the line width correction coefficient k3 from the line width Lew. It is characterized by subtracting and correcting. Further, the present invention is an image pickup means for picking up an image of a document having a linear image extending linearly, the first means of a column being a vertical scanning direction and the first row of a row being a horizontal scanning direction perpendicular to the first direction. An image pickup unit having a plurality of light receiving elements forming pixels for detecting a plurality of gradations arranged in a matrix in two directions, a memory for storing an output of the light receiving element of the image pickup unit, and a stored content of the memory are described. The frequency of each density is read out and the frequency of one density side is maximum
Between the first and second densities Ri and Rb in response to the output of the density detection means and the density detection means for obtaining the second density Rb of the mode having the maximum frequency on the other density side. Of the gradation range w (where 0 <r <1) and w = r · (Rb−Ri) are shifted by one or a plurality of predetermined gradations, and are included in each range w. The range detection means for detecting the range w in which the minimum frequency Fm is obtained among the frequencies Fi obtained for each range w and the output of the range detection means, and the length of the linear image is determined. Direction
Virtually, when the state coincides with the first direction, the width ew extending in the second direction of the range w in which the minimum frequency Fm is obtained in the window is obtained, and the edge widths Ew, Ew = ew / r. In response to the output of the edge width computing means for obtaining the value and the range detecting means, the third density Rew having the central frequency in the range w in which the minimum frequency Fm is obtained is set as a threshold value, and the third density Rew is calculated. In response to the output of the binarizing means for binarizing the boundary as the threshold value, the standard deviation σ of the deviation in the second direction with respect to the linear regression line b0 along the first direction of the boundary is set. The resolution measuring apparatus is characterized in that the resolution R, R = 1 / (Ew + σ) is obtained in response to each output of the standard deviation computing means to be obtained and the edge width computing means and the standard deviation computing means. r is, for example, a value of 1 / 1.5 to 1 / 2.5.

The operation will be described. The resolving power is represented by the resolution ratio with respect to the spatial frequency f, that is, the resolution Rf. For example, when a black line, which is a black image portion, exists in a white background portion, for example, when scanning is performed in a direction intersecting the black line, separation and contact with an adjacent black line are determined. Resolution Rf at spatial frequency f
Can be calculated by Equation 1. Rf = (Nf−N1f) / Nf (1) In Formula 1, Nf is the number of black lines to be measured drawn on the conventional white background portion, and N1f is the contact point in this test chart. Shows the number of black lines. This black line is identified by the binarization threshold value R60. In general, the binarization threshold value Rp of p% is expressed by Equation 2.

[0007]

[Equation 1]

When p = 60%, the binarization threshold value R
60 is shown in Equation 3.

R60 = Rb + 0.6 (Ri-Rb) =
0.6 · Ri + 0.4 · Rb (3) The resolution expresses the characteristic of the resolution with respect to the spatial frequency f as described above, but is simply expressed by the spatial frequency f when the resolution is 50%. You may In the present invention, the resolution of such a simple method is measured.

According to the invention, the claims are
An image of a document having a linear image extending linearly as shown in FIG. 2 is picked up by a light receiving element having pixels capable of detecting a plurality of gradations arranged in a matrix and stored in a memory. For example, when a single line-shaped image extending in the first direction, which is the vertical column direction, exists in the section, and when, for example, a black line-shaped image is formed on the white base portion, one of the high-level images is high. The first density Ri of the most frequent value in which the frequency of the pixels on the density side is maximum
And the second density Rb of the mode in which the frequency of the pixels on the low density side of the white background is the maximum, and the density range between the first and second densities Ri and Rb is set to 100%, and The r% density range is set as a gradation range w that is a sampling range of the edge width, and the range w is shifted by one gradation or by a plurality of predetermined gradations, and is included in each range w. The frequency Fi of each pixel is obtained, and the range w in which the minimum frequency Fm is obtained among the frequencies Fi obtained for each range w is obtained.
Is detected. The longitudinal direction of the line-shaped image is virtually
The direction coincides with the direction (for example, the column direction), and this can be easily performed, for example, by processing the data stored in the memory, and the range w in which the minimum frequency Fm thus obtained is obtained. The width ew extending in the second direction (for example, the horizontal row direction) is calculated. The range w in which the minimum frequency Fm is obtained is a steep range in which the amount of change in gradation with respect to the amount of displacement in the second direction is the largest,
Judge that the axis exists. In optimizing the edge width Ew, when the density range between the first and second densities Ri and Rb is set to 100%, the minimum frequency Fm which is the r% density range therebetween is obtained. The extension line of the edge cross-section density gradient in the range w is a value at which the width of the foot at the intersection of the extension line of the first density Ri of the line image and the extension line of the second density Rb of the base portion is minimum. r is selected in the range of 1 / 1.5 to 1 / 2.5, for example, 1/2 (that is, 50%). r
= 1/2, the edge width Ew can be obtained by doubling the width ew. Further, the boundary of the line-shaped image is binarized with the third density Rew having the center frequency in the range w in which the minimum frequency Fm is obtained as a threshold value, and it is obtained in advance along the first direction of the boundary. The standard deviation σ of the deviation in the second direction with respect to the linear regression line B0, that is, the degree of edge unevenness is calculated. Thereby, the resolution R can be obtained as the reciprocal of (Ew + σ). In obtaining the resolution at the boundary between the linear image extending linearly and the base portion, the width ew extending in the second direction of the range w in which the minimum frequency Fm is obtained is defined as the edge width Ew in the equation for obtaining the resolution R. Substitute as

According to the present invention of claim 1, the resolution of the boundary between the curved line and the base portion is measured by the same concept as the measurement of the resolution of the linear image extending linearly. Is also possible. The curve extends in a predetermined first direction, which determines a regression equation, for example by reading the stored contents of memory, which regression equation may be, for example, a quadratic curve, a hyperbola, etc. The regression line B0 of the regression equation thus obtained has a width ew extending along the first direction in the second direction which is a normal line perpendicular to the tangent line at each position along the regression line B0. , The edge width Ew, and the standard deviation σ can be obtained from the deviation in the normal direction, which is the second direction, with respect to the regression line B0, and thus the resolution R can be calculated.

According to the present invention, when the boundaries on both sides of the curve are present in the window, the value r is calculated from the width ew as in the arithmetic processing of the linear image extending linearly.
The edge width Ew may be obtained by using.

Thus, the resolution of an arbitrary image can be measured. The measurement section w of the edge width Ew is the same as the r% attenuation area of the density area of the resolution determination, and the resolution evaluation method using the conventional test chart and the measurement of the approximate resolution can be achieved. In this way, the resolution of an image such as an arbitrary binary boundary or line can be directly estimated and evaluated, and a wide range of image quality can be evaluated.

In particular, according to the present invention, as described above, among the frequencies Fi obtained for each range w by moving the range w of gradations by shifting one or a plurality of gradations in the gradation direction,
Since the range w in which the minimum frequency Fm is obtained is detected and the width ew is obtained, and thereby the edge width Ew is obtained,
The width ew and the edge width Ew can be minimized so that the measurement result can always be stably obtained and the sampling width w at which the maximum value of the density change rate of the edge width Ew is obtained can be found. Become. This makes it possible to stabilize the measurement accuracy of the edge width Ew. That is, since the maximum section w of the density change rate of the density cross section becomes the evaluation area of the edge portion, the measurement accuracy can be stabilized. Therefore, when measuring the edge width Ew using a CCD (charge coupled device) as a light receiving element of the image sensor, the edge width is not affected by the quality of the target image and the ratio of the evaluation area of the image and the base material. It is possible to prevent variations in measurement accuracy of Ew, and hence resolution R, from occurring.

According to the present invention, the unevenness of the edge having the standard deviation σ can be corrected by multiplying the unevenness correction coefficient k1. As a result, the unevenness degree of the edge is corrected by assuming that the regression lines B0 of the line width contour line B1 are arranged at equal intervals and that 50% of the regression lines B0 are in contact with each other, and the edge of the contour line B1 is corrected. The unevenness degree σ of is corrected. In this way, the unevenness correction coefficient k is set so that the product σ · k1 becomes 50%.
By setting 1 to measure the other resolution, the contact rate of the line-shaped image for resolution determination is set to 50%, and the contact rate is approximated to 50%, which is similar to the test chart resolution method. It is possible to make a correction and measure the resolution.

Further in accordance with the present invention, the image pickup means has means for optically enlarging the image, and the edge-shaped correction coefficient k2 is determined corresponding to the enlargement magnification, and the edge of the corrected image is corrected. Width (Ew-k2) is the edge width Ew of resolution R
By substituting into the image capturing means, it is possible to prevent the fluctuation of the edge width value at the time of measurement from becoming significant due to the influence of the pixel size when changing the magnification when measuring with the microscope in the image capturing means, that is, the image capturing by the CCD camera. It is possible to eliminate the disadvantage that the measured value becomes large due to the pixel size depending on the measurement magnification of the means, that is, the so-called line thickening tendency tends to occur.
Therefore, when measuring the CCD camera output data, the edge width Ew is measured at each magnification of the microscope, and based on the result,
The pixel width correction coefficient k2 is obtained and the edge width is corrected.
For example, when r = 1/2 is determined, 2 · ew = Ew−k
Let it be 2. This 2 · ew is substituted as Ew in the equation of resolution R. This makes it possible to deal with the magnification of the microscope. In this way, the accuracy of edge width measurement can be improved, and highly accurate edge width measurement can be performed. Further, the third density Rew is set to a value having a central frequency in the range w in which the minimum frequency Fm is obtained, and the third density Rew is used as a threshold value to binarize an image such as a boundary or a line. Thus, the standard deviation σ is obtained, so that the contour of the line width can be prevented from being influenced by the quality of the target image and the ratio of the cloth which is the base portion of the image. This increases the degree of freedom of measurement conditions. This line width contour is used as described above for the standard deviation σ, which is the roughness of the edge.

Further according to the invention, the edge width Ew is
It can be calculated based on the above-described minimum frequency Fm and the length L of the window in the first direction and the length A2 of the pixel in the second direction. In addition, the line width Lew of the line image or the curved image that extends linearly is the number of pixels B or the third that is equal to or larger than the third density Rew of the image of the line or the like
It can be obtained by using the number of pixels B that is less than the density Rew. When the background is white and the linear or curved line-shaped image is black, the pixel number B is level discriminated at the third density Rew or higher, that is, the black image has high density, and therefore the reflectance is high. small. When the background portion is black and the straight or curved image is white, the number of pixels B is
The number of pixels that are less than the third density Rew. In this case,
The image has low density and high reflectance. Further, for this line width Lew, the pixel size correction value k3 corresponding to the magnification of the image pickup means such as a CCD camera is used, and the measurement result of the low-magnification line width can be obtained close to the high-magnification line width measurement value. To enable that. In this way, the cause of the line weight of the line width measurement value of the CCD camera can be eliminated by correcting the pixel size.

[0018]

FIG. 1 is a block diagram showing a typical embodiment of the present invention. The imaging means 13 is a CC
The D camera head 1 and the microscope 2 having a magnifying function, which is provided in front of the objective lens of the D camera head 1, are installed on the stand 14. A copy machine, a printer,
A printed matter that has been copied by a digital copier, a facsimile machine, a scanner, or the like, and has a linear image extending linearly on its upper surface.
5 is illuminated by the light source 3 and the illuminator 4 that illuminates so as not to produce shade. The light source means 5 controls the illumination of the illuminator 4. The CCD camera head 1 detects a plurality of grayscales arranged in a matrix in a first direction of a column which is a vertical scanning direction and a second direction of a row which is a horizontal scanning direction perpendicular to the first direction. It has a plurality of light receiving elements, and this light receiving element is realized by a CCD (charge coupled device).

The operation of the camera head 1 is controlled by the control unit 6, and the image obtained by picking up the image is given to the resolution processing circuit 7 to calculate the resolution and the like, and the image picked up by the camera head 1 and The calculation processing result in the resolution processing circuit 7 is displayed on the image monitor 8 realized by a liquid crystal or a cathode ray tube. Furthermore, the output of the image processing circuit 7 is an arithmetic processing circuit 9 realized by a personal computer or the like.
And the calculation is performed. The calculation processing is input and controlled by the keyboard 11. The calculation results of the resolution processing circuit 7 and the calculation processing circuit 9 are printed on recording paper by the printer 12 and output.

The operation of the resolution processing circuit 7 is shown in FIG. This makes it possible to measure the resolution as well as the edge width Ew and the line width Lew.

FIG. 3 is a view showing a window 16 of the image of the original 15 detected by the camera head 1. The image of the window 16 is given from the control unit 6 to the resolution processing circuit 7 and stored in the memory 17. It A line image A extending in a black straight line is drawn on a white background portion C in the window 16. The outline of one boundary of the line-shaped image A is indicated by reference numeral B1, and its linear regression line is indicated by reference numeral B0. The linear regression line B0 is parallel to the column direction, which is the vertical scanning direction of the window 16, that is, the first direction by the image processing in the memory 17, and the horizontal scanning direction perpendicular to the first direction is the row direction. The second direction. In this embodiment, the vertical length L in the first direction is the horizontal length X in the second direction, and in the present embodiment, for example, L = X
It is.

FIG. 4 is a diagram for explaining the resolution. The difference ΔV1 between the maximum value V1 and the minimum value V2 of the density of each pixel along the second direction of the row of the window 16 shown in FIG.
Is large and the resolution is assumed to be 100%, for example, when the difference ΔV2 between the maximum value V3 and the minimum value V4 of the density becomes small as shown in FIG. ,

[0023]

[Equation 2]

It is said that the spatial frequency of the resolution Rv% is satisfied. The spatial frequency is thus 5
When it is 0% or less, it can be said that the resolution is poor and that it is not resolved.

Returning to FIG. 2 again, the resolution processing circuit 7 starts the calculation processing operation of the resolution from step s0.
At step s1, the output of the camera head 1 is stored in the memory 17 according to the gradation, and this memory 17
The density correction calculation of the stored contents of is performed.

As shown in FIG. 5, the output of each pixel of the camera head 1 is discriminated into a total of 256 gradations from gradation 0 to 255. For example, the reflectance of 98% of the white background portion is gradation 227. Correspondingly, the density correction calculation is performed so that the reference curves d1 and d2 match and the correction curve 17 is obtained so that the black reflectance 11.2% corresponds to the gradation 28. In this way, it is possible to calibrate each gradation from 0 to 255 of a pixel having a reflectance of 0 to 100% so as to be roughly represented by a linear line.

In step s2, a histogram showing the relationship between the gradation and the number of pixels, that is, the frequency in the window 16 shown in FIG. 3 is obtained. In the resolution processing circuit 7, as shown in FIG. 6, the frequency for each density is calculated based on the contents stored in the memory 17, and the first density Ri which is the mode having the maximum frequency on one high density side is the maximum. , And the second density Rb, which is the most frequent value in which the frequency on the other low density side is maximum, is obtained. First
The density Ri corresponds to the black line image A, and the second density R
b corresponds to the white base portion C. Such first and second densities Ri and Rb are obtained in steps s3 and s4, respectively.

In the next step s5, the density range of the first and second densities Ri and Rb is set to 100%, and the gradation range w between them is set as follows: w = r. (Rb-Ri) (5) To do. Here, r is set to 0 <r <1 (6), and in this embodiment, for example, r = 1/2 =
It is set at 50%.

In order to obtain the edge width Ew, steps s6 to s8 are executed. In the above equation 5, r = 0.
By selecting 5, the following expression 7 is established.

W = 0.5 (Rb−Ri) (7) The frequency Fi included in this range w is calculated.

[0031]

(Equation 3)

This equation 8 is the area shown by hatching between the first and second concentrations Ri to Rb in FIG. 7 which is a simplified version of FIG. fi represents the frequency for each gradation.

The frequency Fi is obtained while shifting the first and second gradations Ri to Rb by one gradation in this embodiment. As a result, of the frequencies Fi calculated for each range w, the range w in which the minimum frequency Fm is obtained is detected. In the range w in which the minimum frequency Fm is obtained in this manner, the amount of change in gradation with respect to the amount of horizontal displacement in the second direction is the maximum.

FIG. 8 is a graph showing the relationship between the horizontal position of the window 16 shown in FIG. 3 along the second direction and the gradation. In FIG. 8, gradations on a single horizontal scanning line are shown. In the present invention, the edge width Ew is obtained as follows based on the range w in which the minimum frequency Fm is obtained.

FIG. 9 is a diagram showing the edge width Ew of the window 16 shown in FIG. Second row 1 which is a horizontal row
9, the white background portion C has the second density Rb and the black image portion A has the first density Ri. This image part A is 1
It has two boundaries 20, 21.

FIG. 10 is a diagram showing changes in gradation along the second horizontal direction 19 near one boundary 20 of the image portion A. In FIG. 10, the range in which the above-mentioned minimum frequency Fm is obtained is indicated by reference numeral w. The width ew extending in the second direction (the left-right direction in FIG. 10) of the range w in which the minimum frequency Fm is obtained is obtained in step s7 in FIG. Extension lines 25 and 26 of the edge cross-section concentration gradient 24 including the range w in FIG.
Of the image density Ri intersects the extension line of the image density Ri and the intersection 23 intersects the extension line of the density Rb of the base portion C. The foot width between the intersections 22 and 23 is defined as the edge width Ew.

FIG. 11 is a diagram showing a part of the window 16 in a simplified and enlarged manner. The vertical length of the pixel 27 in the first direction is A1, and the horizontal length of the pixel 27 in the second direction is a2.
N pieces are arranged in the direction and m pieces are arranged in the second direction, and L = a1 · n (9) X = a2 · m (10) In this embodiment, a1 = a2, but they are different from each other. May be. Although n = m in this embodiment, they may be different from each other.

The edge width Ew is expressed by equation 11.

[0039]

(Equation 4)

Here, as described above, for example, r = 0.
When set to 5,

[0041]

(Equation 5)

When the range w in which the minimum frequency Fm is obtained is detected in the above step s5, the next step s
In 9, the third density Rew having the central frequency in the range w in which the minimum frequency Fm is obtained is set as the threshold value for binarization.

Rew = zi + 0.5w (13) xi in this equation 13 is the gradation from the first density Ri of the image to the range w in which the minimum frequency Fm is obtained, as shown in FIG. Indicates.

FIG. 12 is a diagram showing a simplified concentration distribution in the second direction in FIG. 8 described above. 1 in FIG.
/ Fn50 represents the length or width in the second direction with respect to a resolution of 50%, and is the sum of 2 · ew and σ · k1 (2 · ew +
σ · k1). Here, σ is the standard deviation described with reference to FIGS. 13 and 14, in other words, it represents the edge unevenness, and k1 is a correction coefficient for the unevenness σ.

An image obtained by binarizing the window 16 shown in FIG. 13 is obtained by using the above-mentioned third density Rew as a threshold value. On the other hand, the linear regression line B0 of the boundary 20 in the image of the window 16 after binarization is calculated and obtained. Boundary 2
A specific straight line of 0 is indicated by reference numeral B1.

FIG. 14 is an enlarged view of the boundary 20.
Along the one straight line 19 in the second direction, the linear regression line B0
By calculating the distance (XB0-XB1) between the coordinates XB0 and XB1 between the position B1 and the position B1 of the boundary 20 and performing the calculation based on the equation 14 along a total of n horizontal scanning lines 19 over the first vertical direction, In step s10, the standard deviation σ is obtained over the vertical length L.

[0047]

(Equation 6)

In step s11, the resolution R is obtained based on the edge width Ew and the standard deviation σ obtained in step s8.

R = 1 / (Ew + σ) (15) The unevenness degree correction coefficient k1 is the standard deviation σ and the unevenness degree correction coefficient k.
The product σ · k1 with 1 is determined to be 50%, and once this unevenness degree correction coefficient k1 is determined, this product σ · k1 is previously described for the resolution measurement of another document. It is used by substituting for σ in Expression 15.

As a result, the standard deviation σ is multiplied by the edge unevenness correction coefficient k1, and the contact rate of the line image for resolution determination is set to 50%. Will be able to do.

The camera head 1 is equipped with a microscope 2 and has an edge width correction coefficient k2 corresponding to the magnification.
Determine.

2 · ew = Ew−k2 (16) As a result, the measurement of the image by the camera head 1 is affected by the pixel size when the magnification when measuring with the microscope 2 is changed, and the edge width at the time of measurement is changed. In order to prevent the fluctuation of Ew from becoming significant, in order to remove the effect, a pixel size correction coefficient k2 corresponding to this magnification is determined. In this way, the accuracy of edge width measurement can be improved regardless of the magnification.

FIG. 15 is a diagram showing the window 16 in a simplified manner, showing the line width Lew of the black line image A. The window 16 binarized by the above-mentioned third density Rew
In this image, the number of pixels B included in the image A is obtained, and the line width Lew can be obtained from this.

Lew = a2 · B / L (17) When the image A has a high density of black, the number of pixels of the image A shown by hatching has a gradation equal to or higher than the third density Rew, and On the contrary, when the white line-shaped image A is present on the black background portion C, the line-shaped image A has a gradation lower than the third density Rew.

A line width correction coefficient k3 for correcting the line width Lew is calculated in accordance with the magnification of the microscope 2.

Lew = a2 · B / L = k3 (18) With this, the pixel size correction corresponding to the magnification of the camera head 1 is performed, and the high-magnification line width measurement is performed even with the low-magnification line width measurement value. A measurement result close to the value can be obtained.
In this way, the cause of the line weight of the line width measurement value of the camera can be eliminated by the pixel size correction.

In the above embodiment, typically r =
Although the edge width Ew is determined by setting the width to 0.5 and doubling the width ew, the width ew and the standard deviation σ at each of the left and right boundaries 20 and 21 are calculated as another embodiment of the present invention. Then, the resolution R may be obtained by using the sum thereof.

Further, according to the present invention, when only one boundary 20 exists in the window 16, the resolution σ may be calculated and calculated based on the edge width Ew and the standard deviation σ of only the boundary 20. . In the above-described embodiment, the linear image A has a straight line shape, and its linear regression line B
The calculation is performed by setting 0 to be parallel to the first direction in the column direction of the window 16. In another embodiment of the invention, the image A has at least one curved boundary 20 or 21.
Also in the case of having two, the direction along each of the boundaries 20 and 21 is the first direction, the normal direction perpendicular to the tangent line at each point along the first direction is the second direction, the edge width Ew and the standard deviation σ. Can be obtained in the same manner as described above to calculate the resolution R.

[0059]

According to the present invention, the resolution of an unspecified arbitrary image such as a linear image extending linearly, one boundary of the linear image, and a curved image and one boundary is evaluated. For the first time, measurement can be performed without using a ladder pattern drawn on a conventional test chart.

Further, according to the present invention, the minimum frequency Fm is moved while moving the gradation range w in the window in the gradation direction.
Detecting the range w in which is obtained, the width ew and the edge width E
Since w is obtained, and the standard deviation σ is obtained by performing binarization with the third density Rew having the central frequency in the range w where the minimum frequency Fm is obtained as a threshold value. It is possible to estimate and measure the resolution of any desired image, and a wide range of image quality can be evaluated by directly estimating and evaluating the resolution of any binary image. Further, by selecting the value r to be, for example, 1/2, the measurement section w of the edge width Ew can be made to be the same as the 50% attenuation range of the density range of the resolution judgment, and the resolution using the conventional test chart is thereby obtained. It becomes possible to measure the resolution similar to the evaluation method.

Further, according to the present invention, the standard deviation σ, and therefore the unevenness degree, is corrected by using the unevenness degree correction coefficient k1, whereby the contact rate of the image for resolution determination is set to, for example, 50%. It is possible to make a correction that approximates the contact rate that is close to the above to 50% upper probability.

Further, the edge width Ew is corrected by the edge width correction coefficient k2 corresponding to the enlargement magnification of the means for optically enlarging the image of the microscope or the like provided in the image pickup means, whereby the measurement error of the pixel size is obtained. To reduce
It is possible to improve the accuracy of resolution measurement of the edge width Ew.
The range w in which the above-mentioned minimum frequency Fm is obtained in this way
By detecting, the measurement standard of the edge width Ew and the standard deviation σ can be unified, and the resolution R can be calculated and obtained, and the measurement accuracy becomes stable.

Further, according to the present invention, as the line width Lew, the central frequency in the range w in which the above-mentioned minimum frequency Fm is obtained is obtained as a threshold value for binarization, and thereby the target image is judged as good or bad. Also, it is possible to prevent variations in measurement accuracy due to the influence of the ratio of the evaluation area of the image and the base portion,
The degree of freedom of measurement conditions is increased.

Further, the line width correction coefficient k3 corresponding to the line width Lew corresponds to the enlargement magnification of the means for optically enlarging the image.
By correcting the line width measured value Lew by
By correcting the pixel size corresponding to the magnification in the enlarging means of the imaging means such as a camera head, it is possible to obtain a measurement result close to the high-magnification linewidth measurement value even with the low-magnification linewidth measurement value. It becomes possible to avoid the adverse effect of the pixel size due to the magnification change and the adverse effect of the measurement system,
It is possible to measure the line width with high accuracy, and it is possible to eliminate the cause of the line width of the line width measurement value in the image pickup means such as a CCD camera by the pixel size correction.

As the edge width Ew, the width ew extending in the second direction when the minimum frequency Fm of the range w is obtained as described above is obtained, and the edge width Ew is calculated by this. It is possible to obtain stable measurement results at all times, prevent the badness of the target image and the adverse effect of the ratio of the image and the background portion in the window, and the shape of the histogram drawn depending on the distribution of the frequency greatly differs. Even in the image inside, the edge width Ew can be obtained with stable measurement accuracy.

Further, according to the present invention, the edge width Ew can be corrected so that the edge width can be measured with high accuracy even when the magnification of the enlarging means provided in the image pickup means is changed. The edge width Ew can be obtained with stable measurement accuracy.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of an embodiment of the present invention.

FIG. 2 is a flowchart for explaining the operation of a resolution processing circuit 7 shown in FIG.

FIG. 3 is a diagram showing a window 16 of image information stored in a memory 17.

FIG. 4 is a diagram for explaining resolution.

FIG. 5 is a diagram for explaining the operation of calibrating the gradation of the camera head 1.

FIG. 6 is a histogram showing the relationship between gradation and frequency in window 16.

FIG. 7 is a diagram showing a simplified version of the histogram shown in FIG.

FIG. 8 is a second view along the horizontal scanning line in the window 16.
It is a figure which shows the position and gradation of a direction.

FIG. 9 is a diagram for explaining an edge width of a window 16.

10 is a diagram for explaining an edge width Ew of a boundary 20. FIG.

FIG. 11 is a diagram showing a pixel 27 in the window 16.

FIG. 12 is a diagram showing in simplified form the gradation on a single horizontal scanning line in FIG. 8 described above.

13 is a diagram for explaining a linear regression line B0 of the linear image A. FIG.

FIG. 14 is a diagram for explaining an operation for calculating a standard deviation σ.

FIG. 15 is a window 16 for explaining the line width Lew.
FIG.

[Explanation of symbols]

 1 CCD camera head 2 Microscope 3 Light source 5 Light source means 6 Control unit 7 Resolution processing circuit 8 Image monitor 9 Arithmetic processing circuit 13 Imaging means 14 Stand 15 Original 16 Window 17 Memory 20, 21 Border B0 Primary regression line

Claims (8)

[Claims] 1. A window is composed of a plurality of pixels each having a plurality of gradations, and the window includes a boundary of an image to be binarized which extends in a predetermined first direction. The frequency for each density of each image is obtained, and the first density Ri of the mode value in which the frequency on the high density side is maximum,
The second density Rb, which is the most frequent value at which the frequency on the low density side is maximum, is obtained, and while moving the range w of the predetermined density difference between the first and second densities Ri and Rb, The frequencies Fi included in the range w are obtained, and the range w in which the minimum frequency Fm is obtained among the frequencies Fi obtained for each range w is detected, and the minimum frequency Fm is obtained in the window. The width ew extending in the second direction orthogonal to the first direction of the obtained range w is obtained as the edge width Ew, and the third density Rew having the central frequency in the range w having the minimum frequency Fm is set as the threshold value. Then, the boundary is binarized with the third density Rew as the threshold value, and the standard deviation σ of the deviation of the boundary in the second direction with respect to the regression line B0 along the first direction is calculated, and the resolutions R and R = 1 Resolution measurement characterized by finding / (Ew + σ) Law. 2. A document having a linear image extending linearly is imaged by an image pickup device, and the image pickup device is arranged in a matrix in a first direction and a second direction perpendicular to the first direction. It has a plurality of light receiving elements that configure each pixel that detects each gradation, stores the output of the light receiving element in a memory, reads the stored contents of the memory, obtains the frequency for each density, and The first density Ri of the mode having the maximum frequency on the side and the second density of the mode having the maximum frequency on the other density side
The density Rb is obtained, and the range w of gradation between the first and second densities Ri and Rb (where 0 <r <1), w = r · (Rb−Ri) is set to one or a plurality of predetermined levels. The frequency Fi included in each range w is calculated while moving in a stepwise manner, and the range w in which the minimum frequency Fm is obtained among the frequencies Fi calculated for each range w is detected. When the longitudinal direction of the contour image virtually coincides with the first direction, the width ew extending in the second direction of the range w in which the minimum frequency Fm is obtained is obtained in the window to obtain the edge width. Ew, Ew = ew / r is obtained, and the two values of the boundary are set with the third density Rew having the central frequency in the range w in which the minimum frequency Fm is obtained as the threshold value and the third density Rew as the threshold value. Conversion to
A resolution measuring method characterized in that the standard deviation σ of the deviation in the second direction with respect to the linear regression line B0 along the first direction of the boundary is obtained, and the resolutions R and R = 1 / (Ew + σ) are obtained. 3. When the unevenness correction coefficient is k1, the product σ · k1 of the standard deviation σ and the unevenness correction coefficient k1 is 50.
The unevenness correction coefficient k1 is determined so that
The method for measuring resolution according to claim 1 or 2, wherein k1 is used for calculating the resolution R. 4. An image pickup unit for picking up an image has a unit for optically enlarging the image, defines an edge width correction coefficient k2 corresponding to the enlargement magnification, and sets the edge width Ew to the edge width correction coefficient k2. 2. The subtraction is used only for calculation of the resolution R.
Alternatively, the resolution measuring method described in 2. 5. An original having a linear image extending linearly is picked up by an image pickup means, and the image pickup means is arranged in a matrix in a first direction and a second direction perpendicular to the first direction. It has a plurality of light receiving elements that configure each pixel that detects each gradation, stores the output of the light receiving element in a memory, reads the stored contents of the memory, obtains the frequency for each density, and The first density Ri of the mode having the maximum frequency on the side and the second density of the mode having the maximum frequency on the other density side
The density Rb is obtained, and the range w of gradation between the first and second densities Ri and Rb (where 0 <r <1), w = r · (Rb−Ri) is set to one or a plurality of predetermined levels. The frequency Fi included in each range w is calculated while moving in a stepwise manner, and the range w in which the minimum frequency Fm is obtained among the frequencies Fi calculated for each range w is detected. When the width Ew is the length L of the window in the first direction and the length a2 of the pixel in the second direction, Ew = a2 · Fm / L is calculated, and the edge width of the line image is measured. . 6. A document having an image on a line is imaged by an image pickup means, and the image pickup means is a first direction which is a vertical scanning direction parallel to the longitudinal direction of the image on the line, and a horizontal direction perpendicular to the first direction. It has a plurality of light receiving elements forming pixels each having a plurality of gradations arranged in a matrix in the second direction which is the scanning direction, and stores the output of the light receiving element in a memory. Is read out to find the frequency for each density, and the first density Ri of the mode having the maximum frequency on the high density side is the maximum.
And the second concentration Rb of the mode having the maximum frequency on the low concentration side.
And moving the range w, w = (Rb-Ri) / 2 between the first and second densities Ri and Rb by shifting one gradation at a time, and moving the frequency Fi within each range w. Of the frequencies Fi obtained for each range w, the range w in which the minimum frequency Fm is obtained is detected, and the third frequency having the center frequency in the range w in which the minimum frequency Fm is obtained is detected. When the number of pixels B of the line-shaped image that is equal to or greater than or equal to the third density Rew in the window is calculated using the density Rew as a threshold value and the length a2 of the pixel in the second direction is defined, line widths Lew, Lew = a2 A method for measuring the line width of a line image, which is characterized by calculating B / L. 7. An image pickup unit for picking up an image has a unit for optically enlarging the image, obtains a line width correction coefficient k3 corresponding to the enlargement magnification, and calculates only the line width correction coefficient k3 from the line width Lew. 7. The method for measuring the line width of a line image according to claim 6, wherein the line width is corrected by subtraction. 8. An image pickup unit for picking up an original having a linear image extending linearly, the first section of a column being a vertical scanning direction and the first row of a row being a horizontal scanning direction perpendicular to the first direction. An image pickup unit having a plurality of light receiving elements forming pixels for detecting a plurality of gradations arranged in a matrix in two directions, a memory for storing an output of the light receiving element of the image pickup unit, and a stored content of the memory are The frequency is read out to obtain the frequency for each density, and the first density Ri of the mode having the maximum frequency on one density side and the second density of the mode having the maximum frequency on the other density side.
The first and second concentrations R in response to the output of the concentration detecting means for obtaining the concentration Rb and the concentration detecting means.
The range w of gradation between i and Rb (where 0 <r <1), w = r · (Rb−Ri) is shifted by one or a plurality of predetermined gradations, and each range w A frequency detecting unit for detecting the frequency Fi included in each of the ranges, and detecting the range w in which the minimum frequency Fm is obtained among the frequencies Fi calculated for each range w; and responding to the output of the range detecting unit, When the longitudinal direction of the line-shaped image is virtually aligned with the first direction, the width ew extending in the second direction of the range w in which the minimum frequency Fm is obtained is obtained in the window to determine the edge. An edge width calculation means for obtaining the widths Ew and Ew = ew / r, and a third density Rew having a central frequency in the range w in which the minimum frequency Fm is obtained in response to the output of the range detection means.
As a threshold value, and a binarizing means for binarizing the boundary with the third density Rew as the threshold value, and a linear regression line along the first direction of the boundary in response to the output of the binarizing means. a standard deviation calculating means for obtaining a standard deviation σ of a deviation in the second direction with respect to b0; and a resolution R, R = 1 / (Ew + σ) in response to each output of the edge width calculating means and the standard deviation calculating means. Characteristic resolution measuring device.