JPH0998292A - Mtf measurement method and mtf correction device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce a measurement error due to noise in the case of MTF measurement by averaging plural line spread functions(LSF) in a frequency region obtained from a digital image so as to calculate an MTF. SOLUTION: When slit chart image data are given to an MTF measurement section 20, an LSF detection section 21 detects plural LSFs, a Composite LSF generating section 22 calculates plural Composite LSFs, which are subject to discrete Fourier transformation and the real part and the imaginary part are separately added. Then the added real part and imaginary part are separately subject to noise elimination processing. An MTF calculation section 23 calculates the MTF based on the real part and the imaginary part. Plural LSFs are measured from the image data and the result is averaged as to a frequency region to reduce the error due to noise.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an MTF measuring method for a digital image pickup system such as a scanner and an MTF correcting apparatus for correcting an MTF of an input image by using an MTF measurement value, and more particularly to an MTF due to noise superimposed on the input image. The present invention relates to a technique for reducing measurement error.

[0002]

2. Description of the Related Art Conventionally, MTF (Mo
duplication Transfer Function
n) As a measuring method, a method of obtaining from the amplitude of a sine chart or a line spread function from a knife edge chart or a slit chart; LSF (Line Spread Fu)
A method is known in which the Fourier transform is performed after the calculation of (n. However, in the digital imaging system, there is a problem that an accurate MTF cannot be calculated due to the aliasing effect of the sampled image data.

To solve this problem, for example, IEEE T
RANSATION ON MEDICAL IMA
GING, VOL. 11, NO. 1, the MTF measurement method of digital radiography described in Page 34-39 is known. This method is
When obtaining the SF, the image data is sampled by inclining the slit at an angle of 2 ° or less from an axis perpendicular to the main scanning direction as shown in FIG. 26 showing the positional relationship between the read pixel and the slit chart. As shown in FIG. 27, a plurality of LSFs out of phase are obtained. And Composite L
As shown in FIG. 28 showing SF, LSF is combined with image data smaller than actual sampling to remove the aliasing effect. Combining this synthesized LSF
It is called a position LSF, and its sampling interval Δ
X'is the LSF combined with the actual sampling interval ΔX
The number n is expressed by the equation (1).

ΔX '= ΔX / n (1) Further, the angle Θ of the slit is determined by n according to the equation (2). Θ = tan ⁻¹ (1 / n) (2) On the other hand, the MTF is obtained by the discrete Fourier transform of the Composite LSF and is obtained from the following equations (3) and (4).

[0005]

[Equation 1]

[0006]

[Equation 2]

Here, h (iΔX ') is Composi
te LSF, where L is the number of image data samples. Further, Hr (f) and Hi (f) represent the real part and the imaginary part of the discrete Fourier transform.

[0008]

However, when noise is superposed on the image read by the image pickup system in the reading optical system or the electric system, a measurement error occurs in the MTF.
To improve this measurement error, it is necessary to read the slit chart a plurality of times and obtain an average value, or to obtain an average value from a plurality of MTFs.
Since the position LSF is out of phase at each measurement position, it is practically extremely difficult to average the measurement values at the same phase position.

The optical image of the slit used for the measurement is ideally an impulse, but in reality, an optical image having a width is used, which deteriorates the MTF. The fact is that no measures have been taken in the prior art to deal with these defects. The present invention has been made in view of such conventional problems, and provides an MTF measuring method capable of reducing a measurement error due to noise at the time of MTF measurement and an MTF measuring method capable of correcting MTF deterioration due to an optical image of a slit. An object of the present invention is to provide an MTF correction device having the same.

[0010]

Therefore, according to the invention of claim 1, an M of a digital image pickup system having a photoelectric conversion element is provided.
In the TF measuring method, the digital imaging system averages a plurality of line spread functions (LSF) obtained from the digital image obtained by reading the slit chart in the frequency domain to calculate the MTF.

According to this, when obtaining the MTF, it is possible to reduce the error due to noise by measuring a plurality of LSFs from the image data obtained through the slit and averaging them in the frequency domain. In the invention according to claim 2, when the plurality of line spread functions (LSF) are averaged in the frequency domain, the plurality of line spread functions (LS) are calculated.
Each of F) is divided into a real number part and an imaginary number part and averaged.

According to this, since each of the line spread functions (LSF) is divided into a real part and an imaginary part for averaging, the noise component is canceled and the error due to noise can be reduced. According to the invention of claim 3, the measurement value deterioration due to the optical image is corrected by dividing the MTF measurement value by the frequency characteristic of the optical image from the slit chart.

According to this, by dividing the MTF measurement value by the frequency characteristic of the optical image from the slit chart, it is possible to correct the deterioration of the MTF due to the optical image of the slit. Further, according to the invention of claim 4, in an MTF correction device for performing MTF correction based on the MTF measured for the input image from the image input means and outputting it to the image output means, the digital image obtained by reading the slit chart is displayed. A plurality of line spread functions (LSF) obtained from the inside are averaged in a frequency domain to obtain an MTF measurement value of the image input unit, and a correction filter for MTF correction of the image output unit is stored. A correction filter storage means and a correction filter for the image input means are created from the MTF measurement value of the MTF measurement means, and the MTF correction is performed using the correction filter for the image input means and the correction filter for the image output means. And an image processing unit for calculating.

According to this, the MTF correction for both the image input means and the image output means is performed based on the correction filter of the image input means and the correction filter of the image output means according to the averaged MTF measurement value. It is possible to obtain an output image that faithfully reproduces the sharpness of the input image. According to a fifth aspect of the invention, the correction filter of the image input means is a spatial filter or a frequency filter.

According to this, the correction filter can be applied to the MTF correction of the image input means as a spatial filter or a frequency filter for performing multiplication in the frequency domain.

[0016]

DETAILED DESCRIPTION OF THE INVENTION An MTF correction device according to the present invention will be described below with reference to the drawings. Figure 1 shows MTF
It is a figure which shows schematic structure of the transmission type flat bed scanner which is an image input means of a correction | amendment apparatus. A transparent original such as a photographic film placed on the stage 1 is exposed from the upper side by an illumination 2, and the transmitted light is received while being scanned by a CCD line sensor 3 provided at the lower side, and then photoelectrically converted. In addition,
Two-dimensional image data is obtained by performing main scanning in the longitudinal direction of the CCD line sensor 3 and sub-scanning the CCD line sensor 3 by a drive system (not shown) in a direction perpendicular to the main scanning (shown by an arrow).

The photoelectrically converted electric signal is digitized by an A / D conversion circuit (not shown) and output as image data to an MTF correction device described later. First,
Prior to reading the transparent original, the MT of the image input means
F measurement is performed. FIG. 2 is a schematic block diagram of the image system apparatus, and the MTF correction apparatus will be described with reference to this figure.

In the field of medical images in which an X-ray photograph is read by an image input device such as a scanner, the sharpness of the input X-ray photograph and the sharpness of the output hard copy must match, and accurate frequency reproduction is possible. M required for high accuracy
It is very important to perform TF correction, and in this embodiment, an X-ray photograph is read by an image input device such as a scanner, gradation correction and sharpness correction are performed, and a hard copy is performed by an image output device such as a printer. A description will be given by taking as an example an image system device for creating

First, in the image input section 10, the slit chart 4 is formed.
And stores the image data in the image storage unit 11. The image data is transferred to the MTF measuring unit 20 and the MTF is calculated. Then, the above calculation is performed for main scanning, sub scanning, 45 °, 135 °
The image processing section 30 creates MTF correction filters from MTFs in four directions. FIG. 3 shows the MTF measuring unit 20.
3 is a detailed block diagram of an MTF correction device including an image processing unit 30.

When the image data of the slit chart 4 is input to the MTF measuring unit 20, the LSF detecting unit 21 outputs a plurality of LSFs.
, The Composite LSF creation unit 22 calculates a plurality of Composite LSFs and performs a discrete Fourier transform, and adds the real number part and the imaginary number part separately. Then, the added real number part and the imaginary number part are separately averaged to perform noise removal processing.

The MTF calculator 23 calculates the MTF based on the real part and the imaginary part. When the obtained MTF is input to the spatial filter creation unit 31 of the image processing unit 30, a spatial filter that is an MTF correction filter is created here. On the other hand, the MTF correction filter of the image output unit 12 is stored in the memory 33 in advance. This MTF correction filter can be obtained from a sample printed with image data of a rectangular wave chart or a Sine chart sent from an external terminal or the like (not shown).

Then, instead of the slit chart 4, the image data of the transparent original of the X-ray photograph is stored in the image storage unit 10 and then input to the image processing unit 30.
The MTF correction processing is performed by performing the convolution calculation in the convolution calculation unit 32 using the MTF correction spatial filter of the image input unit 10 created in 31 and the MTF correction filter of the image output unit 12 to output the image. It is output as a hard copy in section 12. As described above, the MTF correction for both the image input unit 10 and the image output unit 12 is performed based on the correction filter of the image input unit 10 and the correction filter of the image output unit 12 according to the averaged MTF measurement value. It is possible to obtain an output image that faithfully reproduces the sharpness of the input image.

As another embodiment, the processing mode of the image processing unit 30 may be changed as shown in the detailed block diagram of the MTF correction apparatus of FIG. That is, the frequency filter creation unit 34 is provided in place of the spatial filter creation unit 31, and MT
Using the MTF input from the F measurement unit 20, an MTF correction filter (frequency filter) for the image input unit 10 is created using frequency data. Then, the image data of the transparent original input to the image processing unit 30 is subjected to the discrete Fourier transform in the discrete Fourier transform operation unit 35, and the frequency filter and M of the image output unit 12 are processed.
Similar results can be obtained even if the inverse discrete Fourier transform is performed by the inverse discrete Fourier transform operation unit 37 after the multiplication by the TF correction filter.

FIG. 5 shows a flowchart of the MTF measuring means, and the measuring method will be described in detail along with the flowchart. In step (hereinafter abbreviated as S) 1, a slit chart is read. For this, the slit chart 4 having the slits 4a with a width of 7 μm shown in FIG. 6 is placed on the stage 1 of the image input means, and is tilted by an angle Θ with respect to the sub-scanning direction as shown in FIG. If this angle Θ is too small, the number of Composite LSFs decreases and the average number decreases, and conversely, if the angle Θ is too large, the MSF measurement value becomes non-uniform due to the phase shift of LSF. It is set between ° and 4 °.

Since the slit chart 7 has a density around the slit 4a, the light of the illumination 2 is not transmitted, but only the slit 4a portion is transmitted and is received by the CCD line sensor 3. . Therefore, the image data when the slit chart 7 is read becomes the image data as shown in FIG. Next, in S2 to S3, the strict angle Θ of the slit 4a is calculated to determine the number n of LSFs. In the past, the angle Θ of the slit 4a was determined from the number n of LSFs used for synthesis, but it is difficult to exactly match the angle of the slit 4a.
The method of obtaining n from is adopted. This can be expressed by equation (5).

N≈1 / tan Θ (5) When n is not an integer, it is converted to an integer by rounding. The calculation of the specific angle Θ will be described. Pixels corresponding to the positions of the slits 4a are plotted from the obtained digital image data on XY coordinates in which the main scanning direction is the Y axis and the sub scanning direction is the X axis. A regression line is obtained for the digital image data, and the angle Θ of the slit 4a is obtained from the inclination. FIG. 8 shows digital image data plotted on the XY coordinates. From this figure, the equation of the regression line in this embodiment is Y = −0.035189X + 89.195, and the absolute value of the slope is tan Θ. . That is, ta
Since n Θ = 0.035189, the number of LSFs is n from equation (5).
= 28.

Next, in S4 to S8, the MTF measurement position is detected, and the averaged MTF is calculated while counting the number of measurements. This is a diagram showing the pixel values on the slit 4a plotted in FIG. 9, and the position where the waveform has a peak is detected from this. This is the MTF measurement position, and 28 lines of LSF centering on the peak are combined to create a Composite LSF. Since the number of lines at the first and last peak positions is less than 28 lines in FIG. 9, Composite LSF is created for all the second to 26th peak positions.
In other words, set the number of measurements to 25 and
e LSF is created, and each Composi is calculated by the equation (3).
te LSF is subjected to discrete Fourier transform, and the real number part and the imaginary number part are added separately. Then, each time the parameter i is set to 1 and similar processing is repeated, the value of the parameter i is incremented (added). Thus the parameter i is 25
When it becomes larger, the process proceeds to S9. In addition, FIG. 10 shows Composite LS created from the second peak position.
F is shown.

In steps S9 to S10, averaging is performed to reduce measurement errors due to noise. The added real number part and imaginary number part are averaged separately, and the MTF is calculated based on the equation (4). As described above, normally, the noise component does not disappear even if the MTF represented by the absolute value is averaged, but in the present embodiment, when obtaining the MTF, a plurality of LSFs are measured from the image data obtained through the slit 4a. By dividing and averaging the real part and the imaginary part in the frequency domain, the noise component can be canceled and the error in the MTF measurement value due to noise can be reduced. FIG. 11 shows the MTF calculated by the above method.

Here, a simulation for showing the effect of averaging will be described. An LSF as shown in FIG. 12 is created by the Gaussian function, and 6 dB of white noise is added to this. This is shown in FIG. 20 in the same way
When individual LSFs are created and individual MTFs are obtained, FIG. 14 is obtained. In FIG. 14, 10 MTFs are shown, but on the other hand, when the MTFs are calculated from 20 LSFs by averaging,
It becomes 15 and it can be confirmed that the averaging improves the MTF measurement accuracy.

FIG. 16 shows a flow chart of the MTF correction processing, and the deterioration correction processing of the MTF measurement value will be described in detail along the flow chart. In this embodiment, the case where the slit 4a of 7 μm is used is described. S11
Then, the width data of the slit 4a, that is, 7 μm is input. Then, from S13 to S21, the same processing as the averaging processing shown in the flowchart of FIG. 5 is performed to calculate the MTF.

In S22, the deterioration characteristic of the MTF is calculated from the Sinc function. That is, the optical image of the slit 4a is assumed by a rectangular function as shown in FIG. 17, and the frequency characteristic S (f) thereof is calculated from the Sinc function of the equation (6). S (f) = sin (πdf) / (πdf) (6) Here, d represents the width of the slit, and d = 0.007.

FIG. 18 shows the deterioration characteristics due to the slits. When the frequency characteristics are expressed up to the Nyquist frequency of the image input device, it can be seen that the MTF measurement value is lowered (deteriorated). In S23, the MTF measurement value is corrected with this frequency characteristic. This is calculated based on the equation (7).

MTF = MTF _meas (πdf) / S (f) (7) Here, MTF _meas represents the MTF measurement value. That is, by dividing by the frequency characteristic of the optical image of the slit 4a, the MTF measurement value is corrected in the frequency domain, and the MTF deterioration due to the optical image of the slit can be corrected.
FIG. 19 shows the Composite obtained from the slit 4a.
7 shows LSF, and this Composite LSF
In FIG. 20, which illustrates the comparison of MTF correction by performing a discrete Fourier transform on the signal and dividing it by S (f) shown in FIG. 18, the frequency after correction shown by the dotted line is compared with the frequency after correction shown by the solid line. It can be seen that the characteristics are slightly higher.

Although the MTF measurement in the main scanning direction has been described in this embodiment, the measurement in the sub-scanning direction and the oblique 45 ° and 135 ° directions can be performed in the same manner. That is, in the measurement in the sub-scanning direction, the direction of the arrow in FIG. 26 is the sub-scanning direction, and the slit 4a is tilted with respect to the main scanning direction. In the measurement in the oblique 45 ° direction, the direction of the arrow is the main scanning direction as shown in FIG.
In the measurement in the 5 ° direction, the direction shown in the figure is the sub-scanning direction, and the slit 4a is tilted from the oblique 45 ° axis.

Then, similarly to the above, the LSF is obtained in the direction in which the MTF is measured, and the MTF is calculated from the Composite LSF. Further, in the present embodiment, the MTF is obtained from one slit image, but the slit chart 4 is read a plurality of times and the LSF is obtained from all digital image data.
Seeking. It is also possible to average the MTF. Further, it is also possible to fix the position of the slit chart 4 and average the digital image data read a plurality of times to obtain the MTF.

Further, although the slit chart 4 is used as the test chart in this embodiment, the MTF in the main scanning direction may be measured using a knife edge chart as shown in FIG. 22. In this case, the test chart is used. Can be easily manufactured. As a measuring method, as shown in FIG. 26, the stage 1 is tilted by 1 ° to 4 ° with respect to the sub-scanning direction.
Then, the knife edge chart is read and digital image data is output. FIG. 23 shows this digital image data. Here, the X and Y axes are sub-scans,
The main scanning direction is shown. If this digital image data is taken as one line on the Y axis, an edge spread function (ESF) as shown in FIG. 24 is obtained. When the ESF is differentially calculated by the equation (8), an LSF as shown in FIG. 25 is obtained.

G (iΔX) = e (iΔX) −e ((i + 1) ΔX) (8) Here, e (iΔX) represents ESF and g (iΔX) represents LSF. This process can be performed for all the lines in the X-axis direction to obtain the MTF. For the measurement in the sub-scanning direction and the diagonal 45 ° and 135 ° directions, the knife edge chart may be tilted.

Further, in this embodiment, the MTF of the medical image is
Although the correction device has been described, the present invention is not limited to this, and the present invention can be applied to fields other than medical images for performing MTF correction, and MTF measurement methods and image processing devices for reflective originals.

[0039]

As described above, according to the first aspect of the invention, when the MTF is obtained, a plurality of LSFs are measured from the image data obtained through the slits and averaged in the frequency domain to reduce noise. It is possible to reduce the error due to. According to the second aspect of the present invention, since each line spread function (LSF) is divided into a real part and an imaginary part for averaging, the noise component is canceled and the error due to noise can be reduced.

According to the third aspect of the invention, it is possible to correct the deterioration of the MTF due to the optical image of the slit.
According to the invention of claim 4, the averaged MTF
An output in which the sharpness of the input image is faithfully reproduced by performing MTF correction on both the image input unit and the image output unit based on the correction filter of the image input unit and the correction filter of the image output unit according to the measured value. Images can be obtained.

According to the invention of claim 5, the correction filter can be applied to the MTF correction of the image input means as a spatial filter or a frequency filter for performing multiplication in the frequency domain.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of a transmission type flat bed scanner according to the present embodiment.

FIG. 2 is a schematic block diagram of an image system device according to the present embodiment.

FIG. 3 is a detailed block diagram of an MTF correction device according to this embodiment.

FIG. 4 is a detailed block diagram of an MTF correction device according to another embodiment.

FIG. 5 is a flowchart of MTF measuring means in the present embodiment.

FIG. 6 is a diagram showing a slit chart in the present embodiment.

FIG. 7 is a diagram showing image data when a slit chart according to the present embodiment is read.

FIG. 8 is a diagram showing digital image data plotted on XY coordinates in the present embodiment.

FIG. 9 is a diagram showing plotted pixel values on a slit in the present embodiment.

FIG. 10 is a Composite LS according to the present embodiment.
Diagram showing F

FIG. 11 is a diagram showing a calculated MTF according to the present embodiment.

FIG. 12 is an LS assumed with a Gaussian function in the present embodiment.
Figure of F

FIG. 13 shows L when noise is superimposed in the present embodiment.
SF diagram

FIG. 14 is a diagram for comparing MTFs with and without noise in the present embodiment.

FIG. 15 is an MTF obtained by averaging in the present embodiment.
Figure comparing

FIG. 16 is a flowchart of MTF correction processing according to the present embodiment.

FIG. 17 is a diagram illustrating an optical image of a slit in the present embodiment.

FIG. 18 is a diagram showing deterioration characteristics due to a slit in the present embodiment.

FIG. 19 shows Co obtained from the slit in the present embodiment.
Diagram showing mposite LSF

FIG. 20 is a diagram illustrating comparison of MTF correction according to the present embodiment.

FIG. 21 is a diagram showing an angle of a slit when measuring an oblique direction in the present embodiment.

FIG. 22 is a diagram showing a knife edge chart in the present embodiment.

FIG. 23 is a diagram showing image data of a knife edge chart in the present embodiment.

FIG. 24 is a diagram showing an ESF in the present embodiment.

FIG. 25 is a diagram showing an LSF according to the present embodiment.

FIG. 26 is a diagram showing a positional relationship between a conventional reading pixel and a slit chart.

FIG. 27 is a diagram showing a conventional LSF.

FIG. 28 is a diagram showing a conventional Composite LSF.

[Explanation of symbols]

 1 Stage 2 Illumination 3 CCD Line Sensor 4 Slit Chart 10 Image Input Section 11 Image Storage Section 12 Image Output Section 20 MTF Measurement Section 30 Image Processing Section

Claims (5)

[Claims] 1. A method for measuring MTF of a digital image pickup system having a photoelectric conversion element, wherein a plurality of line spread functions (LSF) obtained from a digital image obtained by reading the slit chart by the digital image pickup system are obtained in a frequency domain. An MTF measuring method, characterized in that the MTF is calculated by averaging. 2. A plurality of line spread functions (LS) when averaging the plurality of line spread functions (LSF) in the frequency domain.
The MTF measuring method according to claim 1, wherein each of F) is divided into a real part and an imaginary part and averaged. 3. The measurement value deterioration due to the optical image is corrected by dividing the MTF measurement value by the frequency characteristic of the optical image from the slit chart.
Alternatively, the MTF measuring method according to claim 2. 4. An MTF correction device for performing MTF correction based on the measured MTF of an input image from the image input means and outputting it to the image output means, which is obtained from a digital image obtained by reading a slit chart. A plurality of line spread functions (LSF) are averaged in the frequency domain to obtain the MTF measurement value of the image input means M.
TF measurement means, a correction filter storage means for storing a correction filter for MTF correction of the image output means, a correction filter of the image input means is created from the MTF measurement value of the MTF measurement means, and An MTF correction device comprising: an image processing unit that performs an MTF correction calculation using the correction filter of the image input unit and the correction filter of the image output unit. 5. An MTF correction device, wherein the correction filter of the image input means is a spatial filter or a frequency filter that performs multiplication in the frequency domain.