JPH0855209A - Method and device for radiation image processing - Google Patents

Abstract

PURPOSE:To obtain the method and device for image processing which performs inspection processing conforming with the actual shape of a subject by making a radiation image which is picked up corresponding to the shape of a subject. CONSTITUTION:The analog output from a radiation image pickup system is processed by an image processing system and the digital image is stored in a memory 1; and a correction expression or correction coefficient for correcting transmission characteristics of the image pickup system and image processing system so that gain characteristics become nearly constant with a specific spatial frequency range is found and stored in a memory 3; and a CPU 2 performs correction operation for the respective pixels of the digital image by using the correction expression or correction coefficient. This constitution performs the inspection processing conforming with the actual shape.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a radiation image processing method and apparatus, and more particularly to an image processing method applied to a digital image.

[0002]

2. Description of the Related Art In general, a radiation, especially an X-ray image has an excellent characteristic in principle that density information according to the thickness of a subject can be obtained. However, due to the response characteristic of the radiation detector and the transfer characteristic of the signal processing system, the gain characteristic in the high spatial frequency portion is poor (see the response characteristic example of the image intensifier in FIG. 11). Therefore, unlike the relationship between the density information at a high spatial frequency and the thickness of the object, and the relationship between the density information at a low spatial frequency and the thickness of the object, an inspection based on the actual shape is performed using a radiation image. I have a problem above. Therefore, various correction methods have been developed.
For example, JP-A-2-81276 and JP-A-2-81276.
In Japanese Patent No. 166882, the response characteristic of a logarithmic amplifier which is a signal processing circuit is corrected. In addition, JP-A-3-20945
In Japanese Patent No. 0, the response delay of stimulated emission as a detector is compensated by differentiating. In addition, an image sharpness correction method using an edge enhancement method such as differential processing or Laplacian processing is known. Here, the spatial frequency is a concept in which the frequency of a temporal signal is applied to a spatial pattern. For example, a pattern having a sharper shade change has a higher spatial frequency.

[0003]

The conventional radiation image processing method as described above has the following problems. (1) Japanese Patent Application Laid-Open No. 2-81276 and Japanese Patent Application Laid-Open No. 2-1
In Japanese Patent No. 66882, the response characteristic of the logarithmic amplifier, which is a signal processing circuit, is corrected, but the response characteristic of the detector is not taken into consideration. (2) In Japanese Patent Laid-Open No. 3-209450, the response delay of photostimulated luminescence, which is a detector, is compensated by differentiating, but the response characteristic of the detector is generally a non-linear element, which is a linear element. It is difficult to obtain an image having a uniform gain characteristic over the required spatial frequency domain only by the differential processing. (3) The other methods also have the same problems as the above (1) and (2). The present invention has been made in view of the above circumstances, and an object thereof is an image processing method capable of performing inspection processing in accordance with an actual shape by associating a captured radiation image with the shape of a subject. And to provide the device.

[0004]

In order to achieve the above object, the first invention is a radiation image processing method for obtaining a digital image by processing an analog output from a radiation image capturing system by an image processing system. The radiation image processing method is characterized in that the transfer characteristics of the image pickup system and the image processing system are corrected so that the gain characteristics become substantially constant within the spatial frequency range. Also, the second
In a radiation image processing method for obtaining a digital image by processing an analog output from a radiation image capturing system by an image processing system, the invention of claim 1 and a gain of frequency characteristics for a digital image of a subject of a known shape having a predetermined spatial frequency, Radiation characterized by obtaining a gain characteristic by multiplying with a target gain, and correcting the transfer characteristics of the image capturing system and the image processing system so that the gain characteristic becomes substantially constant within a predetermined spatial frequency range. This is an image processing method. Furthermore, it is a radiographic image processing method for performing a correction calculation for each pixel of the digital image using a correction equation of the transfer characteristic such that the gain characteristic is substantially constant within the predetermined spatial frequency range.

A third aspect of the present invention is a radiographic image processing method for obtaining a digital image by processing an analog output from a radiographic image pickup system by an image processing system to obtain a digital image of a subject having a known spatial frequency. The data and the target data are converted to the spatial frequency domain to obtain the spatial frequency components of both data, and the ratio of the spatial frequency components of both data is adjusted so that the gain characteristic becomes substantially constant within the above-mentioned predetermined spatial frequency range. Compute a correction factor consisting of
The radiation image processing method is characterized in that a correction calculation is performed for each pixel of the digital image using the correction coefficient. Furthermore, by multiplying the correction coefficient by the spatial frequency component obtained by converting the digital image data of the object of known shape having the predetermined spatial frequency into the spatial frequency domain, and inversely transforming the multiplication result, The radiation image processing method is characterized by performing a correction calculation for each pixel of the digital image. Furthermore, in the radiation image processing method, the correction coefficient is set to zero at the predetermined spatial frequency or higher. Furthermore, it is a radiographic image processing method for providing target data of an object of known shape having the above-mentioned predetermined spatial frequency with an MTF chart.

A fourth aspect of the present invention is a radiographic image processing apparatus for obtaining a digital image by processing an analog output from a radiographic image pickup system by an image processing system, wherein a digital image of a subject having a predetermined spatial frequency is formed. Transforming means for transforming the data and the target data into the spatial frequency domain to obtain respective spatial frequency components of both data, and each space for both data so that the gain characteristic is substantially constant within the predetermined spatial frequency range. A calculation means for calculating a correction coefficient composed of a ratio of frequency components, a storage means for storing the correction coefficient, a multiplication means for multiplying the correction coefficient by each spatial frequency component of the digital image data, and a result of the multiplication. The radiation image processing apparatus is characterized by comprising an inverse conversion means for performing an inverse conversion. Further, a fifth invention is a radiation image processing method for obtaining a digital image by processing an analog output from a radiation image pickup system by an image processing system, wherein the digital image is cut out in a predetermined block unit to create block data. , The block data is converted to the spatial frequency domain to obtain each spatial frequency component, and a correction coefficient is applied to each spatial frequency component of the block data so that the gain characteristic is substantially constant within the predetermined spatial frequency component range. The radiation image processing method is characterized in that a correction operation is performed on each pixel of the digital image by performing multiplication, inversely transforming the multiplication result, and synthesizing. Furthermore, it is a radiation image processing method in which the block data is data cut out by overlapping adjacent blocks. Further, the radiographic image is characterized in that the overlap width of the adjacent blocks is determined so that the difference in image density at the block boundary when the correction calculation is performed for each pixel of the digital image falls within a predetermined error range. It is a processing method.

According to a sixth aspect of the present invention, in a radiation image processing apparatus for obtaining a digital image by processing an analog output from a radiation image pickup means by an image processing system, a digital image of a subject is cut out in a predetermined block unit and block data is extracted. Block data generating means for converting the block data and the target data into a spatial frequency domain to obtain each spatial frequency component of both data, and a gain characteristic within the predetermined spatial frequency component range. A calculation means for calculating a correction coefficient composed of a ratio of the spatial frequency components of both data so as to be substantially constant, a storage means for storing the correction coefficient, and a correction coefficient for each spatial frequency component of the block data. A multiplication means for multiplying by, an inverse transformation means for inverse transformation of the multiplication result, and a synthesis of the inverse transformation result. And by a radiographic image processing apparatus characterized by comprising comprises a correction means for performing correction calculation for each pixel of the digital image. Furthermore,
In the radiation image processing apparatus, the block data created by the block data creating means is data cut out by overlapping adjacent blocks. Further, the block data creation means determines the overlap width of the adjacent blocks so that the difference in image density at the block boundary when the correction calculation is performed on each pixel of the digital image falls within a predetermined error range. A radiation image processing apparatus including an overlap width determining unit.

[0008]

According to the first aspect of the present invention, when the analog output from the radiation image pickup system is processed by the image processing system to obtain a digital image, the gain characteristic is substantially constant within a predetermined spatial frequency range. The transfer characteristics of the image capturing system and the image processing system are corrected. In this case, since the gain characteristics in a predetermined spatial frequency range are corrected to be substantially constant for subjects having various spatial frequencies, the shape of the subject and the density information can be associated with each other. According to the second invention, when the analog output from the radiation image pickup system is processed by the image processing system to obtain a digital image, the gain of the frequency characteristic of the digital image of the object of a known shape having a predetermined spatial frequency is obtained. And the target gain are multiplied to obtain a gain characteristic, and the transfer characteristics of the image pickup system and the image processing system are corrected so that the gain characteristic becomes substantially constant within a predetermined spatial frequency range. In this way, the radiation image capturing system and the image processing system are combined by approximating the relationship between them by inputting the digital image of the object of which the shape is known in advance and having the predetermined spatial frequency and outputting the object shape. It is possible to easily obtain a correction expression of the transfer characteristic including the. Further, the correction calculation for each pixel of the digital image is performed by using the transfer characteristic correction equation such that the gain characteristic is substantially constant within the predetermined spatial frequency range. In this way, the correspondence between the shape of the subject and the density information can be easily obtained.

According to the third and fourth inventions, when the analog output from the radiation image pickup system is processed by the image processing system to obtain a digital image, digital image data of a subject having a known shape having a predetermined frequency. And the target data are converted into the spatial frequency domain to obtain the respective spatial frequency components of both data. A correction coefficient composed of the ratio of each spatial frequency component of both data is calculated so that the gain characteristic is substantially constant within the predetermined spatial frequency range. A correction calculation is performed on each pixel of the digital image using the correction coefficient. In this way, since a digital image of a subject whose shape is known in advance having a predetermined spatial frequency and the subject shape are converted into a spatial frequency domain to obtain a correction coefficient, the radiation image capturing system and the image processing system are included. The correction coefficient is obtained for each spatial frequency. Further, the digital image data of the object of known shape having the predetermined spatial frequency is converted into the spatial frequency domain and the obtained spatial frequency component is multiplied by the correction coefficient, and the multiplication result is inversely transformed, A correction calculation is performed on each pixel of the digital image. In this way, since the digital image is converted into the spatial frequency domain for correction, the spatial frequency band in which the gain characteristic is substantially constant can be easily set. Furthermore, above the predetermined spatial frequency, if the correction coefficient is set to zero, image noise is removed. As a result, an image with less noise and having substantially constant gain characteristics within a predetermined spatial frequency range is restored. Further, the target data of the object of known shape having the above-mentioned predetermined spatial frequency is MT
When given as an F chart, this MTF chart is manufactured with high precision in spatial frequency and thickness, so that a precise correction coefficient or equation can be obtained.

According to the fifth and sixth aspects of the invention, when the analog output from the radiation image pickup system is processed by the image processing system to obtain a digital image, the digital image is cut out in a predetermined block unit and is divided into blocks. Data is created.
The block data is converted into the spatial frequency domain to obtain each spatial frequency component. Each spatial frequency component of the block data is multiplied by the correction coefficient so that the gain characteristic becomes substantially constant within the predetermined spatial frequency range. A correction calculation is performed on each pixel of the digital image by synthesizing the results of the multiplication after the inverse conversion. In this way, by cutting out the digital image in block units, correcting in the spatial frequency domain, and combining the correction results, the pixels near the block boundaries become common with the adjacent blocks, so that the adjacent blocks cross the block boundaries. And become correlated. As a result, the problem of image quality deterioration near the block boundary can be solved and a simpler corrected image with high sharpness can be obtained. Furthermore, by using the block data as data cut out by overlapping adjacent blocks, the correlation between pixels beyond the block boundary becomes more remarkable. Further, if the overlap width of the adjacent blocks is determined so that the difference in image density at the block boundary when the correction calculation is performed on each pixel of the digital image falls within a predetermined error range, the block width exceeds the block boundary. The correlation between the pixels is most remarkable. As a result, it is possible to obtain an image processing method and apparatus that can perform an inspection process according to the actual shape by associating the captured radiation image with the shape of the subject.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments embodying the present invention will be described below with reference to the accompanying drawings for the understanding of the present invention. The following embodiments are examples of embodying the present invention and are not intended to limit the technical scope of the present invention. 1 is a block diagram showing a schematic configuration of a radiation image processing system A ₁ according to one embodiment of the present invention, and FIG. 2 is a schematic configuration of a radiation image processing system A ₂ according to another embodiment of the present invention. The block diagram and FIG. 3 show the radiation image processing system A.
₁ , a comparison diagram of the density information of the image obtained by A ₂ and the density information of the required image, FIG. 4 is a flow chart showing the processing procedure for obtaining the correction formula, and FIG. 5 is a flow showing the processing procedure of the correction calculation. FIG. 6, FIG. 6 is a flow chart showing the processing procedure for deriving a correction coefficient, FIG. 7 is a flow chart showing each processing procedure for deriving a correction coefficient and correction calculation, and FIG. 9 is a flow chart showing a processing procedure for image correction, and FIG. 10 is a flow chart showing a processing procedure for obtaining the overlap width.

Each of the first to sixth inventions is similar to the conventional example in that an analog output from the radiation image pickup system is processed by an image processing system to obtain a digital image. However,
The first aspect of the invention differs from the conventional example in that the transfer characteristics of the image capturing system and the image processing system are corrected so that the gain characteristics are substantially constant within a predetermined spatial frequency range. Further, the second invention is positioned as a more specific form of the first invention. That is, here, when the digital image is obtained, the gain characteristic is obtained by multiplying the gain of the frequency characteristic of the digital image of the object of a known shape having a predetermined spatial frequency by the target gain, and the gain characteristic is set to the predetermined value. The transfer characteristics of the image capturing system and the image processing system are corrected so as to be substantially constant within the spatial frequency range, but this is different from the conventional example. Further, the correction calculation for each pixel of the digital image may be performed by using the transfer characteristic correction expression such that the gain characteristic is substantially constant within the predetermined spatial frequency range. Different from FIG. 1 shows a radiation image processing system A _{1 according} to an embodiment of the first and second inventions. Here, the digitized radiation image before correction can be obtained by a well-known technique such as A / D conversion of an analog signal output via a radiation detector such as an image intensifier. The description is omitted. Further, FIG. 1 is an example of a configuration in which the correction of the transfer characteristic is realized by software, and a memory 1 for storing a digital image obtained by radiography in the figure, a CPU 2 for performing a correction calculation, and a description given below. And a memory 3 for storing a correction formula or a correction coefficient.

First, the derivation of the correction formula will be described. A subject (corresponding to a subject having a known shape) whose shape (thickness) is known in advance having a predetermined spatial frequency is radiographed to obtain a digital radiation image thereof. For example, a subject having a rectangular cross-section has a spatial frequency of 0 to ∞, but the density information of an image obtained by radiographing the subject having the rectangular cross-section has an X-ray imaging system (corresponding to a radiation image capturing system) and signal processing. The transfer characteristic of the system (corresponding to the image processing system) is as shown in FIG. That is, compared to the required density information (a), the density information has edge portions which are blunted as shown in (b).
The sagging of the edge portion is mainly due to the response characteristics of the radiation detector, and changes depending on the type of the detector. Hereinafter, a method for obtaining the correction formula for the required density information (a) from the density information (b) will be described. The density information (b) for each pixel is represented by U = (..., u- _{n, -n} , ..., u _0,0 , ..., un _{, n} , ...) (1) It is assumed that the information (a) is represented by Y = (···, y _{−n, −n} , ···, y _0,0 ··, y _{n, n} , ···) (2). In order to formulate the response characteristic from the concentration information (a) to (b), the following equation is applied and its coefficient (a- _{m, m} , ..., a _0,0 , ..., am _{, m} ) May be calculated by the method of least squares. The order m may be determined, for example, so that the least-square error and the error between pixels are small and appropriate values. By this calculation, the following response characteristic formula is obtained. u _{k, k} = a- _{m, m} y- _{m, m} + ··· + a _0,0 y _0,0 + ··· + am _{, m} y _{m, m} (3) The above formula (3) is converted into Z conversion format. Then, it can be expressed as follows. u _{k, k} = a _{-m, m} z ₁ ^-m z ₂ ^-m + ... + a _0,0 z ₁ ⁰ z ₂ ⁰ + ... + am _{, m} z ₁ ^m z ₂ ^m (4) In the above formula (4), z ₁ and z ₂ correspond to the vertical and horizontal directions of the image, respectively.

On the other hand, the correction from the above equation (1) to the equation (2)
As for the formula, the same as above, the following formula is applied to each pixel.
Therefore, the coefficient (b_{-L, -L}, ..., b_0,0, ...
b_{L, L}) Is calculated by the least squares method. y_{k, k}= B_{-L, -L}u_{-L, -L}+ ・ ・ + B_0,0u_0,0+ ・ ・ + B_{L, L}u_{L, L} ... (5) Express the equation (5) in the Z conversion format as in the equation (3).
For example, y_{k, k}= B_{-L, -L}z₁ ^-Lz₂ ^-L+ ・ ・ + B_0,0z₁ ⁰z₂ ⁰+ ・ ・ ＋ b_{L, L}z₁ ^Lz₂ ^L (6) Here, the order L of the above equation (5) is determined as follows.
U That is, equation (5) is applied to the least squares method for some L.
The obtained coefficient (b_{-L, -L}, ..., b _0,0,
., B_{L, L}), The gays of the above equations (4) and (6)
Characteristics are shown on the Bode diagram, and the equations (4) and (5)
The characteristics multiplied by the IN characteristics are almost equal within the specified spatial frequency range.
It suffices to obtain an order L that is constant. The above processing procedure
Is shown in FIG.

In FIG. 1, the memory 3 storing the correction coefficient has a coefficient (b- _{L , -L} , ..., B _0,0 ,
.., b _{L, L} ) are stored, and the density information U of each pixel is sequentially stored from the memory 1 in which the radiation image is stored.
_,, u _{-n, -n} , ..., u _0,0 ,. ． , B _{n, n} , ...) Is called, the CPU 2 executes the calculation of the above formula (5) to perform the correction. The above processing procedure is shown in FIG. Thus, according to the first invention,
Since the gain characteristics within a predetermined spatial frequency range are corrected to be substantially constant for objects having various spatial frequencies, the shape of the object and the density information can be associated with each other. Also,
According to the second aspect of the present invention, a radiation image capturing system is provided by approximating the relationship between the digital image of a subject of which the shape is known in advance having a predetermined spatial frequency and the subject shape as an output by an approximate expression. A transfer characteristic correction equation including the image processing system can be easily obtained. Further, since the correction calculation for each pixel of the digital image is performed by using the correction equation of the transfer characteristic such that the gain characteristic is substantially constant within the predetermined spatial frequency range, the shape of the subject and the density information are Correspondence can be taken easily.

Next, the third invention will be described. In the third invention, when the digital image is obtained, the digital image data of a known object having a predetermined spatial frequency and the target data are converted into the spatial frequency domain to obtain the spatial frequency components of both data. , A correction coefficient consisting of the ratio of the spatial frequency components of both data is calculated so that the gain characteristic is substantially constant within the predetermined spatial frequency range, and the correction coefficient is used to correct each pixel of the digital image. It differs from the conventional example in that calculation is performed. That is, the third
In the invention described above, the steps up to the X-ray imaging of a subject (corresponding to a subject having a known shape) whose shape (thickness) is known in advance are the same as those of the second invention, but here, the X-ray density information U
And Y are converted into the spatial frequency domain, the gain for each spatial frequency is obtained, and the correction coefficient is calculated. As a method of converting the X-ray density information into a spatial frequency, there are a digital Fourier transform (hereinafter referred to as DFT) and a discrete cosine transform (hereinafter referred to as DCT), which are expressed as follows.

[Equation 1] Here, the density information U and Y are given by the above equation (7) or (8).
Apply to the formula and calculate the gain for each spatial frequency. _Assuming that the gains of the density information U and Y for a certain spatial frequency ω _K are u _kl and y _kl (both correspond to F _kl ), the correction coefficient K _kl that becomes substantially constant within the predetermined spatial frequency range is It is represented by. K _kl = y _kl / u _kl (9) Correction calculation is performed using the correction coefficient K _kl obtained by performing the above calculation. The processing procedure is shown in FIG.

As described above, according to the third aspect of the present invention, since the digital image of the object having a predetermined spatial frequency and the shape of which is known in advance and the object shape are converted into the spatial frequency domain to obtain the correction coefficient, A correction coefficient including the radiation image capturing system and the image processing system is obtained for each spatial frequency. Furthermore,
If the digital Fourier transform or the discrete cosine transform is used for the above conversion, the above coefficient can be obtained with high accuracy. Further, by multiplying the correction coefficient by the spatial frequency component obtained by converting the digital image data of the known shape object having the predetermined spatial frequency into the spatial frequency domain, and inversely converting the multiplication result, A correction calculation may be performed on each pixel of the digital image. This method can be realized with the same configuration as in FIG. However, the difference from the above-described method of the third invention is the content of the correction coefficient and the correction calculation, which will be described in detail below. First, a block image of n × n pixels is read out from the memory 1 in which a radiation image is stored, and the conversion processing into the spatial frequency domain is executed via the CPU 2 by either of the equations (7) and (8). , _Derive the gain A _kl for each spatial frequency. On the other hand, the memory 3 in which the correction coefficient is stored stores the correction coefficient K _kl corresponding to each spatial frequency of the equation (9) calculated by either the equation (7) or the equation (8). . As the correction calculation, the CPU 2 first performs the following calculation for each spatial frequency. F _kl = A _kl × K _kl (10)

Next, in order to reproduce the corrected radiation image,
Perform the following inverse transformation.

[Equation 2] The above calculation is carried out for all pixels in n × n pixel block units, and the processing procedure is shown in FIG. 7. In this way, since the digital image is converted into the spatial frequency domain and the correction is performed, it is possible to easily set the spatial frequency that makes the gain characteristic substantially constant. Further, by using the transform digital Fourier transform and the inverse digital Fourier transform as the inverse transform, the software by the CPU 2 can perform the digital Fourier transform (DFT) / inverse digital Fourier transform (IDF).
Since the T) process is performed, the above correction calculation can be realized without requiring any specific hardware. Furthermore, if a discrete cosine transform is used for the above transform and an inverse discrete cosine transform is used for the above inverse transform, a commercially available discrete cosine transform (DCT) / inverse discrete cosine transform (IDCT) IC used for image compression or the like is applied. Therefore, the correction calculation can be realized in real time. FIG. 2 shows a system configuration example A ₂ using a commercially available hardware chip. For the conversion to the spatial frequency domain, various methods such as the Karhunen-Leve conversion have been proposed in addition to the above, and any conversion processing may be performed. Further, above the predetermined spatial frequency, the correction coefficient K _kl may be zero. That is, for frequencies above a predetermined spatial frequency, the image noise is removed by setting the correction coefficient K _kl = 0 from the viewpoint of image noise removal, and the image noise is substantially constant within the predetermined spatial frequency range. An image having a gain characteristic can be restored.

Further, the target data of the object of known shape having the predetermined spatial frequency may be given by an MTF chart. Here, MTF means MODULAR TRAN
It is an abbreviation for SFER FUNCTION, and the MTF chart corresponds to a gauge for measuring the radiation transfer characteristic. Since this MTF chart is manufactured with good spatial frequency and thickness, it is possible to obtain an accurate correction coefficient or formula. Further, if the correction calculation is a product-sum calculation of each coefficient of the correction equation and each pixel of the digital image, an image having a substantially constant gain characteristic within a predetermined spatial frequency range can be accurately restored. It The fourth invention is an apparatus to which the method of the third invention can be applied. In this device, when the analog output from the radiation image pickup system is processed by the image processing system to obtain a digital image, the digital image data and the target data of the known shape subject having a predetermined spatial frequency are set in the spatial frequency domain. Transforming means for converting and calculating each spatial frequency component of both data, and calculating a correction coefficient composed of a ratio of each spatial frequency component of both data so that the gain characteristic is substantially constant within the predetermined spatial frequency range. And a storage unit for storing the correction coefficient, a multiplication unit for multiplying the correction coefficient by each spatial frequency component of the digital image data, and an inverse conversion unit for inversely converting the upper limit result. ing. The converting means and the multiplying means are, for example, executable programs, and are executed by the CPU 2 in FIG. 1, and the memory means is also executed by the memory 3 in FIG. Is obtained. However, some or all of the above components may be embodied by another hardware configuration. Subsequently, the fifth invention will be described.

According to a fifth aspect of the present invention, when the digital image is obtained, the digital image is cut out in a predetermined block unit to create block data, and the block data is converted into a spatial frequency domain to extract each spatial frequency component. Then, each frequency component of the block data is multiplied by a correction coefficient so that the gain characteristic becomes substantially constant within the predetermined spatial frequency limit range, and the multiplication result is inversely transformed and then combined to obtain the digital signal. This is different from the conventional example in that it is configured to perform a correction calculation for each pixel of the image. Further, the block data may be data cut out by overlapping adjacent blocks. Less than,
The procedure will be described in detail. The size of the block for converting the digital image into the spatial frequency domain and correcting the spatial frequency is N × N. Each data of the image before correction is I
It is assumed that the width of overlapping the image blocks of (x, y) is C pixels (C is a multiple of 2) in both the X and Y directions. The N × N block is cut out from the image I and the spatial frequency is corrected. The upper right coordinates of each image block when this block is cut out are defined as follows. Image I (((N−C) × n), ((N−C) × n)):
n = 0,1,2,3, ..., m = 0,1,2,3, ... After correction of the N × N image cut out in this way, as shown in FIG. An image of a block having a size of −C) × (N−C) is adopted as a correction image, and other correction image data is truncated because it causes image deterioration due to a block boundary. To set the cutout coordinates,
This is taken into consideration when setting. If the overlap area is taken into consideration when the image of each block is cut out as described above, the corrected image can be easily recombined by writing the corrected image data of each block in the storage device in order for the X axis and the Y axis. can do. This processing procedure is shown in FIG.

As described above, the digital image is cut out in block units, corrected in the spatial frequency domain, and the correction results are combined, so that the pixels near the block boundary become common to the adjacent blocks. Beyond the boundaries, they become correlated. As a result, the problem of image quality deterioration near the block boundary can be solved and a simpler corrected image with high sharpness can be obtained. Furthermore, by using the block data as data cut out by overlapping adjacent blocks, the correlation between pixels beyond the block boundary becomes more remarkable. Further, the overlap width of the adjacent blocks may be determined so that the difference in image density at the block boundary when the correction calculation is performed on each pixel of the digital image falls within a predetermined allowable error range. The procedure will be described in detail. An object whose shape is known in advance having a predetermined spatial frequency is X-rayed to obtain a digital X-ray transmission image. For example, in FIG.
An image obtained by X-ray photographing a subject having a rectangular cross-section whose shape is known as shown in (1) is shown in (b) compared to the density information (a) due to the transfer characteristics of the X-ray imaging system and the signal processing system. As described above, the point where the edge portion has sloping density information is as described above.

Here again, the density information of (b) for each pixel is U = (..., u _{-n, -n} , ..., u _0,0 , ..., u _{n, n} ,
..), and the concentration information of (a) is Y = (..., y _{-n, -n} , ..., y _0,0 , ..., Y _{n, n} ,
・ ・） Shall be represented. Further, the density information of FIG. 3B corrected for the spatial frequency by the above method is T = (..., t- _{n, -n} , ..., T _0,0 , ..., t _{n, n} ,
..), and the difference between the density information Y and the density information T is D = (..., d _{-n, -n} , ..., d _0,0 , ..., d _{n, n} ,
・ ・） Shall be represented. First, if the maximum difference D _TrMax on the block boundary of the difference D between the required density information Y when C = 2 and the corrected density information T is within a predetermined allowable error range, C = 2 To decide. When it does not fall within the allowable error range, the value of C is increased by a multiple of 2 until the maximum difference D _TrMax falls within the allowable error range. The processing procedure of this method is shown in FIG. If the overlap width of the adjacent blocks is determined in this way, the correlation between pixels beyond the block boundary becomes the most prominent.

A sixth invention is an apparatus to which the method of the fifth invention can be applied. In this device, when the analog output from the radiation image pickup system is processed by the image processing system to obtain a digital image, a block data creation means for cutting out a digital image of a subject in predetermined block units to create block data, Transforming means for transforming the block data and the target data into the spatial frequency domain respectively to obtain the spatial frequency components of both data, and for both data so that the gain characteristic becomes substantially constant within the predetermined spatial frequency component range. Correction means for calculating a correction coefficient composed of the ratio of each spatial frequency component of the above, storage means for storing the correction coefficient, multiplication means for multiplying the correction coefficient by each frequency component of the block data, and the multiplication result. By inverting the result of the inverse transformation with the inverse transformation means for inverse transforming It is provided with a correction means for correcting calculation for. The block data creating means, the converting means, the calculating means, the multiplying means, the inverse converting means, and the correcting means are, for example, executable programs and are executed by the CPU 2 in FIG. 1, and the storage means in FIG. By being executed by the memory 3, the function and effect of the third invention can be obtained.
However, some or all of the above components may be embodied by another hardware configuration. As a result, it is possible to obtain an image processing method and apparatus capable of performing an inspection process according to the actual shape by associating the captured radiation image with the shape of the subject in each case. The following modifications are also possible. For example, even in the case of one-dimensional data such as voice data, there is deterioration of data due to the transfer characteristics as described above, and it may be required to correct this in the spatial frequency domain. In that case, the above correction method can be applied to the frequency correction of one-dimensional data. The one-dimensional case is simpler than the two-dimensional case, and the above method for one coordinate axis may be applied to the input data string.

[0024]

Since the radiation image processing method and the apparatus therefor according to the first aspect of the present invention are configured as described above, gain characteristics within a predetermined spatial frequency range are obtained for subjects having various spatial frequencies. Is corrected to be substantially constant, so that the shape of the subject and the density information can be associated with each other. According to the second aspect of the invention, a digital image of a subject having a predetermined shape having a predetermined spatial frequency is input, the subject shape is output, and the relationship between the two is approximated by an approximate expression. A correction formula for the transfer characteristics including the processing system can be easily obtained. Further, if a correction calculation is performed on each pixel of the digital image using the correction equation of the transfer characteristic such that the gain characteristic is substantially constant within the predetermined spatial frequency range, the correspondence between the shape of the subject and the density information is obtained. Can be taken easily. In the third and fourth aspects of the invention, since the digital image of the object having a predetermined spatial frequency and the shape of which is known in advance and the object shape are converted into the spatial frequency domain to obtain the correction coefficient, the radiation image capturing system A correction coefficient including the image processing system is obtained for each spatial frequency. Further, the digital image data of the object of known shape having the predetermined spatial frequency is converted into the spatial frequency domain and the obtained spatial frequency component is multiplied by the correction coefficient, and the multiplication result is inversely transformed, A correction calculation is performed on each pixel of the digital image. In this way, since the digital image is converted into the spatial frequency domain for correction, the spatial frequency band in which the gain characteristic is substantially constant can be easily set.

Further, above the predetermined spatial frequency, if the correction coefficient is set to zero, image noise is removed.
As a result, an image with less noise and having substantially constant gain characteristics within a predetermined spatial frequency range is restored. Further, when the target data of the object of known shape having the predetermined spatial frequency is given by the MTF chart, this M
Since the spatial frequency and the thickness of the TF chart are manufactured with high accuracy, a highly accurate correction coefficient or correction formula can be obtained. Further, in the fifth and sixth inventions, the digital image is cut out in block units, corrected in the spatial frequency domain, and the correction results are combined, so that pixels near the block boundary become common to the adjacent blocks. Then, it comes to be correlated beyond the block boundary. As a result, the problem of image quality deterioration near the block boundary can be solved and a simpler corrected image with high sharpness can be obtained. Furthermore, by using the block data as data cut out by overlapping adjacent blocks, the correlation between pixels beyond the block boundary becomes more remarkable. Further, if the overlap width of the adjacent blocks is determined so that the difference in image density at the block boundary when the correction calculation is performed on each pixel of the digital image falls within a predetermined error range, the block width exceeds the block boundary. The correlation between the pixels is most remarkable. As a result, it is possible to obtain an image processing method and apparatus that can perform an inspection process according to the actual shape by associating the captured radiation image with the shape of the subject.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a radiation image processing system A ₁ according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a schematic configuration of an image processing system A ₂ of the present invention according to another embodiment of the present invention.

FIG. 3 is a comparison diagram of image density information obtained by the radiation image processing system A ₁ and required image density information.

FIG. 4 is a flowchart showing a processing procedure for obtaining a correction formula.

FIG. 5 is a flowchart showing a processing procedure of correction calculation.

FIG. 6 is a flowchart showing a processing procedure for deriving a correction coefficient.

FIG. 7 is a flowchart showing each processing procedure of correction coefficient derivation and correction calculation.

FIG. 8 is an explanatory diagram of clipping a digital image in block units.

FIG. 9 is a flowchart showing a processing procedure of image correction.

FIG. 10 is a flowchart showing a processing procedure for obtaining an overlap width.

FIG. 11 is a diagram showing an example of response characteristics of an image intensifier.

[Explanation of symbols]

1 ... Memory for storing radiation image 2 ... CPU (corresponding to block data creating means, converting means, calculating means, multiplying means, inverse converting means, and correcting means) 3 ... Memory for storing correction equation or correction coefficient (storage means Equivalent to)

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06T 5/00 H04N 1/40 H04N 1/40 Z (72) Inventor Eiji Yoshida Arai Town, Takasago City, Hyogo Prefecture 2-3-1 Nihama Kobe Steel Works, Takasago Works, Ltd. (72) Inventor Shigeru Yuki 2-33-1 Niihama, Arai-cho, Takasago-shi, Hyogo Prefecture Kobe Steel Works Takasago Works (72), Inventor Shiro Koike Hyogo 2-3-1, Niihama, Arai-cho, Takasago-shi, Kochi Kobe Steel Co., Ltd. Takasago Works

Claims (14)

[Claims] 1. A radiation image processing method for obtaining a digital image by processing an analog output from a radiation image pickup system by an image processing system, wherein the image pickup is performed so that a gain characteristic is substantially constant within a predetermined spatial frequency range. A radiation image processing method, characterized in that the transfer characteristics of the image processing system and the image processing system are corrected. 2. A radiographic image processing method for obtaining a digital image by processing an analog output from a radiographic image capturing system by an image processing system, the gain of frequency characteristics of a digital image of a subject having a predetermined spatial frequency and having a known shape. And a target gain to obtain a gain characteristic, and the transfer characteristics of the image capturing system and the image processing system are corrected so that the gain characteristic becomes substantially constant within a predetermined spatial frequency range. Radiation image processing method. 3. The radiation according to claim 1, wherein a correction calculation is performed on each pixel of the digital image using a correction equation of the transfer characteristic such that a gain characteristic is substantially constant within the predetermined spatial frequency range. Image processing method. 4. A radiographic image processing method for obtaining a digital image by processing an analog output from a radiographic image pickup system by an image processing system, wherein digital image data and a target data of an object of a known shape having a predetermined spatial frequency are obtained. Convert to the spatial frequency domain to obtain each spatial frequency component for both data, and calculate a correction coefficient consisting of the ratio of each spatial frequency component for both data so that the gain characteristic becomes approximately constant within the above specified spatial frequency range. Then, the radiation image processing method is characterized by performing a correction calculation for each pixel of the digital image using the correction coefficient. 5. A spatial frequency component obtained by converting digital image data of a known object having a predetermined spatial frequency into a spatial frequency domain is multiplied by the correction coefficient, and the multiplication result is inversely transformed. The radiographic image processing method according to claim 4, wherein a correction calculation is performed on each pixel of the digital image according to. 6. The radiation image processing method according to claim 4, wherein the correction coefficient is set to zero at the predetermined spatial frequency or higher. 7. The radiation image processing method according to claim 4, wherein the target data of the object having a known shape having the predetermined spatial frequency is given by an MTF chart. 8. A radiation image processing apparatus for obtaining a digital image by processing an analog output from a radiation image pickup system by an image processing system, wherein digital image data of a subject having a known shape having a predetermined spatial frequency and target data are obtained. It is composed of a converting means for converting into the spatial frequency domain to obtain respective spatial frequency components of both data, and a ratio of the respective spatial frequency components of both data so that the gain characteristic becomes substantially constant within the predetermined spatial frequency range. Computation means for computing a correction coefficient, storage means for storing the correction coefficient, multiplication means for multiplying the correction coefficient by each spatial frequency component of the digital image data, and inverse conversion means for inversely converting the multiplication result. A radiation image processing apparatus comprising: 9. A radiographic image processing method for obtaining a digital image by processing an analog output from a radiographic image pickup system by an image processing system to create block data by cutting out the digital image in predetermined block units, The data is converted into the spatial frequency domain to obtain each spatial frequency component, and each spatial frequency component of the block data is multiplied by a correction coefficient so that the gain characteristic becomes substantially constant within the predetermined spatial frequency component range, A radiation image processing method, characterized in that a correction calculation is performed on each pixel of the digital image by inverting and combining the multiplication results. 10. The block data is data cut out by overlapping adjacent blocks.
The described radiation image processing method. 11. The overlap width of the adjacent block is determined so that the difference in image density at the block boundary when the correction calculation is performed on each pixel of the digital image falls within a predetermined error range. Claim 10
The described radiation image processing method. 12. A radiographic image processing apparatus for obtaining a digital image by processing an analog output from a radiographic image pickup means by an image processing system. Block data for cutting out a digital image of a subject in predetermined block units to create block data. Creating means, transforming means for transforming the block data and the target data into a spatial frequency domain to obtain respective spatial frequency components of both data, and a gain characteristic being substantially constant within the predetermined spatial frequency component range. Further, a calculating means for calculating a correction coefficient composed of a ratio of each spatial frequency component for both data, a storage means for storing the correction coefficient, and a multiplying means for multiplying the correction coefficient by each spatial frequency component for the block data. And an inverse transform means for inverse transforming the multiplication result, and the above-mentioned inverse transform result by combining them. A radiation image processing apparatus comprising: a correction unit that performs a correction calculation for each pixel of a digital image. 13. The radiation image processing apparatus according to claim 12, wherein the block data created by the block data creating means is data cut out by overlapping adjacent blocks. 14. The overlap width of the adjacent blocks is set so that the difference between the image densities at the block boundaries when the block data creating means performs the correction calculation for each pixel of the digital image falls within a predetermined error range. 14. The radiation image processing apparatus according to claim 13, further comprising an overlap width determining means for determining.