JPH01287782A - Method and device for image processing - Google Patents

Abstract

PURPOSE:To improve the picture quality of an image obtained by increasing the sharpness of a medical digital image by emphasizing or suppressing a local variation component of image information on an original image by a specific arithmetic expression. CONSTITUTION:A measured value distribution B which is obtained by smoothing a measured value distribution O of various physical quantities of the original image and the local variation component V of the measured value distribution O are found and arithmetic shown by P=B+f(B;V) where the measured value distribution at the time of image reproduction is denoted as P is carried out to emphasize or suppress the local variation component of the measured value. Here, f(B;V) is a function of V including B as a parameter and a monotonous increase function in a broad sense as to such V that when B is constant, f(B; V1)<=f(B;V2) at all times for V1<V2. Further, when f(B;0)=0 and V is a constant positive value, f(B1;V)/¦f(B1;-V)¦>=f(B2;V)/¦f(B2;-V)¦ for B1<B2. Consequently, high picture quality is obtained in sharpness increasing processing.

Description

Detailed Description of the Invention [Industrial Application Field] The present invention reads image information recorded in a 1M image, converts it into an electrical signal, and reproduces the electrical signal as an image by locally converting the image information into one image. The present invention relates to an image processing method and apparatus for emphasizing or suppressing variable components, and particularly relates to an image processing method and apparatus suitable for sharpening processing of medical digital images such as digitized X-ray images.

[Conventional technology]

Conventional image processing methods that sharpen images by emphasizing high spatial frequency components included in the image include Rosenfeld,
Co-authored by Kirk, translated by Makoto Nagao et al. [Digital Image Processing] (Kinda Kagakusha, published in 1978), pages 185 to 196
As discussed in 1999, the Laplacian filter method in real space and the unsharp mask method in single-frequency space have been used. In the edges of images processed with these filters, where the density and brightness change,
Overshoots and undershoots occur in changes in density and brightness, and these are factors that increase the contrast in the edge area and sharpen the image. However, when these filters are used to process images that inherently have high contrast, overshoot and undershoot become excessive, resulting in a phenomenon in which one contour becomes double (hereinafter referred to as "false contour"). Ta. Also, due to these overshoots and undershoots.

The variation range of the density and brightness of the processed image greatly deviates from the variation range of the original image, and since the variation range of the density and brightness of the silver halide film or CRT that displays the processed image is finite, there is a large overshoot. A phenomenon (hereinafter referred to as "blurring") occurs in which image information on top of undershoots or undershoots is lost because it exceeds the variation range of the density or brightness of the display medium.

The phenomenon of such false contours and bulges will be explained with reference to FIGS. 8 and 9. Here, for the sake of simplicity, a -dimensional image will be explained as an example.

First, the solid broken line in FIG. 8 represents the density change A of the original-dimensional image taken as an example. ,
The density changes as shown by the solid line curve in FIG. In this case, as shown in the figure, false contours E and F occur due to large overshoots and undershoots, respectively.

In addition, in FIG. 9, the portions (hatched portions) exceeding the upper limit G and lower limit H of the concentration of the display medium indicated by the dashed line are as follows:
Image information is lost due to whelks.

Although the occurrence of such false contours E and F and burrs differs in appearance and degree, it cannot be avoided even with sharpening processing using other methods such as the Laplacian filter method or the high-frequency emphasis filter method. Can not.

On the other hand, as an image processing method to alleviate the above-mentioned drawbacks, a method for suppressing the above-mentioned excessive emphasis has been proposed as published in Japanese Patent Application Laid-Open No. 56-91735. . That is, for the value of a simple emphasis component given as the difference between an image that has undergone simple sharpening processing as shown by the solid line curve in Figure 9 and the original image shown by the solid broken line in Figure 8, Then, the value of the suppressed emphasis component is calculated as the 10th
As shown in relationship J in the figure, the relationships are made so that excessive emphasis is suppressed. Then, from the above simple emphasis component, the first
Using the relationship shown in Figure 0, the suppressed emphasized component is found and added to the original image. When the original-dimensional image shown in FIG. 8 is processed in this manner, the density changes shown by the dashed straight line and the dashed curve in FIG. 9 are obtained. In this case, as shown in the figure, the false contours E and F in the simple sharpening process are relaxed.

[Problem to be solved by the invention]

However, in the image processing method described in JP-A-56-91735, although the occurrence of false contours can be suppressed, as shown in FIG. (see section) were still in existence. Namely.

The problem of tsubusi remained essentially unsolved. Rather, as a secondary problem, the emphasis is suppressed even for signals on intermediate density or signals on intermediate brightness that generally contain a lot of image information and are unrelated to the blurring. Therefore, it has not been possible to achieve sufficiently high image quality in, for example, sharpening processing of medical digital images.

Therefore, 5 the present invention solves these problems, suppresses the occurrence of false contours in images, removes blur, and does not suppress the emphasis of signals on intermediate density or intermediate brightness that are unrelated to these problems. The object of the present invention is to provide an image processing method and apparatus.

[Means to solve the problem]

The above purpose is to scan an original image in which measured values of various physical quantities are recorded as a two-dimensional distribution, read the image information recorded in this original image, convert it into an electrical signal, and reproduce this electrical signal as an image. At this time, from the measurement value distribution 0 of various physical quantities of the original image, the measurement value distribution 0 itself or a measurement value distribution B obtained by smoothing the measurement value distribution O, and local fluctuation components of the measurement value distribution 0. Calculate V and take the measured value distribution at the time of one image playback as P. (% formula %) (However, f(B; V) is a function of V with B as a parameter, and if B is constant, At some point, ■.

< V
(B; Vz) is a monotonically increasing function of V in a wide sense. Furthermore, when f(B;0)=O and ■ is a constant positive value, f (Bi;V) /I f (B□;- V) 1≧f
It has the property that (Bl :v) /I r (B, ;-v) 1. ) is achieved by an image processing method that emphasizes or suppresses local fluctuation components of the measured values.

In addition, the image processing device used to implement the above image processing method scans an original image in which measured values of various physical quantities are recorded as a two-dimensional distribution, reads image information recorded in this original image, and generates electrical signals. an image reader that converts into
In an image processing apparatus comprising an image reproducing device that reproduces an output signal from the signal processing device as an image, the signal processing device calculates, from the measurement value distribution O of various physical quantities of the original image,
The measurement value distribution B obtained by smoothing the measurement value distribution O itself or the measurement value distribution 0, and the local fluctuation component V of the measurement value distribution O are calculated, and the measurement value distribution at the time of one image reproduction is determined as P.
Then, the calculation formula % formula %) (However, f (B; V) is a function of V with B as a parameter, and when B is constant, for the set of V such that V □ × vt It is a monotonically increasing function with respect to V that always satisfies f(B;V)≦f(B;V,).Also, f(B
oo)=O, and when ■ is a constant positive value, B,
For the set B <B2, always f(B□;V) /I f(B,;-V) 1≧f (
Bi;v) /l f(B, -V) 1. ) There is a device configured to perform the calculation expressed as .

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The image processing method according to the present invention includes optical density and brightness.

The original image is a broadly defined image in which measured values of various physical quantities such as X-ray absorption coefficient, hydrogen atom concentration, radioisotope concentration, ultrasonic reflection amount, temperature, etc. are recorded as a two-dimensional distribution, and this original image is scanned and recorded. This is to emphasize or suppress local fluctuation components of the measured values when reading the image information of the image and converting it into an electrical signal, and reproducing the electrical signal as an image.

In order to perform such image processing, the present invention utilizes the concept of ground and figure in visual perception. That is, letter 5 mentioned above.
As stated on page 9, if the visual field, which is the target image of visual perception, is composed of two areas of different brightness, visual perception treats one of the two areas as meaningless.
The other is often recognized as a meaningful diagram. In this case, in other words, the area judged to be a diagram is treated with more emphasis in visual perception. The idea is that when increasing or decreasing the local variation component of an image, find the density or brightness of the local ground, and increase the variation toward black, which is considered as Figure 5, in areas where the background is white, and also increase the variation toward white. The goal is to reduce the fluctuations in the area where the ground is black, and to increase the fluctuations in the white direction, which is considered to be one figure, and to reduce the fluctuations in the black direction.

More specifically, the position coordinate on the two-dimensional image is X.

y, the density distribution of the ground obtained by smoothing the density distribution 0 (X, y) from the measured value distribution of various physical quantities of the original image, for example, the density distribution 0 (X + y) of optical density, is the ground density distribution B OC + y) and the local variation component V (xt y) of the above density distribution 0 (xty), and calculate the density distribution P (xt y) of the image during image reproduction, that is, after image processing.
and are related as shown in the following equation (1).

p (x, y) = B (xt y) + f(B (
xt y) ;v (xt y) )...(1) Here, f(B;V) is the ground concentration distribution B (xt y)
It is a function of the fluctuation component V (xt y) with V (xt y) as a parameter. Note that the above ground density distribution B(x, y) may be the density distribution ○(xty) of the original image itself. Also, f
(B;V) is when B is constant, vl<vx
...(2) For the set of values, f(B; vl)≦f(B; v,)
It is a monotonically increasing function in a broad sense that satisfies the relationship: '-(3). Since ■ is a fluctuation component, it takes positive and negative values, but
Regardless of the value of B, f(B;0)=O...(4) and when ■ is a constant positive value.

Bx<B,
...(5) For the set of values, f (B,;v) /If(B1;-V) l≧f(B
, ;v)/If(Bs;-v)l ・(6) Equation (6) is as follows: When the concentration of the ground increases from B1 to B2, the change f(B;V) that increases the concentration is This means that when compared with the change f(B knee V) that lowers the density, it tends to be relatively suppressed. As a result, in areas where the ground density B is high (black), the change f(B;V
) is relatively suppressed, and the whitening change f(B;-V) is relatively emphasized. Conversely, in places where the ground density B is low (white), the blackening change f(B;V) is relatively emphasized,
The whiter change f(B; -V) is relatively suppressed. This will suppress both the changes that make white parts whiter and the changes that make black parts blacker in edge areas that have a large contrast between white and black, suppressing excessive overshoot and undershoot. False contours are alleviated. In addition, the blurring phenomenon can be alleviated.

Furthermore, in the present invention, in order to completely eliminate the above-mentioned blurring phenomenon, the following property is given to the function f(B;V). That is, f
(B; V) is convex upward with respect to V in the positive region of V, and convex downward with respect to V in the negative region of . In other words, f(B
; The slope of V) with respect to V becomes a monotonically decreasing function in a wide sense in the positive region of ■, and a monotonically increasing function in a wide sense in the negative region of ■. Then, when B is kept constant, f (B; Vmax)≦Omax-B for the maximum value vmax that ■ can take.
− (7) is satisfied, and the minimum possible value of ■ V w
f(B;vlLin)≧0i−B for in
・=(8) should be satisfied. Here, OmaX in equation (7) is the maximum density that the original image can take, and ○win in equation (8) is the minimum density that the original image can take. In this way, steps (1), (7), (8)
From the equation and monotonically increasing property of f(B;V), 0m1rISP(x,y)≦○wax
-(9), and the density distribution P (X? y) of the processed image when reproduced as an image does not exceed the density range that the original image can take (Owin, Oma error, so the blurring phenomenon can be completely removed. .

As a function f(B:V) that satisfies the above properties.

For example, the following can be considered: That is, f (B
; V) = (2y/π) arctan ((s
/2)βV/-1]・(10), and if γ is positive or O, γ=Omax-B
-01), and when ■ is negative, γ=B-0-...(12) In equation (10), β is the parameter that determines the degree of emphasis for small signals. and a-rc
tan is an arctangent function in radians, and is the pi ratio.

Next, the operation of the image processing method configured as described above will be explained with reference to FIGS. 1 to 4.

Here, the 81st image is used as the original image! Let us take as an example a one-dimensional image shown by a solid broken line in l. First, the original image □ with the density change A shown in Figure 8 is smoothed with a non-sharp mask of mask size C to obtain the ground density change M shown in Figure 1. In Figure 1, the density upper limit G and lower limit H of the display medium are also written in dashed lines. 0 The density variation range of the display medium is 1 Normal original image processing □
Therefore, the value of the upper limit G is set as Omax in equation (11), and the value of the lower limit H is set as O+ain in equation (12). 88th original image 11
By subtracting the smoothing image factor 2 in FIG. 1 from , the local variation component N shown by the broken line in FIG. 2 is obtained. From the density change M of the ground image in Figure 1 (corresponding to B in equations (11) and (12)) and the fluctuation component N shown by the broken line in Figure 2 (corresponding to V in equation (10)) , Otmax as defined above.

When f(B;V) is obtained from equations (10), (11), and (12) using Owin, the processed local fluctuation component Q is shown as the solid line in FIG. In addition, in this calculation, the (10th
) The parameter β in the equation was set to 2°0. Furthermore, the V dependence of f(B;V) when B is fixed is shown in FIG. 1 for the level B process.

Level B2. When looking for level B, we get f in Figure 3.
(B, ;V) , f (B, ;V) , f (B3
;V), from Figure 3, (10) and (11
), the function defined by equation (12) is f(B
; It can be seen that the condition V) is satisfied.

Next, according to equation (1), the density change M of the ground shown in FIG.
When the local fluctuation component Q after processing shown by the solid line in FIG. 2 is added to , the density change R of the image after processing according to the present invention shown by the solid line in FIG. 4 is obtained. The broken line in FIG. 4 shows the density change of the image after processing by the conventional simple non-sharp mask method shown in FIG. Moreover, the - dotted chain line is the upper limit G and lower limit H of the density of the display medium. As can be seen from FIG. 4, in the density change R shown by the solid line, excessive overshoot and undershoot caused by the density change shown by the broken line are effectively suppressed.

In addition, since the concentration upper limit G and concentration lower limit H shown by the - dotted chain line are not exceeded, the blurring phenomenon is completely eliminated. Moreover, the emphasis on signals on intermediate densities is not substantially suppressed.

Next, an image processing apparatus used to implement the above image processing method will be described with reference to FIG.

In the figure, an image reader 1 scans an original image in which measured values of various physical quantities are recorded as a two-dimensional distribution, reads image information recorded in this original image, and converts it into an electrical signal. It consists of a film digitizer that reads the density of a transmission film image by scanning it and converts it into a digital signal. The signal processor 2 inputs the digital signal output from the image reader 1 and performs necessary processing. For example, an image regenerator consisting of a spatial filter receives the output signal from the signal processor 2. It is played back as an image, and includes an image display device 3 that converts the digital signal after signal processing into an analog video signal and displays it on a monitor screen, and a laser beam whose brightness is modulated according to the processed signal and scans it on film or photographic paper. It consists of a laser printer 4, etc., which creates a hard copy of the image.

Using such an image processing device, optical density, brightness,
In order to perform image processing on a broadly defined image in which measured values of various physical quantities such as linear absorption coefficient, hydrogen atom concentration, radioisotope concentration, ultrasonic reflection amount, temperature, etc. are recorded as a two-dimensional distribution as an original image, first, image reading is performed. The machine scans the original image, reads the image information recorded in the original image, and converts it into a digital signal. Next, the digital signal output from the image reader 1 is input to the signal processor 2 and stored in a first image storage device 15a provided therein. In this image storage device 5a, image information is stored as a digital value 0(1+j) corresponding to the density value of each pixel (XI J). Here, Cxej) is a pixel number represented by 2 in a matrix format, and represents, for example, the i-th pixel to the right and the j-th pixel below from the upper left corner of the original image.

Next, the output signal 0(i) of the image storage device 5a.

j) is input to the smoothing calculator 6, smoothed using a rectangular mask, and the output signal 5 (II j) is stored in the second image storage device 5b. That is, the smoothing calculator 6 performs the operation expressed as j0g-n/2) (13). However, in the above formula, n is the mask size, and the decimal part of the division n/2 is rounded down. It will be done. The mask size n is selected from the n value group 7 by the first switch 8a. Note that the third image storage device 5C
Output signals S(l, j) with different mask sizes n are stored in , but this can be omitted depending on the case.

After making the above preparations, actual image processing is performed. That is, first, the original image density 0 (x, j) read from the first image storage device 5a and the smoothed image density 5 (II j) read from the second image storage device 5b are Input to the first subtractor 9a and subtract V (11j) =O(xy J) -8Cx+ J)
- Perform (14). The output V of this subtractor 9a
(II j) corresponds to the local variation component.

At the same time, the second switch 8b selects a signal B( x, j). This signal B (IIJ)
are respectively input to the second subtractor 9b and the third subtractor 9c, and two types of subtraction Omax-B (i, j)
-(15)B (L j) -〇min
...(16) is performed, where Omax is the upper limit of the variation range of image density, O+m
in is the lower limit.

Signal V(i,j) output from the first subtractor 9a
is compared with "0" by the comparator 10.

Then, the output of this comparator 10 is input to the selector 11,
Either the output of the second or third subtractor 9b, 9c is selected and output. That is, the comparator 10 and the selector 11 determine that when V(it j) is positive or 0, the output of the selector 11 becomes the result of equation (15), and when it is negative, it becomes the result of equation (16). I am trying to get the result of the formula.

Here, the output of the selector 11 is the above-mentioned (11).

This is the same as γ defined by equation (12). This γ gives the upper and lower limits of the fluctuation component after processing, but as shown in Figure 3, even if γ is increased to about 1.2 times, the required fluctuation range cannot be reached in practice. It will never be surpassed. Therefore, in this embodiment, γ in equations (11) and (12) is
Instead, if the signal V(i, j) is positive or 0, γ(L j) = a (O1!1aX-B (it j
) ) −(17), and if negative, γ(1?J)=α(B (11j)−α−) ・
...(18) A signal γ(it J) is used. Then, α is used as a parameter and can be selected depending on the image. That is, the third switch 8c selects the parameter α from the α value group 12, and the first multiplier 13a multiplies this α by the output of the selector 11 to obtain the signal γCxp j). .

On the other hand, the signal β(11j) corresponding to β, which is a parameter related to the degree of emphasis in equation (10), is switched between the output of the β table group 14 and the fixed β by the fourth switch 8d.
Selected from value group 15.

The β table group 14 is used to change the degree of emphasis according to density. That is, when useful image information in one image is localized in a specific density region, the value of β is increased in that region to increase the degree of emphasis. The β-table group 14 includes a plurality of β-tables that are look-up tables made up of IC memories. 61st! ! !
l is an example of the input/output relationship of the β table. That is, I.C.
Depending on the input concentration corresponding to the memory address input, I
The output β value corresponding to the data output of the C memory is correlated. In the example shown in FIG. 6, when the input density is included in the β increasing density region S, an output β value is obtained that is increased from the reference β value T by an amount determined by the β increasing amount U. The β table group 14 includes a β increasing density region S and a reference β value T.

It includes various β tables with different values of β increment U, and can be used depending on the image. Also,
The input density of the β table is selected by the fifth switch 8e from the first to third image storage devices 5a.

It is possible to select from the original image density read out from 5b and 5c and two types of smoothed image density.

The reciprocal table 16 is a look-up table constructed from an IC memory, similar to the β table described above, and is adapted to obtain a signal corresponding to the reciprocal of the input value. The signal β(1
1J), the signal V(1,j) corresponding to the local fluctuation component, and the output of the reciprocal table 16 are multiplied together by the second multiplier 13b and the third multiplier 13c, and as a result β
(L j) V (L j) /γby J)
. . . becomes a signal corresponding to (19), and outputs it to the arctangent table 17.

The arctangent table 17 is a look-up table constructed from an IC memory, similar to the β table described above.
When the input is a and °.

b= (2/z) arctan (?Ca/2)
The output specified by ``'' (20) can be obtained.

Here, arctan is an arctangent function in radians, and π is pi.

The output of the arctangent table 17 is sent to the fourth multiplier 13d.
The signal γ(i,j) is multiplied by the signal γ(i,j), and the result is added to the signal B(IIJ) by the adder 18. and,
The output of this adder 18 becomes the output of the signal processor 2 of the present invention.

As is clear from the above explanation, the output of the adder 18 is: B (L j) + (2/π) γ(i, j) a
rctan ((7C/2)×β(11J)V (iy
j)/γ(xtj)) (21), which corresponds to substituting the right side of equation (1o) into the right side of equation (1). Therefore, the signal processor 2 of this embodiment has the characteristics and effects as previously explained using FIGS. 1 to 4.

Furthermore, in the signal processing device 2 of this embodiment, the signal B(x, j) corresponding to the ground density is converted into the output of the first to third image storage devices 15a, 5b, and 5c as described above. You can choose from among them. Now, as B(xtj), O(
xt j), even if the emphasis of the fluctuation component is suppressed, except for the edge area with large contrast,
B is not suppressed below the variation of the original image.
(i.

j) is the output of the second image storage device 5b, and is input to the first subtractor 9a to obtain the fluctuation component V(i.j).

By using the density of the smoothed image used to calculate j), it is possible to emphasize or suppress the fluctuation component more than the original image according to the value of the signal β(i, j).

In particular, by using the β table, setting the standard β value T in Figure 6 to 1 or less, and making the sum of the standard β value T and β increment U 1 or more, a specific density area can be sharpened and other density areas can be smoothed. It becomes possible to convert into Furthermore, B(i, j) is the output of the third image storage device 5G, and the fluctuation component V(II j
) can be used to construct a signal processor with special frequency characteristics, such as a bandpass filter.

Note that since the signal processor 2 of this embodiment is equipped with various β tables, it is possible to select sharpening and smoothing suitable for one image.

Moreover, the signal processor 2 of this embodiment can also be realized by a general-purpose electronic computer. However, with electronic computers, multiplication, division, arctangent function calculations, etc. take time. On the other hand,
In this embodiment, multipliers 13a to 13d, reciprocal table 16
, the arctangent table 17 is used, so high-speed processing is possible.

FIG. 7 is a block diagram of main parts showing a second embodiment of the signal processor 2 according to the present invention. The difference between the signal processor 2 according to this embodiment and the signal processor 2 shown in FIG. 5 is that in FIG. As shown in FIG. A multiplier 9d, a multiplier 13e,
It consists of an adder 18b, a switch 8c, and a multiplier 13a.

First, the signal V (
i, j) are input into the step table 20. This step table 20 is a look-up table constructed from an IC memory like the β table described above, and when the input signal value is positive or O, II I T# is set, and when it is negative, "
-1'' is output.

On the other hand, the signal B (xt j) selected by the second switch 8b shown in FIG.
+ax+0m1n)/2. Here, Omax is the upper limit of the variation range of image density, and Onin is the lower limit. The output of the step table 20 and the subtracter 9
The outputs of d are multiplied together in a 1 multiplier 13s, and the output of this multiplier 13e is added to the value (Omax
−0m1n)/2.

Then, as in the embodiment shown in FIG. 5, the switch 8c
Parameter α selected from 12 values.

The signal γ(ly j) is obtained by multiplying the output of the adder 18b by the 1 multiplier 13a.

As described above, in this example, γ(i, j) is V(i, j
) and B (i, j), γ(i, j)=a((○111ax-Oain)/2
+ CB (L j) + (omX+○m1n) /2]
s t (V (L j) ) ) (22). Here, the function 5t(a) is a step function that has a value of "1" when a is positive or O, and has a value of 111+1 when a is negative. γ(i, j) obtained in this way is the (17th).

γ(i
, j) is clearly the same. Furthermore, in this embodiment, the comparator 10 and selector 11 shown in FIG. 5 are unnecessary, and the types of parts can be reduced.

The embodiments of the image processing method and device of the present invention have been described above, but the present invention is not limited to the above embodiments and can be modified in various ways. Although the signal V(lpj) corresponding to the fluctuation component is obtained using the non-sharp mask method, it may also be obtained using other methods such as the Laplacian filter method or the high-frequency emphasis filter method in the frequency space. Further, in the above embodiment, the smoothed image is obtained by calculation according to equation (13), but it is also possible to obtain it by digitizing an image obtained by optically blurring the original image. Furthermore, although the above embodiments are constructed from digital memories and digital circuits, they may also be constructed from analog circuits such as analog frame memories and operational amplifiers. In addition, the function (2/ x ) using the arctangent function used in the above example
arctan [(π/2)a) can be replaced with a function g(a) having the following properties or a function approximated by a broken line. That is, g (0) = O - (23) nm g (a) = 1
...(24)1jJg (a) =-1...(2
5) d g/d a≧O−(26) d g/da1. ,. =l ・
(77)a·(d"g/da")≦O·(28). A function other than the arctangent function that satisfies these properties is 1, for example, g(a) = (2/(1+exp(-2a)))+1, which is similar to a titration curve.
・ (29) etc. In the above formula, exp
represents an exponential function. In the above example, the image information is given as optical density, but this can be expressed as brightness, X-ray absorption coefficient, hydrogen atom concentration, radioisotope concentration, ultrasonic reflection amount, temperature, etc. It is also possible to replace it with

〔Effect of the invention〕

Since the present invention is configured as described above, by emphasizing or suppressing local fluctuation components of the image information of the original image using a predetermined arithmetic expression, the occurrence of false contours in the image is suppressed and whelks are removed. In addition, image processing can be performed without suppressing the emphasis of signals on intermediate density or intermediate luminance that are unrelated to these. Therefore, for example, in sharpening processing of medical digital images, it is possible to achieve high image quality of processed images and obtain good diagnostic information.

[Brief explanation of the drawing]

Figure 1 is an explanatory diagram showing the state in which the original image consisting of a one-dimensional image has been smoothed, and Figure 2 is an explanatory diagram showing the state of change between the local fluctuation component of the original image and the fluctuation component after processing it. figure,
FIG. 3 is a graph showing functions used in image processing according to the present invention, FIG. 4 is an explanatory diagram showing the effects of image processing according to the present invention, and FIG. 5 is a block diagram showing an embodiment of an image processing apparatus according to the present invention. , Fig. 6 is an explanatory diagram showing an example of the input/output relationship of the β table, Fig. 7 is a block diagram of the main part showing the second embodiment of the signal processor, and Fig. 8 is an illustration of the original image consisting of a one-dimensional image. FIG. 9 is an explanatory diagram showing density changes; FIG. 9 is an explanatory diagram showing an example of processing the original image of FIG. 8 using a conventional image processing method; FIG.
The figure is a graph for explaining suppression of emphasis by a conventional image processing method. 1... Image reader, 2... Signal processor, 3...
・Image display device, 4...Laser printer, I...Original image, Step 2...Smoothed image.

Claims (5)

[Claims] (1) When scanning an original image in which measured values of various physical quantities are recorded as a two-dimensional distribution, reading the image information recorded in this original image and converting it into an electrical signal, and reproducing this electrical signal as an image, From the measurement value distribution O of various physical quantities of the original image, the measurement value distribution B itself or a measurement value distribution B obtained by smoothing the measurement value distribution O, and the local fluctuation component V of the measurement value distribution O. , and when the measurement value distribution at the time of image reproduction is P, the calculation formula P=B+f(B;V
) (However, f(B;V) is a function of V with B as a parameter, and when B is constant, V_1<V_2.
Always f(B;V_1)≦f(B;V_2) for the set
It is a monotonically increasing function with respect to V in a wide sense. Also, f
When (B; O)=0 and V is a constant positive value,
For the set B of B_1<B_2, f(B_1;V)/|f(B_1;-V)|≧f(B_
2;V)/|f(B_2;-V)|. ) is performed to emphasize or suppress local variation components of the measured values. (2) In the region where the slope with respect to the fluctuation component V is positive, the function f(B;V) becomes a monotonically decreasing function in a broad sense with respect to V,
2. The image processing method according to claim 1, wherein V is a function having the property of being a monotonically increasing function in a wide sense regarding V in a negative region, or a polygonal line approximation thereof. (3) When the function f(B;V) holds the measurement value distribution B constant, f(B;Vmax)≦Omax−B for the maximum value Vmax that V can take and the maximum value Omax that O can take. 2. The image processing method according to claim 1, wherein f(B;Vmin)≧Omin-B is satisfied for the minimum value Vmin that V can take and the minimum value Omin that O can take. (4) The function f(B;V) is g(0)=0 ▲There are mathematical formulas, chemical formulas, tables, etc.▼ ▲There are mathematical formulas, chemical formulas, tables, etc.▼ dg/da≧0 ▲Mathematical formulas, chemical formulas, tables, etc. There is ▼ a・(d＾2g/da＾2)≦0 When g(a) and V are positive or 0, γ=α( Omax-B), and when V is negative, the minimum possible value of O
min and the same constant α as in the above equation, γ = α (B - Omin) It is a function of the parameter γ and the measurement value distribution O of the original image, or a function or constant of the measurement value distribution B obtained by smoothing O. The image processing method according to claim 1, wherein the parameter β is defined by the following arithmetic expression f(B;V)=γg(βV/γ). (5) An image reader that scans an original image in which measured values of various physical quantities are recorded as a two-dimensional distribution, reads the image information recorded in this original image, and converts it into an electrical signal, and an electric signal from this image reader. In an image processing device comprising a signal processor that processes a signal and an image reproducer that reproduces an output signal from the signal processor as an image, the signal processor is configured to generate a measured value distribution O of various physical quantities of the original image. Then, the measurement value distribution B itself or the measurement value distribution B obtained by smoothing the measurement value distribution O, and the local variation component V of the measurement value distribution O are determined, and the measurement value distribution at the time of image reproduction is calculated. When P, the calculation formula P=B+f(B;V) (However, f(B;V) is a function of V with B as a parameter, and when B is constant, V_1<V_2.
Always f(B;V_1)≦f(B;V_2) for the set
It is a monotonically increasing function with respect to V in a wide sense. Also, f
(B;0)=0, and when V is a constant positive value,
For the set B of B_1<B_2, f(B_1;V)/|f(B_1;-V)|≧f(B_
2;V)/|f(B_2;-V)|. ) An image processing device characterized in that it is configured to perform an operation represented by: