JP2006061597A - Image processing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To realize signal-to-noise ratio (SN ratio) needed in X-ray film image reader over a wide range of density in an X-ray film image reader, controlling image quality deterioration caused by MTF deterioration. <P>SOLUTION: The apparatus has a spatial filter that can output different characteristics or a plurality of spatial filters having a different characteristics, a means to calculate an average value of density in the vicinity of the elementary pixels, and a means to select the output of the spatial filter based on the calculated average value. The apparatus can make an output based on signal-to-noise ratio needed in the X-ray film image reader by switching the filter with different spatial filter characteristics. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image processing apparatus suitable for use in processing a digitized image signal obtained by optically reading a medical X-ray film image.

  In recent years, a large number of image reading apparatuses have been developed. In the medical field, there are apparatuses for detecting and digitizing medical images, particularly X-ray film images, in order to perform electronic filing, remote diagnosis, computer aided diagnosis, and the like. Has been developed.

Such an image reading apparatus irradiates light to a X-ray film from a light source such as a halogen lamp or a fluorescent lamp, and receives transmitted light from the X-ray film by a solid-state imaging device such as a CCD linear sensor, for example. Image data is obtained by scanning the film.
In medical X-ray film readers, it is necessary to obtain information with high gradation that is faithful to the film, and gradation identification in the apparatus is required to match visual characteristics. The discrimination for visual gradation is not linear with the amount of light transmitted from the film. Density D = −log ₁₀ (T) T: Transmittance (1)
It is known that it is possible to discriminate even with a faint difference when the amount of light is small, that is, close to black. Therefore, a medical X-ray film reader usually outputs a density value.

  A CCD line sensor, which is an image sensor, outputs an electrical signal having a magnitude proportional to light, but includes noise components such as reset noise and shot noise in the output. Also, noise components due to amplifiers and A / D are included in the AD output. Except for shot noise, these noises do not depend on the amount of light and are almost constant.

The density D is expressed by the formula (1) as described above. When this equation is fully differentiated,
dD = − (10 ^D log _e ) dT (2)
Where noise is the variance σ ²
If the normal distribution conforms to the above, the dispersion value of the density D is (10 ^D log _e ) ² σ ² .

That is, assuming that the noise is the standard deviation σ, the noise converted into the density value is proportional to the exponential function (10 ^D log _e ).

  Since this noise can be regarded as an independent random number, conventionally, noise has been suppressed by performing spatial filtering using a moving average or the like.

For example, if the set of pixels is {H _i | i = 1 to N} and the moving average score is P,
The moving average filter is
H _j ′ =; 1 ≦ j ≦ N−P (3)
The dispersion value is 1 / P.

  That is, the noise becomes 1 / P by moving average filtering.

  FIG. 5 shows the relationship between density value and noise using the moving average score as a parameter.

  In FIG. 5, reference numeral 41 denotes a signal-to-noise ratio required for an X-ray film image reading apparatus described later, which is converted into a noise level, 42 denotes a noise level when filtering is not performed, and 43 denotes movement. The noise level when the average score is 2 is shown, and 44 is the noise level when the moving average score is 4.

  It can be seen that the noise level can be reduced by increasing the moving average score.

JP 05-137712 A Japanese Patent Application Laid-Open No. 07-135552

Originally, the signal-to-noise ratio required for an X-ray film image reader is the document High-resolution, high-performance radiographic film scanner (SPIE
According to Vol.1231 Medical Imaging IV), depending on the S / N ratio of the film, it is necessary to be one third or less of the noise of the film.

  In order to achieve this S / N ratio within the read density range, when multi-point spatial filtering is performed, noise is suppressed, but the MTF is lowered and the image is blurred. Further, if the number of spatial filtering is reduced in order to suppress the decrease in MTF, there is a problem that the SN ratio originally required for the X-ray film image reader cannot be achieved within the read density range.

  Accordingly, an object of the present invention is to provide an image processing apparatus capable of realizing an S / N ratio originally required for an X-ray film image reading apparatus over a wide density range while suppressing deterioration in image quality due to a decrease in MTF. It is in.

  In order to achieve the above object, the present invention provides a filter means having a plurality of spatial filter characteristics to which an image signal is input, a detection means for detecting a minimum density value in the vicinity of a target pixel in the image signal, and the detected Selecting means for selecting one of a plurality of spatial filter characteristics in the filter means based on a minimum density value.

  According to the present invention, it is possible to realize the SN ratio required for the X-ray film image reading apparatus over a wide density range without reducing the resolution in the low density portion.

  Embodiments of the present invention will be described below.

  FIG. 1 is a block diagram of an X-ray film image reading apparatus according to the present invention.

  In FIG. 1, light from a light source 1 such as a fluorescent lamp or a halogen lamp passes through an X-ray film 2, is condensed by an optical system lens 3, and forms an image on a CCD line sensor 4.

  Since the X-ray film 2 is sequentially conveyed in the direction of the arrow by a conveyance means (not shown), the CCD line sensor 4 receives a one-dimensional image in a direction orthogonal to the conveyance direction, so that the X-ray film 2 is detected by the CCD line sensor 4. The whole image is read by scanning.

The light received by the CCD line sensor is photoelectrically converted and output as a voltage value for each pixel. The output of the CCD line sensor is amplified by an amplifier 5 and noise is reduced by a noise reduction circuit (CDS) (not shown), and then input to an A / D converter 6 to input n-bit digital data D _i (1 ≦ i ≦ pp: the number of pixels in one line).

In the black correction circuit 7, the dark distribution Bi previously collected as an offset component from the corresponding pixel data by the subtractor 7a.
Is subtracted. The dark distribution Bi is stored in the dark distribution memory 7b.

Assuming that the output of the black correction circuit is Di ′, D _i ′ = D _i −B _i (1 ≦ i ≦ p) (4)
It becomes.

  The output of the black correction circuit 7 is input to a logarithmic conversion lookup table 8 for performing division.

The logarithmic conversion lookup table 8 is an n-bit input and an n-bit output. Specifically,
Yi = Round (A · log ₁₀ (D _i '+1) +0.5) (1 ≦ i ≦ p) (5)
A: (2 ⁿ −1) / log ₁₀ (2 ⁿ )
It becomes. Here, Round (*) means to round off the decimal part.

The logarithmic conversion lookup table 8 performs the conversion of the above equation (5) and outputs Y _i .

Next, in the white correction circuit 9, the output Y _i of the logarithmic conversion lookup table 8 is subtracted from the corresponding pixel of the light distribution correction data L _i ′ collected in advance by the subtractor 9 a. The light distribution correction data L _i ′ is stored in the light distribution memory 9b. If the light distribution correction data L _i 7 is output data from the A / D converter 6 collected without passing through L _i ,
L _i ′ = Round (A · log ₁₀ (L _i −B _i +1) +0.5) (1 ≦ i ≦ p) (6)
A: (2 ⁿ −1) / log ₁₀ (2 ⁿ )
It is represented by

  Accordingly, a division for calculating the transmittance of the film is executed by the subtraction performed by the subtracter 9a. Since this output is a logarithmic value, it becomes a density output.

The density output Z _i is
Z _i = L _i ′ −Y _i (1 ≦ i ≦ p) (7)
It becomes.

  Through the above processing, shading correction and conversion into density values are performed. The density value obtained in this way is input to the density adaptive filter 10.

  The output of the density adaptive filter 10 is written in the memory 13 and then output to an external device such as a PC.

  The memory 13 of the above black correction circuit 7, logarithmic conversion lookup table 8, white correction circuit 9, and density adaptive filter 10 is connected to the control circuit 11 via the system bus 12, and the control circuit 11 is described above. Control the process.

  Next, the density adaptive filter 10 will be described.

  FIG. 2 shows an example of the configuration of the density adaptive filter 10.

  The density adaptive filter 10 includes shift registers SR0 to SR4, weighted moving average filters 21 to 23, a minimum value detection circuit 24, and a filter selection circuit 25.

  The filter 21 is composed of multipliers 21 to 25 and adders 30 to 33 having coefficients of h10 to h14, and the filter 22 is composed of multipliers 26 to 28 and adders 34 to 35 having coefficients of h20 to h22. The filter 23 is constituted by a multiplier 29 having a coefficient h30.

  The outputs of the filters 21 to 23 and the output of the minimum value detection circuit 36 are input to the filter selection circuit 37.

  FIG. 3 is a graph showing the magnitude of each coefficient of the multiplier, that is, the weight of each pixel.

  FIG. 3A is a graph showing the magnitudes of the coefficients h10 to h14 of the filter 21, and the coefficients are set so that the weight of the pixel of interest at the center is the largest and the weight decreases as the distance from the center increases.

  The coefficients h10 to h14 have a relationship of h10 + h11 + h12 + h13 + h14 = 1.

  FIG. 3B is a graph showing the magnitudes of the coefficients h20 to h22 of the filter 22, and like the filter 21, the weight of the central pixel of interest is set to the largest value.

  The coefficients h20 to h22 have a relationship of h20 + h21 + h22 = 1.

  FIG. 3C is a graph showing the size of the coefficient h30 of the filter 23. The coefficient h30 is set to h30 = 1. Therefore, the filter 23 outputs the value of the center pixel of interest as it is.

  Next, the operation of the density adaptive filter 10 will be described.

  The image data Zi obtained as described above is input to the shift register SR4 and sequentially moved to SR3, SR2, SR1, SR0. When image data is input to all the shift registers, the outputs of the shift registers SR4 to SR0 are input to the multipliers 20 to 29.

Here, assuming that the output values of the shift registers SR4, SR3, SR2, SR1, SR0 are Z _{i 2} , Z _{i 1} , Zi, Z _{i + 1} , Z _{i + 2} (3 ≦ i ≦ p),
The output X1 _{i of the} filter 21 constituted by the multipliers 21 to 25 and the adders 30 to 33 is
X1 _i = h10 × Z _{i 2} + h11 × Z _{i 1} + h12 × Z _i + h13 × Z _{i +1}

+ H14 × Z _{i +2} (8)
(3 ≦ i ≦ p-2)
It becomes.

The output X2i of the filter 22 composed of the multipliers 26 to 28 and the adders 34 to 35 is
X2 _i = h20 × Z _{i 1} + h21 × Z _{i 2} + h22 × Z _{i +1}
... (9)
(3 ≦ i ≦ p-2)
It becomes.

The output X3i of the filter 23 constituted by the multiplier 29 is
X3 _i = h30 × Z _i (10)
(3 ≦ i ≦ p-2)
It becomes.

For example, when the filter coefficient is the value shown in FIG.
In the filter 21, √ ((1/9) ²
+ (2/9) ² + (3/9) ² + (2/9) ² + (1/9) ² ) = 0.484
In the filter 22, √ ((1/4) ²
+ (2/4) ² + (1/4) ² ) = 0.612
In the filter 23, √ ((1/1) ² ) = 1.00
Doubled.

Further, in the minimum value detection circuit 36, the minimum value among the output values Z _i -2, Z _{i -1} , Z _i , ..., Z _{i +1} , Z _{i +2 of} the shift registers SR4, ..., SR1, SR0. Value Zmini
Is detected and output.

  The outputs X1i to X3i of the filters 21 to 23 and the output Zmini of the minimum value detection circuit 36 are input to the filter selection circuit 37.

  The filter selection circuit 37 selects and outputs one of the outputs X1i to X3i of the filters 21 to 23 based on the output value Zmini of the minimum value detection circuit 36 and values corresponding to densities Q0 and Q1 shown in FIG. To do.

Specifically, the output Wi of the filter selection circuit 37 is
Wi = X1 _i (when P1 <Zmin _i ≦ 2n−1)
Wi = X2 _i (when P0 <Zmin _i ≦ P1)
Wi = X3 _i (when 0 ≦ Zmin _i ≦ P0)
P0: (2 ⁿ −1) · Q0 / l og ₁₀ (2 ⁿ )
P1: (2 ⁿ −1) · Q1 / l og ₁₀ (2 ⁿ ) (11)
It becomes.

That is, in the filter selection circuit 37, the minimum density value near the pixel of interest is within the high density region (P1 <Zmini).
In the case of ≦ 2n−1), the output of the filter 21 is selected, and the minimum density value in the vicinity of the target pixel is within the medium density region (P0 <Zmini).
In the case of ≦ P1), the output of the filter 22 is selected, and in the low concentration region (0 ≦ Zmini ≦ P0), the output of the filter 23 is selected.

  FIG. 4 shows the noise characteristics of the X-ray film image reading apparatus obtained by the above means.

  In FIG. 4, 51 indicates the noise conversion value of the S / N ratio required for the X-ray film image reader according to the above document, 52 indicates the noise characteristic when passing through the filter 23, and 53 indicates the noise characteristic when passing through the filter 22. , And 54 indicates the noise characteristics when the filter 21 is passed through.

  55 shows the noise characteristic of the X-ray film image reading device obtained by switching the filter used at the point of density Q0 from the filter 23 to the filter 22 and switching the filter used at the point of density Q1 from the filter 22 to the filter 21. Yes.

  The density Q0 is a density at which the noise level of the X-ray film image reading apparatus when the filter 21 is used is equal to the noise level required for the apparatus, and the density Q1 is the X-ray film when the filter 22 is used. This is the density at which the noise level of the image reading device is equal to the noise level required for the device.

  According to the present embodiment, when there is low-density pixel data in the vicinity of the pixel of interest, a filter that suppresses the degradation of the MTF is selected and the pixel data is output. The S / N ratio originally required for the film image reading apparatus can be realized over a wide density range.

  In the present embodiment, the description has been made using three types of filters. However, the present invention is not limited to this, and two or more filters may be used.

  In this embodiment, the outputs of a plurality of spatial filters having different characteristics are selected and output. However, a single spatial filter capable of outputting different characteristics can be used.

  In the present embodiment, the one-dimensional spatial filter has been described. However, the above-described filter can be a two-dimensional spatial filter using a line buffer or the like.

  The recording medium 14 storing the program according to the present invention may be supplied to another system or apparatus, and the computer of the system or apparatus may read and execute the program code stored in the recording medium 14.

  Here are some examples of embodiments of the present invention.

  Filter means having a plurality of spatial filter characteristics to which an image signal is input, detection means for detecting a minimum density value near the target pixel in the image signal, and a plurality of filters in the filter means based on the detected minimum density value An image processing apparatus comprising: selecting means for selecting one of the spatial filter characteristics.

  The image processing apparatus according to the first embodiment, wherein the filter unit performs filtering only in one direction of the two-dimensional image.

  The image processing apparatus according to the first embodiment, wherein the filter unit sets a target pixel in the image signal as an intermediate point of sampling pixels.

  The image signal is obtained by reading an X-ray film with an image reading unit, and the filter unit includes a plurality of signals corresponding to an S / N ratio required for the image reading unit calculated from a noise level of the X-ray film. The image processing apparatus according to the first embodiment, which has a spatial filter characteristic.

It is a block diagram showing the structure of the image process part of the image reader using this invention. It is a block diagram showing the structure of a density | concentration adaptive filter. It is a figure showing the relationship of the coefficient of a filter. It is a figure showing the relationship between the density value obtained using this invention, and noise. It is a figure which shows the relationship between the density value which used the moving average score as a parameter, and noise.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Light source 2 X-ray film 3 Optical system 4 CCD linear sensor 5 Amplifier 6 A / D converter 7 Black correction circuit 7a Subtractor 7b Dark distribution memory 8 Logarithmic conversion look-up table 9 White correction circuit 9a Subtractor 9b Bright distribution memory 10 Concentration adaptive filter 11 Control circuit 12 Bus 13 Memory 14 Storage medium SR4-0 Shift register 21-23 Filter 24 Minimum value detection circuit 25 Filter selection circuit 26-34 Multiplier 35-40 Adder

Claims (2)

  Filter means having a plurality of spatial filter characteristics to which an image signal is input, detection means for detecting a minimum density value near the target pixel in the image signal, and a plurality of filters in the filter means based on the detected minimum density value An image processing apparatus comprising: selecting means for selecting one of the spatial filter characteristics.   The image signal is obtained by reading an X-ray film by an image reading unit, and the filter unit includes a plurality of signals corresponding to an S / N ratio required for the image reading unit calculated from a noise level of the X-ray film. The image processing apparatus according to claim 1, wherein the image processing apparatus has a spatial filter characteristic.

Images