JPH01109483A - Image processor - Google Patents

Abstract

PURPOSE:To prevent a noise from being prominent by executing suitable density emphasis even when there is a large part in which the density of an image is uniform. CONSTITUTION:A noise component included in picture data is reduced by a digital filter operating part 12 and next, a variance value is obtained from the density value of a pixel. Continuously, the variance value is added to variance value data (be prepared in a separate memory 6) corresponding to the density of the respective pixels. Based on this result, the probability integrated value of the sum is calculated in a probability integrated value calculating part 5 and succeedingly, based on this probability integrated value, a window curve is determined by an image density converting part 8. Then, a density converting processing is executed. Thus, concerning the picture data whose noise is reduced, since the window curve can be determined to execute a weighting in correspondence to the uniformity of the density (namely, the size of dispersion), the suitable density emphasis can be executed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to an image processing apparatus that performs window processing on X-ray image data and the like and displays the results as an image.

(Prior Art) Conventionally, a mask image taken before a contrast agent is mixed into a desired diagnostic site of a subject and a contrast image taken after a contrast medium is mixed into the diagnostic site are input into a drawing device. ,
This subtraction device subtracts the density values of each of the images (thereby extracting an image of only the region where the contrast agent is present, that is, a subtraction image).

By the way, by exposing the subject to X-rays, the subject's Such conversion processing is called window processing. Conventionally, to perform such window processing, the operator inputs the window value [window value] via a switch, etc. provided on the operation panel. Width (WW) and Window Level (WL)], however, there were drawbacks such as it took time to set the desired window value and the operation became difficult.

Therefore, in order to improve these problems, the present applicant has previously filed an invention as Japanese Patent Application No. 61-41933, which allows window processing conditions to be automatically set without operator intervention. This invention calculates the appearance frequency (histogram) of the image density for the captured image data to obtain an appearance frequency curve as shown in 15 in FIG. After obtaining the curve, this is used as the window processing condition as a window curve as shown in FIG. 9 to perform the density conversion processing of the image data. By performing each of the above operations using a plurality of calculation means, the window processing conditions can be automatically set, so that the results can be quickly displayed as an image.

(Problems to be Solved by the Invention) By the way, as a processing method for determining a wind curve as shown in FIG. 9 using the probability integral curve 16 obtained based on the appearance frequency curve 15 as in the above invention, A HE (adap
tive histgraa equal 1zati
There is a method called ``on''. However, this processing method has the following drawbacks.

If a large part of the image has a substantially uniform m degree, that density region will be greatly emphasized, making noise (for example, photon noise in the case of an X-ray image) noticeable. In other words, as is clear from FIGS. 8 and 9, image density areas with a large number of corresponding pixels tend to be displayed more emphasized, so the wider the uniform density area, the wider the image density area. The more these areas become concentrated, the more concentrated they become.

The present invention has been made to address these problems, and provides an image processing device that makes noise less noticeable by appropriately emphasizing density even if the image has a fairly uniform density area. The purpose is to

[Structure of the Invention] (Means for Solving the Problems) In order to achieve the above object, the present invention provides a noise reduction means for reducing noise contained in image data, and an image data for the image data with the noise reduced. a first calculation means for calculating a value representing the variation in density corresponding to the density; and a second calculation means for calculating a probability integral value of the sum of the values representing the variation in image density based on the output of the first calculation means. The present invention is characterized in that it comprises a calculation means, and a third calculation means that performs density conversion processing on the image data based on the output of the second calculation means.

(Operation) First, noise components included in the image data are reduced, and then a variance value is determined from the density value of the vixel. Subsequently, the dispersion value is added to the dispersion value data (prepared in a separate memory) corresponding to the density of each pixel. Next, based on this result, a probability integral value of the sum is calculated, and then a wind curve is determined based on this probability integral value, and density conversion processing is performed.

As a result, it is possible to determine a wind curve that is weighted according to the uniformity of density (that is, the magnitude of variation) for image data with reduced noise, so that appropriate density enhancement can be performed.

(Embodiment) FIG. 1 is a block diagram showing an image processing apparatus according to an embodiment of the present invention.
L) 2, first image memory (181)3. Second image memory (IH2) 11, digital filter calculation unit (A3) 12. Arithmetic section (AI) 4 (hereinafter referred to as the computing means of tube 1), probability integral value manipulation section (A2) 5 (hereinafter referred to as the computing means of tube 2) 1. Dispersion histogram data storage memory (HEHI)6. Memory for storing probability integral values (
HEt42) 7, Image density conversion unit (GTA) 8
(Third calculation means), D/A converter 92
It consists of a display device (DISP) 10.

Pass line 1 is a line for exchanging data between each block. The II 60 unit 2 is composed of a CPLJ and the like and is in charge of overall operation control.

The second image memory 11 is a memory for storing various image data including X-ray information and the like before being subjected to noise reduction processing, that is, filter processing, for noise contained in the image data.

Digital filter calculation unit (as a noise reduction means)
A3) When the image data stored in the second image memory 11 and the coefficient values of a digital filter, such as a low-pass filter, from the arithmetic control unit (CTRL) 2 are input, 12 detects noise contained in the image data. It has a function of performing noise reduction processing and outputting image data that has undergone noise reduction processing to the first image memory 3 specified by the calculation t11 control unit (CTRL) 2.

The first image memory 3 is a memory that stores image data that has been subjected to noise reduction processing (filter processing) in the digital filter calculation section (A3) 12 under the control of the calculation control section 2.

When image data is input, the first calculation means 4 divides the image data into a plurality of blocks, for example, 64 blocks B1, 82, B3, . . . as shown in FIG.

It is divided into B64 blocks, and each block has a plurality of pixels, for example, 16 pixels Pl, P, as shown in FIG.
2, P3, . . . , P city, first, each block B1, 82, B3, .

For example, a value representing variation in image density, such as variance, is calculated. That is, the variance value is calculated from the density values of 16 pixels for one block. Next, the first calculating means 4 adds the dispersion value to dispersion data (stored in a dispersion histogram data storage memory 6 to be described later) corresponding to the density of each pixel of each block.

That is, the sum of values representing variations in each image density is calculated.

The dispersion histogram data storage memory 6 stores the calculation results by the first calculation means 4. For example, this is the fourth
They are stored in the correspondence relationship shown in the figure. Note that this memory 6 is initially set to an initial state in which the contents are cleared.

The second calculation means 5 reads out the data stored in the memory 6, and calculates the ratio Pi to the number of pixel density values obtained corresponding to the pixel density value di, as shown in the following equation.

Then, the sum Si of Pi from the minimum pixel density to the current pixel density is calculated for all pixel densities as shown in the following equation.

That is, a probability integral value of the sum of values representing variations in each image RAM degree is calculated.

The probability integral value storage memory 7 stores the exclusion results obtained by the second calculation means 5. This is stored in a correspondence relationship as shown in FIG. 5, for example. - As an example, a case where an image is constructed with 8 bits is shown.

When the image data is input, the third calculation means 8 reads out the data stored in the memory 7, performs density conversion processing, and calculates the pixel density value to be displayed according to the pixel density value of the image. The image is output to the A converter 9 and the corresponding image is displayed on the second display device 10.

Next, the functions and effects of this embodiment will be explained.

The image data to be displayed is stored in the second image memory (1
82) Assume that it is stored in 11.

First, based on the control of the calculation control unit 2, the digital filter calculation unit (A3) 12 has a second image memory (182).
The image data stored in 11 and the coefficient values of the low-pass filter are input. The digital filter calculation unit (A3) 12 performs filter processing on the input image data and stores the processing results in the first image memory (18).
1) Output to 3.

The image data stored in the first image memory 3 is input to the first calculation means 4 under the control of the calculation control section 2.

The first calculation means 4 divides the input image data into a plurality of blocks as shown in FIG. 2, and calculates a variance value from the density value for each pixel in each block as shown in FIG. Next, the variance value is added to the variance data in the memory 6 corresponding to the density of each pixel in each block, and a sum indicating the variation in each image density is calculated based on the correspondence relationship as shown in FIG. This sum 11 is shown in FIG. Note that 22 in the figure is a sum indicating the variation in each image density when no noise reduction means is provided. Subsequently, the second calculation means 5 reads data from the memory 6 and calculates the probability integral value of the sum of values representing variations in image density using the correspondence relationship as shown in FIG. FIG. 6 shows this probability integral -112. Note that 20 in the figure is a probability integral curve when no noise reduction means is provided.

The third calculation means 8 reads data from the memory 7 in which the probability integral value data is stored, determines a wind curve 13 based on the probability integral curve 12 as shown in FIG. A density conversion process is performed on the image data stored in the memory 11, and the image is displayed on the display device 2i10 via the D/A converter 9. Actually, if the display density is 0 to 255 (8 bits), the value obtained by multiplying the integral ISi of the proportion of the dispersed phase by 255 becomes the data for display.

According to this embodiment, probability integral curve 1 in FIG.
2, weighting is performed according to the density variation that reduces the variation due to noise, and this (
Based on this, the wind curve 13 in FIG. 7 is determined, so that appropriate density enhancement (density conversion processing is performed) according to the uniformity of the density. Note that noise reduction means is provided for 2 in the figure. This is the wind curve when there is no wind curve. Therefore, even if there is a wide area with uniform density, appropriate density enhancement is performed without extreme density enhancement. Therefore, since noise is not noticeable, diagnostic performance can be improved.

To be more specific, FIGS. 6 and 7 also show characteristic curves that do not involve noise reduction means, and it can be seen that there are many pixels in the vicinity of the pixel values. These are DS
It shows the background part of the subtraction image in A (digital subtraction angiography).

The value of the background portion of this subtraction image is a portion obtained by subtracting the contrast image from the mask image, so in principle it is zero, and this background portion occupies a large area in the image.

If the noise reduction means is not used, the relationship between each pixel value and the dispersed phase will be as shown by 20 in FIG. 6, and the wind curve in this case will be as shown in 2) in FIG. 7. In this case, the background portion contains a dispersed phase caused by noise. On the other hand, as shown in this example, when using a noise reduction means, the dispersed phase caused by the noise contained in the background part is significantly reduced, and is relatively It can be set to an extremely small value. Therefore, the wind curve can also be made such that the density emphasis of the background portion is further suppressed.

By the way, in this embodiment, even when noise filter processing is performed, image data other than noise hardly changes, so image data that has been filtered through a noise reduction means is used for pixel values or block average values. ing. This aims to shorten calculation time.

On the other hand, if such calculation time is not considered, the variance value of the corresponding block of the image after noise reduction processing is calculated based on the pixel value or block average value of the original image data to be displayed. It may be made to correspond.

It should be noted that the present invention is not limited to the above-mentioned embodiments, and various modifications can be made within the scope of the invention.

For example, in the above embodiments, the dispersion value was shown as the value representing the variation in concentration, but the standard deviation value may also be used.

Alternatively, the absolute value of the data difference between adjacent pixels may be used.

In short, any value can be used as long as it can represent the amount of density variation. Furthermore, the object of calculation does not have to be the entire image, but may be a part of the image. Furthermore, it is not always necessary to divide the data into blocks.

[Effects of the Invention] As described above, according to the present invention, it is possible to perform appropriate density enhancement even if there are many areas with uniform density in an image, so that it is possible to prevent noise from becoming noticeable. . Moreover, from this, it is also possible to improve the diagnostic ability.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an image processing device according to an embodiment of the present invention, FIGS. 2 and 3 are data arrangement diagrams showing the principle of the present invention, and FIGS. 4 and 5 are data calculation examples of this embodiment. FIGS. 6 and 7 are characteristic curve diagrams obtained by this embodiment, and FIGS. 8 and 9 are characteristic curve diagrams of the conventional example. 4...Dispersion histogram calculation unit (first calculation means),
5... Probability integral value calculation unit (second calculation means), 6...
- Memory for storing distributed histogram data, 7... Memory for storing probability integral values, 8... Image*m degree conversion unit (third calculation means), 11... Second image memory, 12...・
Digital filter calculation section (noise reduction means). 7℃...Tsuku no Giri 91 Flies Captive - Straight 16 wide ml' I double layered Koji-like 7-C1w Evening 5P Tsu 9 trenches 4 Turtles Figure 8 aS Mioshi torture straight Figure 9

Claims (3)

[Claims] (1) In an image processing device that performs window processing on image data and displays the result as an image, a noise reduction means that reduces noise included in the image data, and a method that corresponds to the image density of the image data from which the noise has been reduced. a first calculation means for calculating a value representing the variation in density; and a second calculation means for calculating a probability integral value of the sum of the values representing the variation in image density based on the output of the first calculation means; ,
An image processing device comprising: third calculation means that performs density conversion processing on the image data based on the output of the second calculation means. (2) The image processing device according to claim 1, wherein the value representing the variation in image density is a variance or a standard deviation. (3) The image processing device according to claim 1, wherein the value representing the variation in image density is the absolute value of the difference in data between adjacent pixels forming the image.