JPH09179962A - Picture processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To remove a speckle noise by storing a pixel data value exceeding a threshold on each grating point and a processing result indicating a segment exceeding the threshold as new pixel information on the corresponding grating point. SOLUTION: A comparing means 54 compares the sent results of twelve kinds of searching segments with a previously determined threshold, determines the values and directions of respective segments in the descending order of differences from the threshold and sends the determined results to a three- dimensional picture output means 56. The means 56 sets up the maximum value out of the results of the searching segments as a center pixel value and stores respective information as the processing results of grating points corresponding to directional components included in the center pixel based upon the direction of the maximum segment and that of the 2nd longer segment. The threshold of each grating point and a processing result indicating a segment exceeding the threshold are stored as new pixel information on the corresponding grating point, and after obtaining the information of all grating points, a picture is outputted based upon the information.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to image processing technology, and more particularly to image processing for medical images using ultrasonic waves and the like.

[0002]

2. Description of the Related Art Currently, images used for ultrasonic medical image processing are mainly two-dimensional. But,
Three-dimensional image processing is becoming mainstream in order to obtain more accurate information.

[0003]

An ultrasonic medical image generally has more speckle noise than an MRI image or the like, and may actually lack necessary information.

Therefore, it is necessary to remove speckle noise from an ultrasonic medical image from nearby information and interpolate necessary information.

[0005]

In order to solve the above-mentioned problems, the image processing apparatus of the present invention is directed to x, y,
Three-dimensional input of a set of physical quantities observed at each grid point corresponding to the spatial distribution of the physical properties of the target object existing in the three-dimensional space area consisting of z coordinate axes In an image processing device that outputs pixel data by processing pixel data at all grid points based on this three-dimensional pixel data, for each grid point indicating the physical quantity of the above three-dimensional space And a virtual line segment on the x, y, and z axes having a constant length centered on the grid point is at a constant angle on each of three planes including the center point perpendicular to the x, y, and z coordinate axes. Consider a search point sequence consisting of a plurality of search line segments that are rotated one by one and a plurality of auxiliary points that are respectively located on behalf of the plurality of search line segments, and in each of the above search line segments, a search that represents this search line segment Pixel at point A data value is calculated as a data value at a search point by performing interpolation processing from each of the three-dimensional pixel data, and all the search points representing this data value are integrated to obtain a pixel data value of the search line segment. And the pixel data values accumulated for each of the plurality of search line segments at each lattice point from the feature amount accumulating means are compared, and if the comparison result is larger than a predetermined threshold value, the pixel having the larger value is selected. Comparing means for selecting a plurality of data values and their line segments, pixel data values exceeding the threshold value at each grid point from the comparing means, and corresponding grid points for processing results indicating line segments exceeding the threshold value. In the image processing apparatus, an image output unit is provided, which stores the new pixel information in 1) and outputs an image based on the information after all the grid point information is obtained. That.

With the above arrangement, the present invention makes it possible to remove speckle noise and interpolate necessary missing information from nearby information from insufficient images obtained by ultrasonic diagnosis or the like.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION

(Embodiment 1) FIG. 1 shows an image processing apparatus according to a first embodiment of the present invention.

The operation of this device will be described below. In this description, the input three-dimensional pixel data 50 is x, y,
The description will be made assuming that it is composed of grid points of xmax, ymax, and zmax in the respective directions of the z axis.

First, the coordinates of the central pixel as a typical grid point to be processed are set to x = 3, y = 3, and z = 3.

In this image processing apparatus, when the three-dimensional pixel data 50 is input and the pixel data of each grid point is processed, the grid point is set to the center and the xy plane and yz including the center are respectively processed. The points of 5 × 5 pixels near the plane and the z−x plane are sent to the feature amount integrating means 52. Therefore,
At x = 3, y = 3, z = 3, the following data is sent.

Pixel data of x = 3, y = 1-5, z = 1, 5 Pixel data of x = 1-5, y = 3, z = 1-5 x = 1-5, y = 1-5 , Z = 3 pixel data In the feature amount integration means 52, the
In the neighborhood 5x5 around the central pixel 102 sent as shown in (a), 4 for each xy plane, yz plane, and zx plane as shown in FIG. 2 (b).
With respect to the types, that is, 12 types of search line segments in total for 3 planes, the filtering process for calculating the average value is performed for the representative search points, and the results for the 12 types of search line segments are sent to the comparison means 54. . That is, in FIG. 2B, the pixel at the search point f33 on the xy plane that represents each search line segment is located at the center position x on the xy plane (z = 3).
= 3, y = 3 pixels, and the search point f13 on the same plane is x
= 1 and y = 3, and the pixel data value of each search line segment is obtained by accumulating these search points.

The comparison means 54 compares the sent results of the 12 types of search line segments, and in the order of larger value than a predetermined threshold value, the value and the direction thereof are selected. The determination result is sent to the three-dimensional image output means 56. If 12 kinds of results are smaller than the threshold value, it is determined that there is no data as the minimum value of brightness, that is, 0, and the information is sent to the three-dimensional image output means 56.

In the three-dimensional image output means 56, the maximum value of the sent search line segment results is set as the central pixel value,
From the direction of the largest line segment and the direction of the second line segment from the larger order, as the direction component of the central pixel, the corresponding grid point (in this case, x = 3, y = 3, z = 3) These pieces of information are stored as the processing result.

In the above processing, the central pixel is x = 3, y
= 3, z = 3, but by updating x, y, and z one by one, pixel data of all grid points will be processed. The following is a simple explanation of the formula. for (z = 3; k <zmax-2; ++) {for (y = 3; k <ymax-2; y ++) {for (x = 3; k <xmax-2; x ++) {Filter (x, y , Z);-> neighborhood processing described so far}}}} By repeating such an operation, almost all three-dimensional pixel data can be processed. Then, the three-dimensional image output means 56 outputs an image after obtaining all the results. In addition, although the formulas shown here are not applied to the peripheral part of the calculation area, since there is often not much important information in the peripheral part, it is possible to deal with it by using the input image as it is. is there.

Therefore, when the above-described processing is performed on the input three-dimensional image data, an average value is obtained from the central pixel peripheral information, and pixel data that appears regardless of the neighborhood such as speckle noise is detected. It becomes possible to set the threshold value to remove the noise information, and to perform interpolation when there is data in the vicinity and the central pixel itself has no data.

In the present filter, for simplification of description, search points representing search line segments are averaged using the pixel data themselves at the grid points, but as shown in FIG. It seems that the same can be done by accurately considering the distances 142, 144 from the center 140, and by considering the pixel data (subpixelization) 146 at the grid points subdivided therefrom. 1 in FIG.
The straight line 50 corresponds to the processing of the filter 3 in FIG.
4 × 4 sub-pixels 160, 162, 166, 168 including 2, 154, 156, 158 are pixels f51, f4.
2, f24, f15, and the sub-pixel 4 × 4 including the center of the pixel corresponds to f33. If this is done, it is possible to finely set the direction in consideration of the degree of calculation processing as shown in FIG.

(Second Embodiment) FIG. 1 shows an image processing apparatus according to a second embodiment of the present invention. In this description, the three-dimensional pixel data 50 input in the same manner as in the first embodiment has xmax, y, and z axes in the respective directions.
The description will be made assuming that it is composed of lattice points of ymax and zmax.

The operation of this device will be described below. First,
The coordinate of the central pixel is x =
3, y = 3, z = 3.

In this image processing device, the three-dimensional pixel data 50 is input, the point of the pixel to be processed is the center, and the xy plane, the yz plane, and the z- plane including the center are respectively included.
The point of 5 × 5 pixels in the vicinity of the x-plane is sent to the feature amount integrating means 52. Therefore, when x = 3, y = 3, and z = 3, the following points are sent.

Pixel data of x = 1-5, y = 1-5, z = 1, 5 In the feature quantity integrating means 52, the neighborhood 5x similar to that shown in FIG.
The average of the pixels arranged in the search arc among the 5 pixels is calculated, and 8 types, that is, 3 planes, for each of the xy plane, the yz plane, and the zx plane as shown in FIG. 7A. A total of 24 types of search arcs are subjected to a filtering process for obtaining representative values by integrating representative search points, and the 24 types of results are sent to the comparison means 54. At this time, the 24 types of filters have almost the same curvature as a circle having a radius of 7.5 pixels as shown in FIG. 7B.

The comparison means 54 compares the sent 24 types of search arcs, determines the maximum value and the direction thereof, and sends the result to the three-dimensional image output means 56.

In the three-dimensional image output means 56, the maximum value is set as the central pixel value from the sent search circular arc results, and the central pixel has the maximum circular arc from the larger order and the central pixel from the direction of the second circular arc. As the direction component, these pieces of information are stored as the processing results of the corresponding grid points (x = 3, y = 3, z = 3 in this case).

In the above processing, the central pixel is x = 3, y
= 3, z = 3, but by updating x, y, and z one by one, pixel data of all grid points will be processed. The following is a simple explanation of the formula. for (z = 3; k <zmax-2; z ++) {for (y = 3; k <ymax-2; y ++) {for (x = 3; k <xmax-2; x ++) {Filter (x, y , Z);-> neighborhood processing described so far}}}} By repeating such an operation, almost all three-dimensional pixel data can be processed. Then, the three-dimensional image output means 56 outputs an image after obtaining all the results. Also, although the formulas shown here are not applied to the peripheral part of the calculation area, in many cases there is not much important information in the peripheral part, so it is possible to deal with it by using the input image as it is. is there.

Therefore, when the above-described processing is performed on the input three-dimensional image data, an average value is obtained from the central pixel peripheral information, and pixel data that appears regardless of the neighborhood such as speckle noise is detected. It becomes possible to set the threshold value to remove the noise information, and to perform interpolation when there is data in the vicinity and the central pixel itself has no data.

In the present filter, for simplification of description, the search points representing the search arc are averaged by using the pixel data themselves at the grid points, but they are subdivided in the calculation of the pixel data at the grid points. By considering the pixel data (sub-pixels) at the selected grid points, the same processing as in the first embodiment can be performed, and it becomes possible to set the degree of processing for finer angles in consideration of the setting.

Also, by preparing some kinds of curvatures in advance and performing some of the prepared curvatures, it becomes possible to interpolate pixels with a more fitted curvature.

(Third Embodiment) FIG. 1 shows an image processing apparatus according to a third embodiment of the present invention.

The operation of this device will be described below. In this description, the three-dimensional pixel data 50 input as in the first embodiment has xma in the directions of the x, y, and z axes.
The description will be given assuming that the grid points are x, ymax, and zmax.

First, the coordinates of the central pixel as a typical grid point to be processed are set to x = 3, y = 3, and z = 3.

In this image processing device, the three-dimensional pixel data 50 is input, and points of 5 × 5 × 5 pixels in the vicinity centering on the point of the pixel to be processed are sent to the feature amount integrating means 52. The neighborhood 5 × 5 × 5 on which the processing is performed can be numbered from 1 to 5 when the coordinate axes are x, y and z as shown in FIG. The position will be described as f (x, y, z). For example, when the coordinates of the central pixel to be processed are x = 3, y = 3, and z = 3, the following points are sent to the feature amount integrating means 52.

Pixel data of f (x, y, z) x = 1-
5, y = 1-5, z = 1,5 In the feature amount integrating means 52, the central pixel f (3,3,3) 102 of 5 × 5 × 5 sent as shown in FIG. 7 to 15 in which the virtual plane is rotated by 45 degrees with respect to each axial direction in the vicinity 5 × 5 × 5.
Filter processing for accumulating representative search points and obtaining an average value is performed on nine types of search planes as shown in FIG.
Send to

The comparing means 54 compares the sent results of the nine types of search planes, and if the maximum value among the nine types of results is greater than a predetermined threshold value, , Pixel data and the direction of the search plane are determined, and the result is sent to the processing result storage means 56. Also, if the 9 results are smaller than the threshold,
It is determined that there is no data as the minimum value of luminance, that is, 0, and that there is no corresponding search plane is sent to the three-dimensional image output means 56.

In the three-dimensional image output means 56, the value of the central pixel is set based on the maximum value sent from the comparing means, and if there is a plane, the information is used as the direction component of the central pixel. Grid points (in this case, x = 3
These pieces of information are stored as the processing result of y = 3, z = 3).

The processing up to this point is performed for f (3,3,3), but by updating x, y, z one by one, pixel data of all grid points can be processed. Become. The following is a simple explanation of the formula. for (z = 3; k <zmax-2; z ++) {for (y = 3; k <ymax-2; y ++) {for (x = 3; k <xmax-2; x ++) {Filter (x, y , Z);-> neighborhood processing described so far}}}} By repeating such an operation, almost all three-dimensional pixel data can be processed. Then, the three-dimensional image output means 56 outputs an image after obtaining all the results. In addition, although the formulas shown here are not applied to the peripheral part of the calculation area, since there is often not much important information in the peripheral part, it is possible to deal with it by using the input image as it is. is there.

Therefore, when the above-described processing is performed on the input three-dimensional image data, an average value is obtained from the central pixel peripheral information, and pixel data that appears regardless of the neighborhood such as speckle noise is detected. It becomes possible to set the threshold value to remove the noise information, and to perform interpolation when there is data in the vicinity and the central pixel itself has no data.

For simplification of description, the filter used this time averages the search points representing the search plane using the pixel data themselves at the grid points, but as shown in FIG. Accurately consider a concentric circle 142 whose distance is 1 and a concentric circle 144 whose distance is 2,
Pixel data at subdivided grid points (sub-pixelization, each pixel is 4x4x4 for the filter,
The same can be done by considering z = 3) 146. FIG. 16 shows z of the processing of the filter (2) in FIG.
= 3, for example, 152, 154, 15
4 × 4 sub-pixels 160, 16 including 6, 158
2, 166, 168 are pixels f (5, 1, 3), f (4,
Subpixel 4 × 4 corresponding to (2,3), f (2,4,3), f (1,5,3) and including the center of the pixel is f (3,3).
3, 3). For example, in the case of FIG. 12, f
The value corresponding to (5, 1, 3) can be expressed as follows in consideration of the area ratio. (4 * f (4,1,3) + 4 * f (5,1,3) + 4 *
f (4,2,3) + 4 * f (5,2,3)) / 16 By doing this, it is possible to finely set the further rotated angle in consideration of the degree of calculation processing.

(Fourth Embodiment) FIG. 1 shows an image processing apparatus according to a second embodiment of the present invention.

The operation of this device will be described below. In this description, the three-dimensional pixel data 50 input as in the first embodiment has xma in the directions of the x, y, and z axes.
The description will be given assuming that the grid points are x, ymax, and zmax.

First, the coordinates of the central pixel as a typical grid point to be processed are set to x = 3, y = 3, and z = 3.

In this image processing apparatus, the three-dimensional pixel data 50 is input, and points in the neighborhood of 5 × 5 × 5 pixels centered on the point of the pixel to be processed are sent to the feature amount integrating means 52. Similar to the first embodiment, the neighborhood 5 × 5 × 5 is 1 to 5 when the coordinate axes are x, y, and z as shown in FIG.
Can be numbered up to, and in the following description, the position of each pixel will be described as f (x, y, z). Therefore, for example, the coordinates of the central pixel to be processed are x = 3, y = 3, z = 3.
In that case, the following points are sent to the feature amount integrating means 52.

Pixel data of f (x, y, z) x = 1-
5, y = 1-5, z = 1, 5 In the filter of the search curved surface of the present embodiment, the curved surface corresponding to a spherical surface having a radius of 7.5 pixels from the central pixel shown in FIG. In FIG. 17, Z = 3 in 5 × 5 × 5
18 shows the position at the place, but representative ones of the three-dimensionally shown positions are FIG. 18, FIG. 19, and FIG.

The filters shown in FIGS. 18, 19, and 20 were each rotated by 45 degrees in each axial direction. Therefore, it is necessary to perform filter processing on a total of 24 types of search curved surfaces shown below.

The search curved surface filter (1) corresponds to FIG. 18, and the search curved surface filters (2) to (7) are each rotated by 45 degrees with respect to the z axis. Here, the term appearing in each equation corresponds to a search point that represents each search curved surface. F1 = (f (1,1,1) + f (2,2,1) + f (2,3,1) + f (3,4,1) +
f (3,5,1) + f (1,1,2) + f (2,2,2) + f (3,3,2) + f (3,4,
2) + f (3,5,2) + f (1,1,3) + f (2,2,3) + f (3,3,3) + f
(3,4,3) + f (3,5,3) + f (1,1,4) + f (2,2,4) + f (3,3,4)
+ f (3,4,4) + f (3,5,4) + f (1,1,5) + f (2,2,5) + f (2,3,
5) + f (3,4,5) + f (3,5,5)) / 25 F2 = (f (3,1,1) + f (3,2,1) + f (2,3, 1) + f (2,4,1) +
f (1,5,1) + f (3,1,2) + f (3,2,2) + f (3,3,2) + f (2,4,
2) + f (1,5,2) + f (3,1,3) + f (3,2,3) + f (3,3,3) + f
(2,4,3) + f (1,5,3) + f (3,1,4) + f (3,2,4) + f (3,3,4)
+ f (2,4,4) + f (1,5,4) + f (3,1,5) + f (3,2,5) + f (2,3,
5) + f (2,4,5) + f (1,5,5)) / 25 F3 = (f (1,3,1) + f (2,2,1) + f (3,2, 1) + f (4,2,1) +
f (5,1,1) + f (1,3,2) + f (2,2,2) + f (3,3,2) + f (4,2,
2) + f (5,1,2) + f (1,3,3) + f (2,2,3) + f (3,3,3) + f
(4,2,3) + f (5,1,3) + f (1,3,4) + f (2,2,4) + f (3,3,4)
+ f (4,2,4) + f (5,1,4) + f (1,3,5) + f (2,2,5) + f (3,2,
5) + f (4,2,5) + f (5,1,5)) / 25 F4 = (f (1,1,1) + f (2,2,1) + f (3,2, 1) + f (4,3,1) +
f (5,3,1) + f (1,1,2) + f (2,2,2) + f (3,3,2) + f (4,3,
2) + f (5,3,2) + f (1,1,3) + f (2,2,3) + f (3,3,3) + f
(4,3,3) + f (5,3,3) + f (1,1,4) + f (2,2,4) + f (3,3,4)
+ f (4,3,4) + f (5,3,4) + f (1,1,5) + f (2,2,5) + f (3,2,
5) + f (4,3,5) + f (5,3,5)) / 25 F5 = (f (3,1,1) + f (3,2,1) + f (4,3, 1) + f (4,4,1) +
f (5,5,1) + f (3,1,2) + f (3,2,2) + f (3,3,2) + f (4,4,
2) + f (5,5,2) + f (3,1,3) + f (3,2,3) + f (3,3,3) + f
(4,4,3) + f (5,5,3) + f (3,1,4) + f (3,2,4) + f (3,3,4)
+ f (4,4,4) + f (5,5,4) + f (3,1,5) + f (3,2,5) + f (4,3,
5) + f (4,4,5) + f (5,5,5)) / 25 F6 = (f (5,1,1) + f (4,2,1) + f (4,3, 1) + f (3,4,1) +
f (3,5,1) + f (5,1,2) + f (4,2,2) + f (3,3,2) + f (3,4,
2) + f (3,5,2) + f (5,1,3) + f (4,2,3) + f (3,3,3) + f
(3,4,3) + f (3,5,3) + f (5,1,4) + f (4,2,4) + f (3,3,4)
+ f (3,4,4) + f (3,5,4) + f (5,1,5) + f (4,2,5) + f (4,3,
5) + f (3,4,5) + f (3,5,5)) / 25 F7 = (f (5,3,1) + f (5,2,1) + f (3,4, 1) + f (2,4,1) +
f (1,5,1) + f (5,3,2) + f (5,2,2) + f (3,3,2) + f (2,4,
2) + f (1,5,2) + f (5,3,3) + f (5,2,3) + f (3,3,3) + f
(2,4,3) + f (1,5,3) + f (5,3,4) + f (5,2,4) + f (3,3,4)
+ f (2,4,4) + f (1,5,4) + f (5,3,5) + f (5,2,5) + f (3,4,
5) + f (2,4,5) + f (1,5,5)) / 25 F7 = (f (5,1,1) + f (4,2,1) + f (3,4, 1) + f (2,3,1) +
f (1,3,1) + f (5,1,2) + f (4,2,2) + f (3,3,2) + f (2,3,
2) + f (1,3,2) + f (5,1,3) + f (4,2,3) + f (3,3,3) + f
(2,3,3) + f (1,3,3) + f (5,1,4) + f (4,2,4) + f (3,3,4)
+ f (2,3,4) + f (1,3,4) + f (5,1,5) + f (4,2,5) + f (3,4,
5) + f (2,3,5) + f (1,3,5)) / 25 F8 = (f (1,1,1) + f (1,2,1) + f (1,2, 3) + f (1,3,4) +
f (1,3,5) + f (2,1,1) + f (2,2,2) + f (2,3,3) + f (2,3,
4) + f (2,3,5) + f (3,1,1) + f (3,2,3) + f (3,3,3) + f
(3,3,4) + f (3,3,5) + f (4,1,1) + f (4,2,4) + f (4,3,3)
+ f (4,3,4) + f (4,3,5) + f (5,1,1) + f (5,2,5) + f (5,2,
3) + f (5,3,4) + f (5,3,5)) / 25 The search surface filter (9) corresponds to FIG. 19, and the search surface filters (10)-(16) are It is rotated by 45 degrees with respect to the z-axis. F9 = (f (1,1,1) + f (1,2,2) + f (1,2,3) + f (1,3,4) +
f (1,3,5) + f (2,1,1) + f (2,2,2) + f (2,3,3) + f (2,3,
4) + f (2,3,5) + f (3,1,1) + f (3,2,2) + f (3,3,3) + f
(3,3,4) + f (3,3,5) + f (4,1,1) + f (4,2,2) + f (4,3,3)
+ f (4,3,4) + f (4,3,5) + f (5,1,1) + f (5,2,2) + f (5,2,
3) + f (5,3,4) + f (5,3,5)) / 25 F10 = (f (1,1,5) + f (1,2,4) + f (1,2, 3) + f (1,3,2) +
f (1,3,1) + f (2,1,5) + f (2,2,4) + f (2,3,3) + f (2,3,
2) + f (2,3,1) + f (3,1,5) + f (3,2,4) + f (3,3,3) + f
(3,3,2) + f (3,3,1) + f (4,1,5) + f (4,2,4) + f (4,3,3)
+ f (4,3,2) + f (4,3,1) + f (5,1,5) + f (5,2,4) + f (5,2,
3) + f (5,3,2) + f (5,3,1)) / 25 F11 = (f (1,1,3) + f (1,2,3) + f (1,3, 2) + f (1,4,2) +
f (1,5,1) + f (2,1,3) + f (2,2,3) + f (2,3,3) + f (2,4,
2) + f (2,5,1) + f (3,1,3) + f (3,2,3) + f (3,3,3) + f
(3,4,2) + f (3,5,1) + f (4,1,3) + f (4,2,3) + f (4,3,3)
+ f (4,4,2) + f (4,5,1) + f (5,1,3) + f (5,2,3) + f (5,3,
2) + f (5,4,2) + f (5,5,1)) / 25 F12 = (f (1,1,1) + f (1,2,2) + f (1,3 ,, 2) + f (1,4,3) +
f (1,5,3) + f (2,1,1) + f (2,2,2) + f (2,3,3) + f (2,4,
3) + f (2,5,3) + f (3,1,1) + f (3,2,2) + f (3,3,3) + f
(3,4,3) + f (3,5,3) + f (4,1,1) + f (4,2,2) + f (4,3,3)
+ f (4,4,3) + f (4,5,3) + f (5,1,1) + f (5,2,2) + f (5,3,
2) + f (5,4,3) + f (5,5,3)) / 25 F13 = (f (1,3,1) + f (1,3,2) + f (1,4, 3) + f (1,4,4) +
f (1,5,5) + f (2,3,1) + f (2,3,2) + f (2,3,3) + f (2,4,
4) + f (2,5,5) + f (3,3,1) + f (3,3,2) + f (3,3,3) + f
(3,4,4) + f (3,5,5) + f (4,3,1) + f (4,3,2) + f (4,3,3)
+ f (4,4,4) + f (4,5,5) + f (5,3,1) + f (5,3,2) + f (5,4,
3) + f (5,4,4) + f (5,5,5)) / 25 F14 = (f (1,5,1) + f (1,4,2) + f (1,4, 3) + f (1,3,4) +
f (1,3,5) + f (2,5,1) + f (2,4,2) + f (2,3,3) + f (2,3,
4) + f (2,3,5) + f (3,5,1) + f (3,4,2) + f (3,3,3) + f
(3,3,4) + f (3,3,5) + f (4,5,1) + f (4,4,2) + f (4,3,3)
+ f (4,3,4) + f (4,3,5) + f (5,5,1) + f (5,4,2) + f (5,4,
3) + f (5,3,4) + f (5,3,5)) / 25 F15 = (f (1,5,3) + f (1,4,3) + f (1,3, 4) + f (1,2,4) +
f (1,1,5) + f (2,5,3) + f (2,4,3) + f (2,3,3) + f (2,2,
4) + f (2,1,5) + f (3,5,3) + f (3,4,3) + f (3,3,3) + f
(3,2,4) + f (3,1,5) + f (4,5,3) + f (4,4,3) + f (4,3,3)
+ f (4,2,4) + f (4,1,5) + f (5,5,3) + f (5,4,3) + f (5,3,
4) + f (5,2,4) + f (5,1,5)) / 25 F16 = (f (1,5,5) + f (1,4,4) + f (1,3 ,, 4) + f (1,2,3) +
f (1,1,3) + f (2,5,5) + f (2,4,4) + f (2,3,3) + f (2,2,
3) + f (2,1,3) + f (3,5,5) + f (3,4,4) + f (3,3,3) + f
(3,2,3) + f (3,1,3) + f (4,5,5) + f (4,4,4) + f (4,3,3)
+ f (4,2,3) + f (4,1,3) + f (5,5,5) + f (5,4,4) + f (5,3,
4) + f (5,2,3) + f (5,1,3)) / 25 The search surface filter (17) corresponds to FIG. 20, and the following search surface filters (18)-(24) are It is rotated by 45 degrees with respect to the z-axis. F17 = (f (1,1,1) + f (2,1,2) + f (2,1,3) + f (3,1,4) +
f (3,1,5) + f (1,2,1) + f (2,2,2) + f (3,2,3) + f (3,2,
4) + f (3,2,5) + f (1,3,1) + f (2,3,2) + f (3,3,3) + f
(3,3,4) + f (3,3,5) + f (1,4,1) + f (2,4,2) + f (3,4,3)
+ f (3,4,4) + f (3,4,5) + f (1,5,1) + f (2,5,2) + f (2,5,
3) + f (3,5,4) + f (3,5,5)) / 25 F18 = (f (1,1,5) + f (2,1,4) + f (2,1, 3) + f (3,1,2) +
f (3,1,1) + f (1,2,5) + f (2,2,4) + f (3,2,3) + f (3,2,
2) + f (3,2,1) + f (1,3,5) + f (2,3,4) + f (3,3,3) + f
(3,3,2) + f (3,3,1) + f (1,4,5) + f (2,4,4) + f (3,4,3)
+ f (3,4,2) + f (3,4,1) + f (1,5,5) + f (2,5,4) + f (2,5,
3) + f (3,5,2) + f (3,5,1)) / 25 F19 = (f (1,1,3) + f (2,1,3) + f (3,1, 2) + f (4,1,2) +
f (5,1,1) + f (1,2,3) + f (2,2,3) + f (3,2,3) + f (4,2,
2) + f (5,2,1) + f (1,3,3) + f (2,3,3) + f (3,3,3) + f
(4,3,2) + f (5,3,1) + f (1,4,3) + f (2,4,3) + f (3,4,3)
+ f (4,4,2) + f (5,4,1) + f (1,5,3) + f (2,5,3) + f (3,5,
2) + f (4,5,2) + f (5,5,1)) / 25 F20 = (f (1,1,1) + f (2,1,2) + f (3,1, 2) + f (4,1,3) +
f (5,1,3) + f (1,2,1) + f (2,2,2) + f (3,2,3) + f (4,2,
3) + f (5,2,3) + f (1,3,1) + f (2,3,2) + f (3,3,3) + f
(4,3,3) + f (5,3,3) + f (1,4,1) + f (2,4,2) + f (3,4,3)
+ f (4,4,3) + f (5,4,3) + f (1,5,1) + f (2,5,2) + f (3,5,
2) + f (4,5,3) + f (5,5,3)) / 25 F21 = (f (3,1,1) + f (3,1,2) + f (4,1, 3) + f (4,1,4) +
f (5,1,5) + f (3,2,1) + f (3,2,2) + f (3,2,3) + f (4,2,
4) + f (5,2,5) + f (3,3,1) + f (3,3,2) + f (3,3,3) + f
(4,3,4) + f (5,3,5) + f (3,4,1) + f (3,4,2) + f (3,4,3)
+ f (4,4,4) + f (5,4,5) + f (3,5,1) + f (3,5,2) + f (4,5,
3) + f (4,5,4) + f (5,5,5)) / 25 F22 = (f (5,1,1) + f (4,1,2) + f (4,1, 3) + f (3,1,4) +
f (3,1,5) + f (5,2,1) + f (4,2,2) + f (3,2,3) + f (3,2,
4) + f (3,2,5) + f (5,3,1) + f (4,3,2) + f (3,3,3) + f
(3,3,4) + f (3,3,5) + f (5,4,1) + f (4,4,2) + f (3,4,3)
+ f (3,4,4) + f (3,4,5) + f (5,5,1) + f (4,5,2) + f (4,5,
3) + f (3,5,4) + f (3,5,5)) / 25 F23 = (f (5,1,3) + f (4,1,3) + f (3,1, 4) + f (2,1,4) +
f (1,1,5) + f (5,2,3) + f (4,2,3) + f (3,2,3) + f (2,2,
4) + f (1,2,5) + f (5,3,3) + f (4,3,3) + f (3,3,3) + f
(2,3,4) + f (1,3,5) + f (5,4,3) + f (4,4,3) + f (3,4,3)
+ f (2,4,4) + f (1,4,5) + f (5,5,3) + f (4,5,3) + f (3,5,
4) + f (2,5,4) + f (1,5,5)) / 25 F24 = (f (5,1,5) + f (4,1,4) + f (3,1, 4) + f (2,1,3) +
f (1,1,3) + f (5,2,5) + f (4,2,4) + f (3,2,3) + f (2,2,
3) + f (1,2,3) + f (5,3,5) + f (4,3,4) + f (3,3,3) + f
(2,3,3) + f (1,3,3) + f (5,4,5) + f (4,4,4) + f (3,4,3)
+ f (2,4,3) + f (1,4,3) + f (5,5,5) + f (4,5,4) + f (3,5,
4) + f (2,5,3) + f (1,5,3)) / 25 In the feature quantity integrating means 52, the neighborhood 5 × 5 × 5 similar to that in FIG.
The average of the pixels lined up in the curved surface in the pixel is calculated, the representative search points are added to the 24 types of search curved surfaces shown above, the average value is filtered, and the results of the 24 types are compared. Send to means 54.

The comparison means 54 compares the sent results of the 24 types of search curved surfaces, determines the maximum value and the direction thereof, and sends the result to the three-dimensional image output means 56. If 24 kinds of results are smaller than the threshold value, it is determined that there is no data with the minimum value of brightness, that is, 0, and that there is no search surface.
It is sent to the three-dimensional image output means 56.

In the three-dimensional image output means 56, the value of the center pixel is set based on the maximum value sent from the comparison means, and if there is a search curved surface, the information is used as the direction component of the center pixel. The corresponding grid point (in this case x =
These information is stored as the processing result of (3, y = 3, z = 3).

The processing up to this point was carried out for f (3,3,3), but by updating x, y, z one by one, pixel data of all grid points can be processed. Become. The following is a simple explanation of the formula. for (z = 3; k <zmax-2; z ++) {for (y = 3; k <ymax-2; y ++) {for (x = 3; k <xmax-2; x ++) {Filter (x, y , Z);-> neighborhood processing described so far}}}} By repeating such an operation, almost all three-dimensional pixel data can be processed. Then, the three-dimensional image output means 56 outputs an image after obtaining all the results. In addition, although the formulas shown here are not applied to the peripheral part of the calculation area, since there is often not much important information in the peripheral part, it is possible to deal with it by using the input image as it is. is there.

Therefore, when the above-described processing is performed on the input three-dimensional image data, an average value is obtained from the central pixel peripheral information, and pixel data that appears regardless of the neighborhood, such as speckle noise, is obtained. It becomes possible to set the threshold value to remove the noise information, and to perform interpolation when there is data in the vicinity and the central pixel itself has no data. Further, the difference from the third embodiment is that the curved shape of the filter in the feature amount integrating means 52 makes it possible to interpolate even a curved shape.

For simplification of description, the search points representing the search curved surface are averaged by using the pixel data themselves at the grid points in the present filter, but they are subdivided in the calculation of the pixel data at the grid points. By taking into consideration the pixel data (sub-pixels) at the selected grid points, the same operation as in the third embodiment can be performed, and the degree of processing can be set considering the finer angle.

Also, some kinds of curved surfaces are prepared in advance, and some of the prepared curvatures are used to remove speckle noise, and pixels can be interpolated by the fitted search curved surface.

[0050]

As described above, according to the present invention, a filter processing is performed on pixels which are three-dimensionally close to each other, and a part of a necessary image obtained by ultrasonic diagnosis is missing.
Interpolation processing can be performed, and a more realistic image can be reproduced.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of an image processing apparatus according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram of a filter according to the first embodiment of the present invention.

FIG. 3 is an explanatory diagram of an expanded filter according to the same embodiment.

FIG. 4 is an explanatory view of a rotation angle of the filter according to the same embodiment.

FIG. 5 is an explanatory diagram of a filter according to a second embodiment of the present invention.

FIG. 6 is an explanatory diagram of a filter according to a third embodiment of the present invention.

FIG. 7 is an explanatory diagram of a filter according to a third embodiment of the present invention.

FIG. 8 is an explanatory diagram of a filter according to a third embodiment of the present invention.

FIG. 9 is an explanatory diagram of a filter according to a third embodiment of the present invention.

FIG. 10 is an explanatory diagram of a filter according to a third embodiment of the present invention.

FIG. 11 is an explanatory diagram of a filter according to the third embodiment of the present invention.

FIG. 12 is an explanatory diagram of a filter according to a third embodiment of the present invention.

FIG. 13 is an explanatory diagram of a filter according to the third embodiment of the present invention.

FIG. 14 is an explanatory diagram of a filter according to the third embodiment of the present invention.

FIG. 15 is an explanatory diagram of a filter according to a third embodiment of the present invention.

FIG. 16 is an explanatory diagram of a filter according to the third embodiment of the present invention.

FIG. 17 is an explanatory diagram of a filter according to the fourth embodiment of the present invention.

FIG. 18 is an explanatory diagram of a filter according to the fourth embodiment of the present invention.

FIG. 19 is an explanatory diagram of a filter according to the fourth embodiment of the present invention.

FIG. 20 is an explanatory diagram of a filter according to the fourth embodiment of the present invention.

[Explanation of symbols]

50 three-dimensional pixel data to be input 52 feature amount integrating means 54 comparing means 56 three-dimensional image output means 102 center pixel 140 center of pixel 142 concentric circle indicating that pixel distance is 1 144 distance of pixel is 2 Concentric circles shown 146 Pixels (subpixels) obtained by subdividing one pixel into 16 divisions (4 × 4) 150 Straight line for interpolation 152 Intersection at a pixel distance of 2 154 Intersection at a pixel distance of 1 156 Pixel distance at 1 Intersection point at 160 subpixels at a distance of 2 pixels (4 ×
4) 160 subpixels at a distance of 2 (4 ×
4) 162 sub-pixels at a distance of 1 pixel (4 ×
4) Sub-pixel (4 × 4) at a distance of 164 pixels is 0
4) 166 sub-pixels at a distance of 1 (4 × 4)
4) 168 sub-pixels at a distance of 1 (4 × 4)
4) 180 Search line segment indicating the angle of rotation of the virtual line segment 1 182 Search line segment indicating the angle of rotation of the virtual line segment 2 2 184 Search line segment indicating the angle of rotation of the virtual line segment 3 186 Search line segment showing the angle of rotating the virtual line segment 4 188 Search line segment showing the angle of rotating the virtual line segment 5 190 Search line segment showing the angle of rotating the virtual line segment 6 202 Center Pixel

Claims (4)

[Claims] 1. A grid point is constructed in a three-dimensional space area consisting of x, y, z coordinate axes orthogonal to each other, and at each grid point corresponding to a spatial distribution of physical properties of an object existing in the area. In the image processing device, the set of observed physical quantities is set as input three-dimensional pixel data, and the pixel data at all grid points are processed based on the three-dimensional pixel data to output an image. For each of all the lattice points indicating the physical quantity of the space, x, y, with a constant length centered on the lattice point,
A plurality of search line segments, each of which is a virtual line segment on the z axis rotated by a certain angle on each of three planes including the center point perpendicular to the x, y, and z coordinate axes, and the plurality of search line segments are representative. Consider a search point sequence consisting of a plurality of auxiliary points located at each of the search points, and interpolate the pixel data values at the search points that are representative of this search line segment from each of the three-dimensional pixel data. A feature amount integrating unit that calculates as a data value at a running search point and further integrates all search points that are representative of this data value to obtain a pixel data value of a search line segment; and the plurality of grid points at each grid point from the feature amount integrating unit. Pixel data values for each search line segment are compared. If the comparison result is greater than a predetermined threshold value, multiple pixel data values and line segments are selected from the larger one. The comparing means and the processing result indicating the pixel data value exceeding the threshold value at each grid point from the comparing means and the line segment exceeding the threshold value are accumulated as new pixel information at the corresponding grid point. An image processing apparatus comprising image output means for outputting an image based on the information of all the grid points after the information is obtained. 2. A grid point is formed in a three-dimensional space area consisting of x, y, z coordinate axes that are orthogonal to each other, and each grid point corresponds to a spatial distribution of physical properties of an object existing in the area. In the image processing device, the set of observed physical quantities is set as input three-dimensional pixel data, and the pixel data at all grid points are processed based on the three-dimensional pixel data to output an image. A virtual arc on the x, y, and z axes having a constant length and a constant curvature centered on the lattice point is perpendicular to the x, y, z coordinate axes for all the lattice points indicating the physical quantity of the space. Consider a search point sequence consisting of a plurality of search arcs each rotated by a certain angle on three planes including the center point, and a plurality of auxiliary points positioned at each of the search arcs representing the plurality of search arcs. In an arc Then, the pixel data value at the search point that represents this search arc is calculated as a data value at the search point by performing interpolation processing from each of the three-dimensional pixel data, and all search points that represent this data value are integrated to obtain the search arc. Of the pixel data value and the pixel data value integrated for each of the plurality of search arcs at each lattice point from the feature amount integrating means are compared, and the comparison result is a predetermined threshold value. If it is larger, the comparing means for selecting the pixel data value and the plurality of arcs from the larger one, and the pixel data value exceeding the threshold value and the arc exceeding the threshold value at each grid point from the comparing means are shown. The processing result is stored as new pixel information at the corresponding grid point, and after all grid point information is obtained, image output is performed based on that information. The image processing apparatus according to claim providing an image output unit. 3. A grid point is formed in a three-dimensional space area consisting of x, y, z coordinate axes that are orthogonal to each other, and each grid point corresponds to a spatial distribution of physical properties of an object existing in the area. In the image processing device, the set of observed physical quantities is set as input three-dimensional pixel data, and the pixel data at all grid points are processed based on the three-dimensional pixel data to output an image. For each of all the lattice points indicating the physical quantity of the space, xy, yz, z centered on the lattice point
-A plurality of search planes obtained by rotating a virtual square plane having a constant side length on the x plane by a certain angle with respect to the x, y, and z coordinate axes, and the positions representing the search planes. Considering a search point composed of a plurality of auxiliary points, the pixel data value at the search point representing the search plane in each search plane is interpolated from each of the three-dimensional pixel data to obtain a data value at the search point. A feature amount integrating unit that calculates and further integrates all search points representing this data value to obtain a pixel data value of the search plane, and pixels integrated for each of the plurality of search planes at each grid point from the feature amount integrating unit. When the data values are compared, and the comparison result is larger than a predetermined threshold value, the maximum pixel data value and the plane for selecting the plane are compared. And the maximum pixel data value exceeding the threshold value at each grid point from the comparison means and the processing result indicating the plane are accumulated as new pixel information at the corresponding grid point, and all grid point information is stored. An image processing apparatus comprising an image output means for outputting an image based on the information obtained. 4. A grid point is formed in a three-dimensional space area consisting of x, y, z coordinate axes orthogonal to each other, and at each grid point corresponding to the spatial distribution of the physical properties of the object existing in the area. In the image processing device, the set of observed physical quantities is set as input three-dimensional pixel data, and the pixel data at all grid points are processed based on the three-dimensional pixel data to output an image. A virtual curved surface having a constant area and a constant curvature centered on the grid point is rotated by a fixed angle with respect to each of the x, y, and z coordinate axes for each of all the grid points indicating the physical quantity of space. Consider a search point consisting of a plurality of search curved surfaces and a plurality of auxiliary points positioned at each of the plurality of search curved surfaces, and in each of the search curved surfaces, A feature amount accumulating means for calculating a raw data value as a data value at a search point by performing an interpolation process from each of the three-dimensional pixel data, and further integrating all the search points representing this data value to obtain a pixel data value of the search curved surface; , Comparing the pixel data values accumulated for each of the plurality of search curved surfaces at each grid point from the feature amount accumulating means, and if the comparison result is larger than a predetermined threshold value, the maximum pixel data value And a comparison means for selecting the curved surface, and storing the maximum pixel data value exceeding the threshold value at each grid point from the comparison means and the processing result indicating the curved surface as new pixel information at the corresponding grid point. An image processing apparatus comprising image output means for outputting an image based on the information of all the grid points after the information is obtained.