JPH05253215A - Image processing device - Google Patents

Abstract

PURPOSE:To extract the edge of a minute part on a light/dark image by preparing a plurality of curves, which exhibit density variation from picture element to picture element lying on the reference axis intersecting the scanning line of the light/dark image, while the reference axis is moved in the dense/pale image, distinguishing the representing the features among the curves, and thereupon extracting the feature point. CONSTITUTION:Using an emphasized image I1 a profile preparing part draws a plurality of profile lines PL1-PLn in parallel at a certain spacing, and round the rotational center line CP situated approx. in the center thereof a rotation is generated on the emphasized image I1 at certain angular intervals. At each angle of rotating, profiles of (n) different types are drawn for profile lines PL1-PLn and supplied to a feature point extracting part. This part, extracts the feature point from these profiles and draws the feature images I21-I26 and an overlap processing part overlaps these feature images to prepare one sheet of line image I3. This permits complementing the feature points with one another, extracting whereof has not been practicable according to the conventional technique which uses one sheet of feature image.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a grayscale image having different densities according to changes in the radiation transmission coefficient of a tissue such as a radiation transmission image obtained by irradiating a subject with radiation or a CT image obtained by a computer tomography apparatus. The present invention relates to an image processing device that extracts the edges of a desired part image.

[0002]

2. Description of the Related Art A gray-scale image (hereinafter referred to as a gray-scale image) having different densities according to changes in the radiation transmission coefficient of a tissue such as a radiation transmission image obtained by irradiating a subject with radiation or a CT image reconstructed by a computer tomography apparatus The profile method is one of the methods for obtaining the edge image of a desired image from the "original image". Note that, here, a case where a peripheral image of a blood vessel is created from an original image will be described as an example.

This profile method involves the steps of creating a density value profile as shown in FIG. 17 and the characteristic points pa and pb which are points on the edges of the blood vessel image from the profile.
And a step of obtaining a marginal image including only pixels of the original image corresponding to the obtained feature points.

In the profile creating step, for a certain pixel row on the original image, a change curve of the density value of each pixel in the pixel row, that is, a density value profile is created. In addition,
The line that defines the pixel row is referred to as a profile line. FIG. 15 shows an original image I of a blood vessel.
0 is a diagram showing a region of interest (region for creating a peripheral image) ROI and profile lines PLX, PLY. The region of interest ROI is a region for which a peripheral image is desired to be obtained, and the profile is created within the range of the region.
The profile line PLX is translated in the scanning line direction (generally horizontal direction) of the original image I0 within the region of interest ROI, and the movement causes a horizontal profile within the region of interest ROI. You can create multiple. Further, the profile line PLY is parallel to the direction perpendicular to the scanning line direction of the original image I0 within the region of interest ROI, and the movement causes a plurality of vertical profiles within the region of interest ROI. Can be created.

In the feature point extraction step, feature points pa and pb that can be regarded as points on the edge of the blood vessel image are extracted from a large number of horizontal and vertical profiles. As a method of determining whether or not a feature point is included, the gradient value (differential value) of each pixel point of the profile, that is, the slope is calculated, and it is determined whether or not the gradient value is included in the specific range, In that case, a method of using the point as a feature point, or creating a gradient value profile and differentiating again for each point of the profile (second derivative)
This is a method of determining a feature point by using the obtained second derivative value. These methods are methods devised by paying attention to the fact that the density value greatly changes at the border of the blood vessel.

In the edge image creating step, using the obtained many feature points, the feature point image composed of the feature points is directly replaced with the density value of the corresponding pixel of the original image I0 to create the edge image. To do. The edge image of the blood vessel thus obtained is
It is used for the diagnosis of stenosis of the blood vessel, the diagnosis of dilatation, the measurement of the area or volume, and the like.

Here, as the main factors that determine the reliability of the edge image such as whether or not the edge image accurately represents the blood vessel edge, the amount of information constituting one profile,
It is a specific value range used for feature point extraction used in the feature point extraction stage. In addition, the number of blood vessel pixels on one profile line is a factor that determines how small the diameter of a blood vessel can be extracted for a minute portion or a blood vessel. FIG. 16 is a diagram showing the number of pixels. In addition, in this figure, it is assumed that the thick line inner region represents the blood vessel image B having a small diameter, and each portion partitioned in a grid pattern is a single pixel. As shown in the figure, the number of blood vessel pixels on one profile line PLX in the horizontal direction is 3 pixels (hatched portion), and it cannot be said that the amount of information is sufficient for feature point extraction. If a single pixel size is miniaturized, the edge of a blood vessel having a minute diameter can be extracted according to the degree of miniaturization, but this miniaturization has a limit and is not realistic.

[0008]

SUMMARY OF THE INVENTION An object of the present invention is to provide an image processing apparatus capable of extracting a minute portion of a grayscale image, for example, a margin of a minute blood vessel.

[0009]

An image processing apparatus according to the present invention is an image processing apparatus for extracting a feature point having a desired feature from a grayscale image and obtaining a feature image composed of the feature point,

A plurality of curves, which are provided so as to intersect with the scanning lines of the grayscale image, and which show a density change between a plurality of pixels on a reference axis, are created while moving the reference axis on the grayscale image. And a means for extracting the characteristic point by identifying a curve shape representing the characteristic from the density change curve.

[0011]

According to the present invention, a curve showing a density change between a plurality of pixels existing on a reference axis provided so as to intersect a scanning line of a grayscale image is moved on the grayscale image. While creating a plurality, by extracting the characteristic points by identifying the curve shape representing the characteristics from the density change curve, it is possible to increase the number of pixels on the reference axis, as a result, If it is a minute portion or a blood vessel, the edge can be extracted even if it is a very minute blood vessel.

[0012]

Embodiments will be described below with reference to the drawings. The images used in this embodiment are X-ray diagnostic apparatus, X
An image (hereinafter referred to as "original image I0") in which the difference in the radiation transmission coefficient (attenuation coefficient) of the tissue obtained by a line computer tomography apparatus or the like is expressed as density information is used. In addition, the present embodiment is a device that can reliably obtain a marginal image even if it is a minute portion that is difficult to obtain a marginal image by the conventional method. It will be described below. The term "fine blood vessel" as used herein means a blood vessel having a diameter that is not easily captured by the pixel size of the original image I0.

FIG. 1 is a block diagram showing the arrangement of an image processing apparatus according to the first embodiment of the present invention. Pre-processing unit 1
Is a region containing a fine blood vessel image (hereinafter referred to as “region of interest RO”) to which the original image I0 is input and a marginal image including fine blood vessels is to be obtained.
(Hereinafter referred to as “I”), and performs processing for noise reduction on the image of the region of interest ROI, for example, smoothing processing, low-pass filtering processing, or processing for blood vessel image enhancement, for example, nonlinear or linear feature filtering processing. It is a processing unit to perform. The pre-processing unit 1 is not an essential component and can be appropriately removed. This pre-processing unit 1
The processed image of the region of interest ROI obtained in step 1 will be referred to as "enhanced image I1" below. Also, the region of interest ROI
Is a region in which the edge image is desired to be obtained, and the profile is created within the range of the region.

The profile creating section 2 uses the emphasized image I
Create a concentration profile using 1. A line that serves as a reference in creating this density profile is referred to as a profile line (PL). The method of setting the profile line is one of the features of this embodiment. FIG. 2 is a diagram showing a profile line group G used in the head office, and FIG. 3 is a diagram showing this profile line group G.
It is a figure which shows the mode of rotation movement of. This profile line group G includes a plurality of profile lines PL1 to PLn.
Are provided in parallel at a predetermined interval, for example, an interval of one side length of the pixels of the original image I0, and the emphasized image I1 is centered on the rotation center point CP provided at the substantially center thereof.
It rotates on the (original image I0) at every predetermined angle θ. Here, the predetermined angle θ is 30 degrees as shown in FIG.
Make a turn. Note that G1 to G6 shown in FIG. 3 are shown in the long side direction of the profile line group G corresponding to the rotation.

Here, the principle of improving the reliability of edge extraction along with the rotation of the profile line group G will be briefly described. FIG. 7 shows FIG. 16 cited in the description of the prior art.
The diameters of the blood vessel images B shown in both figures are the same. As shown in FIG. 7, the number of pixels existing on the profile line PL in the blood vessel image B (hatched portion; hereinafter referred to as “blood vessel pixel”) is 5 here, and the number is larger than in the conventional case. Is becoming This is because the profile line PL is set in a direction that is not perpendicular to the traveling direction of the blood vessel image B (hereinafter referred to as “blood vessel direction”), that is, obliquely. When this situation is compared between the profile shown in FIG. 8 and the conventional horizontal profile shown in FIG. 17, the distance (the number of pixels) between the overlapping end points pa and pb of the profile line PL and the blood vessel image B is the profile. When the line PL is set obliquely as shown in FIG. 7, it becomes longer. That is, the amount of profile information is increased, and the features of the edges are more likely to be surfaced. However, blood vessels run in all directions, and it is difficult to set the profile line PL obliquely with respect to the blood vessel direction. Therefore, in this embodiment, the profile lines G are rotated to obtain profiles in various directions.

That is, in the profile creating section 2, each time the predetermined angle θ is rotated, the profile line PL1
Create n types of profiles for each PLn. That is, 6
The total number of profiles obtained by turning is n × 6 types. 4 to 6 are views conceptually showing the rotation of the profile line group G by taking the group directions G1, G2 and G4 of FIG. 3 as an example. In the group direction G1, as shown in FIG. 4, the profile line group G is set for the region of interest ROI (emphasized image I1) designated on the original image I0, and each profile line PL1 to PLn in that direction is set. N
Seed profiles are created, and the n kinds of profiles with respect to the group direction G1 are supplied to the feature point extraction unit 3 described later. The profile line group G has a group direction G2.
Similarly, the n kinds of profiles obtained with respect to the group direction G2 are supplied to the feature point extraction unit 3. Furthermore, the group directions G3, G4 (FIG. 6), G5, G6 are sequentially rotated, and n types of profiles are obtained for each rotation.
Supply.

The feature point extraction section 3 is a profile creation section 2
The characteristic points are extracted from the plurality of profiles supplied from the above-mentioned, and characteristic images I21 to I26 are created. The characteristic images I21 to I26 are created every time the profile line group G described above is rotated, and since six rotations are repeated here, six characteristic images I21 are generated for each rotation. ~ I26 is obtained. This feature point is here a point on the edge of the blood vessel because the purpose of the present embodiment is to obtain the edge image, but it may be variously changed according to the change of the purpose. Good. The determination method for extracting the feature points based on the profile is, as described in the description of the related art, a method using the gradient value (differential value) of each pixel point of the profile,
The method based on the second derivative is adopted. The characteristic image I21
I26 are images showing only the presence or absence of the extracted feature points. This is to reduce the amount of handled data in order to improve the data processing efficiency in the processing unit provided behind the feature point extracting unit 3.

As shown in FIG. 9, the superposition processing section 4 superimposes the characteristic images I21 to I26 obtained by the characteristic point extracting section 3,
Create a line image I3. By this superposition, feature points that could not be extracted with one feature image can be complemented with each other, and the reliability of feature point extraction can be improved.

The edge image creating section 5 creates a edge image I4 using the line image I3 and the original image I0 obtained in the overlapping processing section 4. That is, the line image I3 showing only the presence or absence of the blood vessel edge is converted into the edge image I4 showing the shading of the edge line by using the density value of the corresponding pixel of the original image I0.

As described above, with the apparatus of the present embodiment, it is possible to reliably obtain the edge image I4 of a fine blood vessel, that is, to faithfully reproduce the edge of a fine blood vessel. it can. By using the highly reliable marginal image of the microvessel by the apparatus of this embodiment, the degree of stenosis of the blood vessel can be accurately measured. Further, when obtaining a three-dimensional image showing a running state of a minute blood vessel, X-ray irradiation is performed from multiple directions, a marginal image is created for each of a plurality of obtained original images, and a marginal image is created on each marginal image. It is possible to accurately obtain a mutual relationship such as a positional relationship and a thickness relationship of the same microvessel, and by using these relationships, the reconstruction processing of the three-dimensional image can be significantly reduced.

Next, a second embodiment will be described. The apparatus of the present embodiment is the same as the apparatus of the first embodiment except that the profile creation procedure in the profile creation unit 2 is different. Therefore, only the different procedure will be described.
Descriptions of other same parts are omitted. FIG. 10 is a diagram showing a designated line L designated by an operator before profile creation, and FIG. 11 is a diagram showing a profile line group G ′ used in this embodiment and its movement.

The specified line L is linearly specified along with its length substantially parallel to the running direction of the blood vessel to be edge-extracted and substantially above the center line of the blood vessel by an operator's operation. The designated line L is for identifying the moving direction and range of the profile line group G ′ described later and guiding the movement.

Profile line group G used in this embodiment
As shown in FIG. 11, ′ has six profile lines PL1 to PL6, which are radially separated by an angle θ similar to the turning angle of the profile line group G of the first embodiment. It is provided. The lengths of the profile lines PL1 to PL6 can be adjusted arbitrarily. By appropriately changing the lengths of the profile lines PL1 to PL6 and the designated line L,
The range of the edge image can be changed appropriately.

The profile line group G'moves intermittently and linearly on the designated line L from the point P1 to the point Pn at a movement interval of one pixel or several pixels. At the time of this movement, the profile line group G ′ is
It is assumed that one of the profile lines PL1 to PL6, for example, the profile line PL4 and the designated line L are moved while maintaining the parallel state. As a result, it is possible to prevent the reliability of the obtained edge image from changing according to the change in the blood vessel direction, and to always ensure the same degree of reliability. This is because the inclination angle of each line with respect to the blood vessel direction becomes uniform.

At each moving interval, the profile line P
Create a profile for each of L1 to PL6. As a result, if the line spacing of the profile line group G of the first embodiment is the same as the moving spacing, the same number of profiles as in the first embodiment can be obtained. The many profiles thus obtained are supplied to the feature point extraction unit 3 as in the first embodiment, and are passed through the overlap processing unit 4 and the edge image creation unit 5 to obtain the edge image.

According to this embodiment, the same effect as that of the first embodiment can be obtained, and the uniformity of reliability of the obtained edge image can be improved. In this embodiment, the designated line L for guiding the movement of the profile line group G ′ is designated, but the guiding line of the profile line group G ′ is replaced with the designated line L shown in FIG. , The blood vessel core line (one-dot chain line) shown in FIG. 12 obtained by differentiating / thresholding the original image
It may be R. In this case, the movement range is specified by designating the region of interest ROI as in the case of the first embodiment. As a result, the reliability of the edge image can be made more uniform than in the case of the second embodiment in which the linear designated line L is designated.

Next, a third embodiment will be described. The apparatus of the present embodiment is the same as the apparatus of the second embodiment except that the profile line group G ′ used in the profile creating unit 2 is different. Therefore, only the different procedure will be described, and other same portions will be described. Is omitted.

FIG. 13 is a diagram showing a profile line group G ″ used in this embodiment. This profile line group G ″ has a plurality of profiles that maintain an inclination of a predetermined angle θ similar to the angles used in the first and second embodiments with respect to the specified line L that is specified as in the second embodiment. It has lines PL1 to PLm. The profile line group G ″ is the designated line L as in the second embodiment.
Create numerous profiles as you move up. The angles θ of the profile lines PL1 to PLm with respect to the designated line L are set independently, and when the designated line L is replaced with a blood vessel core line as shown in FIG. 12, it is as shown in FIG. Further, the parallel state of the profile lines PL1 to PLm is released.

As described above, according to the present embodiment, the angle θ with respect to the blood vessel direction, which is effective for increasing the profile information amount (the number of blood vessel pixels) in the profile extraction, is created.
By limiting the profile to the profile No. 1, the reliability of the edge image can be improved, and the total number of profiles created can be significantly reduced.

The present invention is not limited to the above embodiments, but can be modified in various ways. For example, first, second
Alternatively, the angle θ used in the third embodiment may be changed appropriately under the instruction of the operator. In the second embodiment, the number of profile lines in the profile line group G ′ naturally changes according to the change in the angle θ.

[0031]

As described above, according to the present invention, a curve showing a density change between a plurality of pixels existing on a reference axis provided so as to intersect with a scanning line of a grayscale image is set to the reference axis. Create multiple while moving on the gray image,
By extracting the characteristic points by identifying the curve shape representing the characteristic from the density change curve,
It is possible to increase the number of pixels existing on the reference axis, and as a result, it is possible to extract the edge of a very small blood vessel, even if it is a minute portion, and thus the reliability of the edge image. An image processing device that can be improved can be provided.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an image processing apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing a profile line group used in the profile creating unit shown in FIG.

FIG. 3 is a diagram showing the rotation of a profile line group in the profile creation unit shown in FIG.

FIG. 4 is a diagram showing an initial position of a profile line group in the profile creating unit shown in FIG.

FIG. 5 is a diagram showing the rotation of the profile line group from the initial position shown in FIG. 3 to the next position.

FIG. 6 is a diagram showing a position when the profile line group shown in FIG. 4 is further rotated and its direction becomes a horizontal direction.

FIG. 7 is a diagram to be compared with FIG. 16 and is a diagram for explaining the improvement in accuracy of feature point extraction with the increase in the amount of information according to the present embodiment.

8 is a diagram to be compared with FIG. 17 and is a diagram showing the increase in the information amount shown in FIG. 7 on a profile.

9 is a diagram showing a superimposing process in a superimposing unit of a plurality of feature point images obtained by the feature point extracting unit shown in FIG.

FIG. 10 is a diagram showing designated lines for guiding the movement of the profile line group used in the second embodiment.

FIG. 11 is a diagram showing a profile line group used in the second embodiment and its movement.

12 is a diagram showing a blood vessel core line used in place of the designated line shown in FIG.

FIG. 13 is a diagram showing a profile line group and its movement according to a third embodiment.

14 is a diagram showing the independence of each profile line of the profile line group shown in FIG. 13 in the curved portion.

FIG. 15 is a diagram showing a conventional profile line.

FIG. 16 is a diagram showing the amount of information for extracting feature points obtained by a conventional profile.

17 is a diagram showing a profile based on the profile line shown in FIG.

[Explanation of symbols]

1 ... Pre-processing unit, 2 ... Profile creation unit, 3 ... Feature point extraction unit, 4 ... Overlap processing unit, 5 ... Edge image creation unit.

Claims (1)

[Claims] 1. An image processing apparatus for extracting a feature point having a desired feature from a grayscale image to obtain a feature image consisting of the feature point, wherein a feature line is provided on a reference line provided so as to intersect a scanning line of the grayscale image. A means for creating a plurality of curves showing the density change of existing plural pixels while moving the reference axis on the grayscale image, and by distinguishing the curve shape representing the characteristic from the density change curve, An image processing apparatus, comprising: means for extracting a feature point.