JP2001297334A - Three-dimensional image constituting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a three-dimensional(3D) image constituting device capable of reproducing image data with fidelity in a short time. SOLUTION: Concerning an algorithm for extracting surface pixels so as to plot the surface of medical image data with dots while using OpenGL and for calculating the inclination of the surface of a correspondent pixel, when plotting the surface of a 3D object, the patch of a polygon corresponding to the pixels is not plotted but one dot is plotted corresponding to one pixel. By plotting dots corresponding to each of pixels, a conventional gap or unnatural ruggedness does not occur.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a three-dimensional image forming apparatus, and more particularly to a three-dimensional image forming apparatus that performs reconstruction and high-speed drawing to reconstruct a three-dimensional image.

[0002]

2. Description of the Related Art Pseudo three-dimensional images (hereinafter, referred to as "three-dimensional images") have a perspective or a three-dimensional reality.
Simply called a three-dimensional image). This three-dimensional image is obtained by stacking a plurality of tomographic images (two-dimensional images) (hereinafter, referred to as volume data), and viewing the stacked image from a certain viewpoint or light source to obtain a perspective feeling such as perspective. This is an image created as a certain image. In the case where there is a gap in the stacking, a method of embedding an interpolated image obtained by interpolating the original images into the gap to increase the stacking density of the stacked image is often adopted.

The above tomographic images include X-ray CT tomographic images and M
There is an RI tomographic image.

A three-dimensional image is an image created artificially, not a stacked image itself. For example,
This is a method called surface rendering which calculates the reflection and transmission of light from the light source based on the positional relationship between the viewpoint, the light source, and the volume data (object), and projects a pixel value corresponding to the inclination of the surface of the object onto a projection plane.

In recent years, the OpenGL technology widely used in the world of 3D computer graphics has been widely used in the field of medical imaging. When the surface rendering method is performed using OpenGL, a method of displaying a three-dimensional object by simulating a polygonal plane on an object surface and setting a normal vector to the plane surface is used.

[0006]

However, in the conventional image construction method, since a polygonal plane is formed, irregularities are generated in gaps and surfaces, and artifacts are generated, so that image data can be faithfully reproduced. could not. Also, the algorithm for generating the planes constituting the surface becomes complicated, the data volume of medical tomographic images is enormous, and it takes time to draw a large number of planes.

[0007] The present invention has been made in view of such circumstances, and has as its object to provide a three-dimensional image forming apparatus capable of faithfully reproducing image data in a short time.

[0008]

According to the present invention, in order to attain the above object, volume data obtained by stacking a plurality of tomographic images is binarized based on an appropriate threshold value to indicate whether or not a target object exists. Means for constructing valued volume data, means for extracting each pixel constituting the surface of the target object based on the binarized volume data,
For each pixel constituting the surface, means for determining a normal vector indicating the inclination of the surface of the target object at the position of the pixel, and the coordinates of each pixel constituting the surface of the target object and each pixel are determined. Means for forming a pseudo three-dimensional image composed of a set of points corresponding to each pixel based on the normal vector.

According to the present invention, there is provided an algorithm for extracting a surface pixel so that a surface of medical image data can be drawn as a point using OpenGL, and calculating an inclination of the surface of the corresponding pixel. When drawing the surface of a three-dimensional object, instead of drawing a polygonal plane corresponding to the pixel, one point is drawn for one pixel. By drawing points corresponding to each pixel in this manner, gaps and unnatural irregularities unlike the related art do not occur.

A pixel in which a target object exists and a pixel in the vicinity of which a target object does not exist may be extracted as a pixel forming the surface of the target object. A vector from each pixel forming the surface of the target object to a pixel not included in one of the neighboring regions may be used as a normal vector.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of a three-dimensional image forming apparatus according to the present invention will be described below with reference to the accompanying drawings.

A description will be given of a three-dimensional object forming method to which the three-dimensional image forming apparatus of the present invention is applied. First, a plurality of tomographic image data shown in FIG. 1A are provided with appropriate thresholds for extracting skin and bone, and are binarized as shown in FIG. 1B. Pixels with data of 1 correspond to the target object. Data is centered on one pixel and its vicinity (n as in FIG. 2)
It is checked whether or not the data of the pixel (in the (× n × n) cube) is 0. If at least one of the data is 0, the pixel of interest is regarded as a pixel forming the object surface. This processing is performed for each pixel for each slice, and the surface of the object is extracted.

At this time, since the search has been completed for the previous slice that has already been processed, it is known before the search whether or not the search target pixel on the previous slice is a pixel forming the surface. When the pixel data to be searched is 1 and does not form a surface, at least n × n ×
In the range of n, all data is filled with 1 as shown in FIG. Therefore, there is no need to search this range. When the search target pixel data 1 on the lower slice is one and the pixel is not the surface of the object, the search for the target pixel may be performed only on the slice surface on the upper slice. In this way, the calculation procedure is reduced and the search range is narrowed, thereby speeding up the object surface extraction.

When checking whether data is 0 in the vicinity of the object, the normal vector of each pixel is simultaneously set as the inclination of the object surface as shown in FIG. This normal vector N is represented by (pi-p0) and is a vector directed from the pixel of interest to a pixel whose data is 0. When there are a plurality of pixels having data of 0, the sum shown in Expression 1 is obtained and normalized.

[0015]

(Equation 1) When accessing the above pixel information, it is necessary to search the tomographic data again. Since most of the tomographic data is data that is not on the surface of the object, the search is wasteful. Here, the coordinate data and the normal vector data of each pixel are secured as a list. FIG.
As shown in (1), the coordinate data list and the normal vector data list are one-dimensional lists in which data is arranged in the order in which the data is recognized as the object surface. The elements of x, y, and z are stored respectively. The numbers in the list are assigned in the order in which they are recognized as object surfaces. By listing the data, there is an advantage that the data access time in drawing can be reduced and the memory capacity used can be saved.

Next, the obtained normal vector and the surrounding normal vectors are smoothed to smooth the object surface. As shown in FIG. 6, in each pixel forming the surface, the vicinity of n × n × n is searched to determine whether there is a pixel forming the surface. If there is, the normal vector NO (x, y, z) of the target pixel and the normal vector Nj (x, y, z) of the searched pixel are averaged. If there is more than one, normalization is performed by taking the sum shown in Equation 2.

[0017]

(Equation 2) As described above, since the coordinate and normal vector data are stored in a one-dimensional list, the n × n ×
It is necessary to search for data near n from the list. Here, neighboring data is searched by a two-dimensional two-branch search method.

The data list of coordinates and normal vectors is arranged in the order in which pixels constituting the surface are recognized. Further, when searching for the pixels forming the surface, since the image is manipulated two-dimensionally, it may be considered that the coordinates are arranged in ascending order as shown in FIG. Since the recognition of each pixel is performed by scanning the tomographic plane, a result sorted in the order of x coordinate, y coordinate, and z coordinate is obtained. Knowing the data amount for each tomographic plane means that data sorted by coordinate values is selected for each tomographic plane. This data is subjected to a two-dimensional binary search. This is a method of performing a binary search in the x-axis direction after performing a binary search in the y-axis direction as shown in FIG. A binary search is performed on the y coordinate to search the list for data having the same value as the target y coordinate. x
A binary search is performed on the coordinates to search the list for data having the same value as the target x coordinate. By performing a binary search, it is possible to search at high speed without having to access all data.

A pseudo three-dimensional object is drawn using a point object of OpenGL from a normal line corresponding to the coordinates of the extracted surface pixels. Then, points corresponding to each pixel are constructed as shown in FIG. 9, and a pseudo three-dimensional object is drawn.
The projection method is an orthographic projection method.

FIG. 10 is a block diagram showing an example of a hardware configuration to which the present invention can be applied. In FIG.
1 is a CPU, 2 is a main memory, 3 is a magnetic disk, 4 is a display memory, 5 is a CRT, 6 is a controller, 7 is a mouse, 9 is a communication interface, 10 is a removable storage medium interface, 11 is various drive devices,
Reference numeral 2 denotes an image processing accelerator, which is connected to a common bus 8.

The magnetic disk 3 stores a plurality of tomographic images, a program for executing the method of the present invention, and the like. The CPU 1 reads the plurality of tomographic images and a program for executing the present invention, performs an operation such as three-dimensional image processing using the main memory 2, sends the result to the display memory 4, and displays the result on the CRT monitor 5. Let it. The mouse 7 connected to the controller 6 is a GUI (Graphical User
Interface) for rotation, translation, and scaling operations.

Reference numeral 9 denotes a communication interface such as Ethernet (registered trademark), which is used to communicate with a CT device or an MR device to acquire image data. 10 is SCSI
MOD, DAT, 8mm
Drive device 11 for tape, CD, CDR, DVD, etc.
And use a medium that is compatible with various drive devices,
Acquires CT and MR data and externally stores and stores processed data. The image processing accelerator 12 performs three-dimensional image processing at high speed. In this system configuration, it is assumed that the processing can be performed by the CPU without the image processing accelerator. The CRT 13 is an extension CRT, and a plurality of CRTs can be provided as needed.

[0023]

According to the present invention, a pseudo three-dimensional object can be displayed by points without using polygons, and pixels can be written one by one. Is drawn. Also, since the normal vectors are averaged, a smooth shape can be obtained. Drawing is faster than when drawing polygons.

[Brief description of the drawings]

FIG. 1 is a diagram showing a state of a binarization process.

FIG. 2 is a diagram showing a state in the vicinity of a pixel of interest (n × n cube).

FIG. 3 is a diagram showing a state when a search process at the time of extracting an object surface is omitted.

FIG. 4 is a diagram showing a method for calculating a normal line.

FIG. 5 is a diagram showing a state of a data list of coordinates of a pixel forming a surface and a normal vector.

FIG. 6 is a diagram showing a normal line smoothing method.

FIG. 7 is a diagram showing a state of a data list of coordinates of pixels forming a surface.

FIG. 8 is a diagram showing a state when a two-dimensional binary search is performed on a data list of coordinates of pixels forming a surface;

FIG. 9 is a diagram showing a state of a pseudo three-dimensional object constructed by point objects.

FIG. 10 is a hardware configuration diagram to which the present invention can be applied.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... CPU, 2 ... main memory, 3 ... magnetic disk, 4 ... display memory, 5 ... CRT, 6 ... controller, 7 ... mouse, 9 ... communication interface, 10 ... removable storage medium interface, 11 ... various drive devices, 1
2. Image processing accelerator

Claims (1)

[Claims] A means for binarizing volume data obtained by stacking a plurality of tomographic images on the basis of an appropriate threshold value to form binarized volume data indicating the presence / absence of a target object; Means for extracting each pixel constituting the surface of the target object based on the converted volume data, and for each pixel constituting the surface, a normal vector indicating the inclination of the surface of the target object at the position of the pixel. Means for obtaining, and means for forming a pseudo three-dimensional image consisting of a set of points corresponding to each pixel based on the coordinates of each pixel constituting the surface of the target object and the normal vector obtained for each pixel, A three-dimensional image forming apparatus, comprising: