CN104766364A - Separation method for attribute similar structure in lower-dimension transfer function space - Google Patents
Separation method for attribute similar structure in lower-dimension transfer function space Download PDFInfo
- Publication number
- CN104766364A CN104766364A CN201510075037.5A CN201510075037A CN104766364A CN 104766364 A CN104766364 A CN 104766364A CN 201510075037 A CN201510075037 A CN 201510075037A CN 104766364 A CN104766364 A CN 104766364A
- Authority
- CN
- China
- Prior art keywords
- edge
- module
- separation method
- data
- pve
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The invention discloses a separation method for an attribute similar structure in lower-dimension transfer function space in the technical field of visualization. Through selecting a region of interest in the lower-dimension transfer function space, pre-separation is performed on attribute similar edges according to space connectivity, and separation is performed on mis-connection edges combined with set operation. The separation method for the attribute similar structure in the lower-dimension transfer function space solves the edge overlapping problem in lower-dimension transfer function design, the problem that area overlapping of the attribute similar edges in the lower-dimension transfer function space is solved, and meanwhile the complexity of a high-dimension transfer function design is avoided. The separation method for the attribute similar structure in the lower-dimension transfer function space is suitable for visualization of edges in complex and actual three-dimensional data, in particular to visualization of edges of different anatomical structures in medical three-dimensional data.
Description
Technical field
The present invention relates to a kind of technology of volume visualization field, particularly relate to the separation method at attribute similarity edge in a kind of low-dimensional transport function space.
Background technology
Volume visualization, as an important three-dimensional data visualization technique, can explore the shape of different structure in three-dimensional data, size and spatial relation.In order to the different structure in correct visualization of 3 d data, a suitable transport function is necessary; Transport function is used for distributing different colors and transparency to different structures, and which structure is transparency determine is visible, and color is used for distinguishing different structures.
Transport function design is extensively studied at present and applies, especially in medical 3 D is visual.But being still faced with a challenge, is exactly that how to design a suitable transport function be actual three-dimensional data.Because structures different in actual three-dimensional data may have similar attribute, or these structures are closely adjacent on three dimensions, easily occur to connect by mistake, and these reasons all result in difficulty and the complexity of Designing Transfer Function.Current most of transport function design is all be separated different structures by the attribute region different at transport function spatial choice.Each transport function space is made up of specific attribute, such as: gray scale, gradient, second derivative, texture, curvature, size etc.According to the difference of transport function space dimensionality, transport function can be divided into low-dimensional transport function and higher-dimension transport function.
The object of low-dimensional transport function by utilizing little attribute to reach classification, popular low-dimensional transport function mainly comprises one-dimensional transfer function and two-dimentional transport function.One-dimensional transfer function is based on a specific attribute, to classify three-dimensional data structure mainly through gray scale.Two dimension transport function is based on two attributes, such as: shade of gray two dimension transport function, and the mean variance two dimension transport function etc. of the less larger two-dimentional transport function of edge variation and regional area.They are all by selecting specific region in transport function space to be separated different edges.The advantage of low-dimensional transport function to provide mutual friendly interface, allows user select intuitively and to adjust feature space, be therefore widely used in volume drawing.But low-dimensional transport function is faced with imbricate problem: the different structure with same alike result has identical region in transport function space.Meanwhile; due to PVE (Partial VolumeEffects; half bulk effect) area data value mixes by other adjacent data values and data value own, the different edges of closely adjacent attribute similarity can occur to connect usually by mistake, and this makes imbricate problem more complicated.This is the maximum technical matters that the design of low-dimensional transport function faces.Higher-dimension transport function is assigned with more attribute to each voxel, and what usually can be separated better compares low-dimensional transport function.Such as: the dendrogram method of improvement, intelligent system method, machine learning method etc.But the design of higher-dimension transport function is also faced with very serious complicacy: the first, is difficult to determine to need which attribute to design a transport function; The second, higher-dimension transport function time overhead is very huge, limits the actual of it and applies; 3rd, higher-dimension transport function operates as a black box, usually can not flexible operating transport function according to actual needs; 4th, the situation that higher-dimension transport function connects in the face of attribute similarity edge by mistake has limited separating power.
Summary of the invention
For prior art above shortcomings, the present invention proposes the separation method of attribute similarity structure in a kind of low-dimensional transport function space, by at low-dimensional transport function spatial choice area-of-interest, and in three-dimensional data, spatial connectivity calculating is carried out to the data point meeting area attribute, isolate different connected set edges.Be separated connecting edge by mistake in conjunction with set operation.Finally give different colors to the edge separated and transparency shows.The present invention is applicable to the visual of edge in complicated actual three-dimensional data, especially different anatomical structures edge visual in medical 3 D data.
The invention provides the separation method of attribute similarity structure in a kind of low-dimensional transport function space, in order to solve the complicacy run in the imbricate problem that runs into and the design of higher-dimension transport function in the design of low-dimensional transport function.
The present invention specifically comprises the following steps:
Step one, input 3-D view, build gradient gray scale two dimension transport function space (SG-TF), in SG-TF space, select rectangular area U, U represents selected gray scale and gradient scope, and rectangle U right boundary represents tonal range V
s=[V
a, V
b], up-and-down boundary represents gradient scope V
g=[V
c, V
d], will the data set Φ of U range of attributes be met simultaneously
uconstitute the structural edge meeting selected properties scope,
Wherein: f (x, y, z) represents gray scale,
represent gradient.
Step 2, to all Φ
uin data point carry out spatial connectivity calculating, extract all connected set C
kalthough each connected set represents a structural edge S
i, but due to the existence in PVE region, two spatially closely adjacent edges can occur to connect by mistake, and namely connected set is two unions that there is the structural edge by mistake connected.
Described spatial connectivity calculates and refers to: connective expression is in three-dimensional regular data grids, and whether data point is adjacent is close to, and adjacent is be communicated with by just illustrating.Travel through all Φ
uin data point, determine which data point is spatially adjacent according to contiguous range, extract all spatially adjacent connected set.
Step 3, is separated two mistakes comprising PVE region by etching operation and connects edge, obtain the structural edge of side not containing PVE region by expansive working and phase reducing; Then the structural edge of opposite side not containing PVE region can be obtained in opposite side reexpansion operation and phase reducing.
Described etching operation refers to: by control rectangle edges of regions, removes the data point in PVE region, reaches the object being separated and connecting edge by mistake, is separated the structural edge of formation two corrosion.
Described expansive working refers to: carry out contiguous range expansion to the corrosion structure edge after corrosion and obtain expansion edge, to reach the object comprising PVE number of regions strong point.
Described phase reducing refers to: because PVE region is contained in expansion edge, therefore connect edge from two by mistake deduct one of them expansion edge, then carry out connective calculate removes fragment and obtaining maximum connected set wherein, i.e. one of them the structural edge not containing PVE region.
Step 4, repeats above-mentioned steps three, until Φ
uin all edge S of comprising
iall separated, finally to different edge S
igive different colors and transparency, realize the visual of the structural edge of image.
The present invention relates to a kind of system realizing said method, comprise: edge collecting module, spatial connectivity computing module, set operation module, wherein: edge collecting module is connected with spatial connectivity computing module and transmits marginal point, spatial connectivity computing module is connected with set operation module and transmits connected set corresponding to edge; Described set operation module comprises corrosion treatment module, expansion process module and subtract each other processing module, wherein: corrosion treatment module is connected to transmit with expansion process module and corrodes marginal point, expansion process module with subtract each other processing module and be connected and transmit expansion marginal point.
Technique effect
Compared with prior art, the invention solves imbricate problem in the design of low-dimensional transport function, and solve the region overlap problem at attribute similarity edge in low-dimensional transport function space, avoid the complicacy of higher-dimension transport function design simultaneously.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of the inventive method.
Fig. 2 is spatial connectivity computing method process flow diagram in the inventive method.
Fig. 3 is present system structural representation.
Fig. 4 is in embodiment of the present invention edge separation result schematic diagram 1, figure: (a) is original graph; B () is process schematic diagram of the present invention; C () is prior art comparison diagram.
Fig. 5 is in embodiment of the present invention edge separation result schematic diagram 2, figure: (a) is original graph; B () is process schematic diagram of the present invention; C () is prior art comparison diagram.
Fig. 6 is in embodiment of the present invention edge separation result schematic diagram 3, figure: (a) is original graph; B () is process schematic diagram of the present invention; C () is prior art comparison diagram.
Embodiment
In low-dimensional transport function space provided by the invention, the separation method of attribute similarity structure is applicable to multiple fields, such as medical threedimensional images is visual, industry 3-D view is visual, and following examples are described in detail for medical image, but are not limited only to medical image.
Embodiment 1
As shown in Figure 1, the present embodiment comprises the following steps:
Step one, first at SG-TF spatial choice rectangular area U, this rectangular area U defines the data point scope that will gather, and rectangle left and right edges represents the interval V of tonal range
s=[V
a, V
b], lower edges represents the interval V of gradient scope
g=[V
c, V
d], only have the shade of gray of data point to meet this rectangular extent just collected and show, image data point set is combined into Φ here
u.
Described SG-TF space refers to first quartile region in XOY coordinate system, and horizontal ordinate represents gray-scale value, and ordinate represents Grad.In three-dimensional data, each data point (x, y, z) has a gray-scale value f (x, y, z) and Grad
grad is calculated by gray-scale value, and computing formula is:
G
x＝(f(x+1，y，z)-f(x-1，y，z))/2
G
y＝(f(x，y+1，z)-f(x，y-1，z))/2
G
z＝(f(x，y，z+1)-f(x，y，z-1))/2
Step 2, to all Φ
uin data point carry out spatial connectivity calculating, extract all connected set C
k.Each connected set represents a structural edge S
ibut, due to the existence in PVE region, two spatially closely adjacent edge S
iand S
jcan occur to connect, so connected set C by mistake
k=S
i∩ S
j, i and j represents the label at different edge respectively here, and value is integer.
Described spatial connectivity calculates and refers to: connective expression is in three-dimensional regular data grids, and whether data point is adjacent is close to, and adjacent is be communicated with by just illustrating.Travel through all Φ
uin data point, determine which data point is spatially adjacent according to contiguous range, extract all spatially adjacent connected set, concrete grammar is as Fig. 2.Territory has 6 neighborhoods, 18 fields, and three kinds, 26 fields etc. are in 3 × 3 × 3 cubes, and we can select any one as contiguous range, here with 18 neighborhood N
18for example.Below neighborhood computing formula:
N
6＝{(i，j，k)：|(x，y，z)-(i，j，k)|≤1}
N
18＝{(i，j，k)：|(x，y，z)-(i，j，k)|≤2}
N
26＝{(i，j，k)：|(x，y，z)-(i，j，k)|≤3}
(x, y, z) is current data point, and (i, j, k) is that data point in 3 × 3 × 3 cubes centered by current point is in three-dimensional mesh data.
The edge that step 3, separation occur to connect by mistake is by set operation, and removal causes the PVE region Λ by mistake connected exactly.Connect for the mistake at two edges, two edge S
1and S
2linked together by Λ by mistake, C
k=S
1∩ S
2∩ Λ.And the data value of PVE region Λ is different from the data value of two neighboring edge junctions, illustrates technically and can remove Λ region to be separated two edges.
Here set operation comprises etching operation, expansive working and phase reducing:
Step 3.1, etching operation: by removing the minimum data point of (or large) of gray-scale value and the data point of PVE region Λ from the left side (or the right), slowly shrink rectangular area U; Described contraction is preferably first carried out connectedness afterwards and is calculated, until obtain the structural edge after two corrosion
with
Step 3.2, expansive working: the structural edge after the corrosion obtain step 3.1 carries out expansion process, be grouped into corrosion edge itself by the ambient data dot-dash of each data point, specific as follows:
The present embodiment adopt expansion radius be 3, and in three-dimensional data centered by current data point 7 × 7 × 7 cube in expand.
The structural edge of the present embodiment first after selective etching
carry out expansive working and obtain expansion edge
Step 3.3, phase reducing: by the connected set C containing wrong connection
kdeduct expansion edge
obtain another edge:
here S is obtained
1to need right before
carry out connectedness to calculate, extract maximum connected set edge i.e. S
1corresponding connected set.
Here why carrying out asking for maximal connected set, is because the data centralization after subtracting each other comprises fragment (little connected set).
Step 3.4, repetition step 3.2 and step 3.3 obtain edge
Step 4, repetition above-mentioned steps 3.1 ~ 3.4, until Φ
uin all edge S of comprising
iall separated, finally to different structure edge S
igive different colors and transparency, realize the visual of the structural edge of image.
As shown in Fig. 4 a ~ Fig. 6 c, the present embodiment by spatially away from edge opened with spatially the adjacent edge occurring by mistake to be connected is all separated, compared with prior art, its advantage is:
First: solve imbricate problem in the design of low-dimensional transport function, the edge occurring by mistake to connect can be separated accurately, and existingly have technology to be undivided.
Second: maintain former marginate slickness and accuracy.
3rd: simple to operate, non-domain expert also can apply this system.
4th: efficiency is high compared to existing technology, mask work can be completed in a short period of time.
5th: the complicacy avoiding the design of higher-dimension transport function.
Claims (7)
1. the separation method of attribute similarity structure in low-dimensional transport function space, is characterized in that, comprise the following steps:
Step one, selects rectangular area U, the data set meeting U range of attributes is constituted the structural edge meeting selected properties scope simultaneously in gradient gray scale two dimension transport function space;
Step 2, carries out spatial connectivity calculating to the data point of data centralization, extracts wherein all connected set;
Step 3, is separated two mistakes comprising PVE region by etching operation and connects edge, obtain the structural edge of connected set side not containing PVE region by expansive working and phase reducing; Then the structural edge of opposite side not containing PVE region can be obtained in the reexpansion operation of connected set opposite side and phase reducing;
Step 4, repeats above-mentioned steps three, until all structural edge that data centralization comprises are all separated, finally gives different colors and transparency to different structural edge, realize the visual of the structural edge of image.
2. separation method according to claim 1, is characterized in that, described structural edge is
Wherein: f (x, y, z) represents gray scale,
represent gradient, the right boundary of rectangular area U selected represent tonal range V
s=[V
a, V
b], up-and-down boundary represents gradient scope V
g=[V
c, V
d].
3. separation method according to claim 1, it is characterized in that, described spatial connectivity calculates and refers to: connective expression is in three-dimensional regular data grids, whether data point is adjacent is close to, adjacent is be communicated with by just illustrating, ergodic data concentrates all data points, determines which data point is spatially adjacent according to contiguous range, extracts all spatially adjacent connected set.
4. separation method according to claim 1, is characterized in that, described etching operation refers to: by control rectangle edges of regions, removes the data point in PVE region, reaches the object being separated and connecting edge by mistake, is separated formation two corrosion structure edges.
5. separation method according to claim 1, is characterized in that, described expansive working refers to: carry out contiguous range expansion to corrosion structure edge and obtain expansion edge, to reach the object comprising PVE number of regions strong point.
6. separation method according to claim 1, it is characterized in that, because PVE region is contained in expansion edge, therefore connect edge from two by mistake deduct one of them expansion edge, then carry out connective calculate removes fragment and obtaining maximum connected set wherein, i.e. one of them the structural edge not containing PVE region.
7. one kind realizes the system of method described in above-mentioned arbitrary claim, it is characterized in that, comprise: edge collecting module, spatial connectivity computing module and set operation module, wherein: edge collecting module is connected with spatial connectivity computing module and transmits marginal point, spatial connectivity computing module is connected with set operation module and transmits connected set corresponding to edge;
Described set operation module comprises corrosion treatment module, expansion process module and subtract each other processing module, wherein: corrosion treatment module is connected to transmit with expansion process module and corrodes marginal point, expansion process module with subtract each other processing module and be connected and transmit expansion marginal point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510075037.5A CN104766364A (en) | 2015-02-12 | 2015-02-12 | Separation method for attribute similar structure in lower-dimension transfer function space |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510075037.5A CN104766364A (en) | 2015-02-12 | 2015-02-12 | Separation method for attribute similar structure in lower-dimension transfer function space |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104766364A true CN104766364A (en) | 2015-07-08 |
Family
ID=53648169
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510075037.5A Pending CN104766364A (en) | 2015-02-12 | 2015-02-12 | Separation method for attribute similar structure in lower-dimension transfer function space |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104766364A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120223939A1 (en) * | 2011-03-02 | 2012-09-06 | Noh Junyong | Rendering strategy for monoscopic, stereoscopic and multi-view computer generated imagery, system using the same and recording medium for the same |
| CN103514623A (en) * | 2013-09-13 | 2014-01-15 | 上海交通大学 | Volume data identification method based on watershed algorithm for two-dimensional transfer function |
| CN103679796A (en) * | 2013-11-27 | 2014-03-26 | 上海交通大学 | Two-dimensional transfer function body data identification method based on distance transformation watershed algorithm |
| CN103914880A (en) * | 2014-03-14 | 2014-07-09 | 上海交通大学 | Volume rendering method of automatic identification overlapping structure based on two-dimensional transfer function |
-
2015
- 2015-02-12 CN CN201510075037.5A patent/CN104766364A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120223939A1 (en) * | 2011-03-02 | 2012-09-06 | Noh Junyong | Rendering strategy for monoscopic, stereoscopic and multi-view computer generated imagery, system using the same and recording medium for the same |
| CN103514623A (en) * | 2013-09-13 | 2014-01-15 | 上海交通大学 | Volume data identification method based on watershed algorithm for two-dimensional transfer function |
| CN103679796A (en) * | 2013-11-27 | 2014-03-26 | 上海交通大学 | Two-dimensional transfer function body data identification method based on distance transformation watershed algorithm |
| CN103914880A (en) * | 2014-03-14 | 2014-07-09 | 上海交通大学 | Volume rendering method of automatic identification overlapping structure based on two-dimensional transfer function |
Non-Patent Citations (1)
| Title |
|---|
| 夏斌: "三维图像边界曲面的体绘制", 《中国优秀硕士学位论文全文数据库，信息科技辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102708370B (en) | Method and device for extracting multi-view angle image foreground target | |
| CN105374033B (en) | SAR image segmentation method based on ridge ripple deconvolution network and sparse classification | |
| CN103020993B (en) | Visual saliency detection method by fusing dual-channel color contrasts | |
| CN106934766A (en) | A kind of infrared image super resolution ratio reconstruction method based on rarefaction representation | |
| CN102246204A (en) | Devices and methods for processing images using scale space | |
| CN104680508A (en) | Convolutional neural network and target object detection method based on convolutional neural network | |
| CN104331873A (en) | Method for detecting road from single image | |
| CN102651132B (en) | Medical image registration method based on intersecting cortical model | |
| CN103902989B (en) | Human action video frequency identifying method based on Non-negative Matrix Factorization | |
| CN104616308A (en) | Multiscale level set image segmenting method based on kernel fuzzy clustering | |
| CN104103052A (en) | Sparse representation-based image super-resolution reconstruction method | |
| CN105335997A (en) | Complex structure point cloud processing algorithm bases on Poisson reconstruction | |
| CN103544488B (en) | A kind of face identification method and device | |
| CN111340080B (en) | High-resolution remote sensing image fusion method and system based on complementary convolution characteristics | |
| CN102663762B (en) | The dividing method of symmetrical organ in medical image | |
| CN104484886A (en) | Segmentation method and device for MR image | |
| CN103927758A (en) | Saliency detection method based on contrast ratio and minimum convex hull of angular point | |
| CN104899857A (en) | Camera calibration method and apparatus using a color-coded structure | |
| CN102867171B (en) | Label propagation and neighborhood preserving embedding-based facial expression recognition method | |
| CN105809182A (en) | Image classification method and device | |
| CN107742302A (en) | Profile testing method based on the multiple dimensioned profile fusion of primary visual cortex | |
| CN103413332A (en) | Image segmentation method based on two-channel texture segmentation active contour model | |
| CN105512663A (en) | Significance detection method based on global and local contrast | |
| CN104778666A (en) | Image segmentation method based on spectral clustering and mixed model clustering | |
| CN104766364A (en) | Separation method for attribute similar structure in lower-dimension transfer function space |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150708 |
|
| RJ01 | Rejection of invention patent application after publication |