CN110276832B - Interactive industrial CT scanning method based on KCF and GrabCT - Google Patents
Interactive industrial CT scanning method based on KCF and GrabCT Download PDFInfo
- Publication number
- CN110276832B CN110276832B CN201910645893.8A CN201910645893A CN110276832B CN 110276832 B CN110276832 B CN 110276832B CN 201910645893 A CN201910645893 A CN 201910645893A CN 110276832 B CN110276832 B CN 110276832B
- Authority
- CN
- China
- Prior art keywords
- calibration
- projection
- images
- image
- tracking
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/277—Analysis of motion involving stochastic approaches, e.g. using Kalman filters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Geometry (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Computer Graphics (AREA)
- Software Systems (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
The invention provides an interactive industrial CT scanning method based on KCF and GrabCT, which utilizes KCF target tracking and GrabCT image segmentation to improve the calibration mode and the scanning method of the existing industrial cone beam CT, and simultaneously adds a human-computer interactive correction method for the condition that the coordinate of the center of a small sphere projection circle is deviated due to the fact that an object which is complex in structure and not easy to penetrate shields the calibration small sphere, and the foreground and the background are marked by manpower for the appointed projection, so that the coordinate of the center of the deviated small sphere projection circle is corrected, the condition that rescanning is needed due to the deviation of the coordinate of the center of the small sphere projection circle is avoided, and the existing scanning process and software are compatible; the invention synchronously completes the scanning and calibration of the object, reduces the scanning time and improves the scanning efficiency.
Description
Technical Field
The invention relates to an industrial CT scanning method, in particular to an interactive industrial CT scanning method based on KCF and GrabCT.
Background
The industrial cone beam CT nondestructive detection is a nondestructive detection technology which can be applied to the extensive fields of minerals, metallurgy, microelectronics, composite materials, biology, military and the like. Since the first CT in the 20 th century and the 70 th era was born, the performance of industrial CT has been improved year by year with the progress of computer science and the development of devices such as detectors and ray sources. As a practical industrial nondestructive testing means, the current industrial cone-beam CT with the flat panel detector component has the remarkable advantages of high control resolution, short data acquisition time, high ray utilization efficiency and the like.
In the industrial cone-beam CT nondestructive testing process, the scanning of the object and the calibration of the object are two crucial steps, wherein the scanning of the object is used to obtain the projection data of the object, and the calibration of the object is used to obtain the position information of the object in the industrial cone-beam CT system, and the results obtained by the two steps usually lay the foundation for obtaining the three-dimensional reconstruction of the scanned object. However, due to the limitations of the calibration algorithm, the systematic correction, the target extraction under a complex background, and the like, the two steps of scanning the article and calibrating the article in the existing industrial cone-beam CT nondestructive testing process are performed separately, so that the testing time of the nondestructive testing process of the article is longer, and the testing efficiency is reduced.
In order to solve the above problems, people always seek an ideal technical solution.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, and provides an interactive industrial CT scanning method based on KCF and GrabCT, which has the advantages of short scanning time, high detection effect and good compatibility.
In order to achieve the purpose, the invention adopts the technical scheme that: an interactive industrial CT scanning method based on KCF and GrabCT, which comprises the following steps:
step 1, simultaneously fixing an object and a double-sphere calibration rod on a bearing plate, wherein the object and the double-sphere calibration rod cannot have a CT (computed tomography) view field under the condition of rotating for a circle, and acquiring a projection image set of the object and the double-sphere calibration rod through an industrial CT acquisition system;
step 2, selecting two calibration pellets in the first frame of projection image, tracking the two calibration pellets by utilizing a KCF target tracking algorithm, and obtaining projection tracking images of the two calibration pellets in all the frame projection images of the projection image set;
step 3, rapidly dividing two calibration pellets in all frame projection tracking images by using a GrabCont image division algorithm to obtain binary projection images of the calibration pellets, and solving the projection center coordinates of the calibration pellets in each frame of binary projection images based on circle detection of Hough transformation;
step 4, carrying out ellipse fitting on the projection circle center coordinates of the calibration pellet in all the obtained frame binary projection images based on a least square ellipse fitting algorithm, automatically detecting whether the projection circle center of the calibration pellet deviates, and if detecting that the projection circle center of the calibration pellet deviates, executing step 5; if not, skipping to execute the step 6;
step 5, determining projection images and projection tracking images corresponding to the projection centers of all the deviated calibration beads one by one, and correcting the projection center coordinates of the projection tracking images in a man-machine interaction mode for all the determined projection tracking images;
step 6, generating a calibration parameter file by utilizing calibration parameter generation software according to the circle center coordinate parameters of the calibration small ball projection;
step 7, reconstructing a three-dimensional reconstruction model containing the calibration small balls by using the calibration parameter file and the three-dimensional reconstruction software;
and 8, cutting the three-dimensional reconstruction model containing the calibration ball by using three-dimensional visualization software to finally obtain the three-dimensional model of the scanned object.
Based on the above, the specific steps of correcting the projection circle center coordinates of the projection tracking chart in the step 5 in a man-machine interaction mode are as follows:
marking a determined foreground image area and a determined background image area in each frame of projection tracking image;
according to the determined foreground image area and the determined background image area marked in each frame of projection tracking image, rapidly segmenting two calibration beads in each frame of projection tracking image by using a GrabCT image segmentation algorithm to obtain binary projection images of the calibration beads;
and (4) calculating the projection center coordinates of the calibration ball in each frame of binary projection image based on circle detection of Hough transformation.
Based on the above, in step 1, the ball distance of two calibration balls in the double-ball calibration rod is set according to the size of the article: when the article is small, the distance between the two calibration small balls is slightly larger than the height of the article; when the article is large, the ball distance of the two calibration small balls is set randomly.
Based on the above, in the step 3, in the process of rapidly and automatically segmenting the small balls in the tracking image by using the GrabCut image segmentation algorithm, the projection tracking images of the two calibrated small balls are uniformly expanded by one time, the expansion images of the projection tracking images of the two calibrated small balls are selected as the source images to be segmented, and the projection tracking images of the two calibrated small balls are selected as the possible foreground image areas.
Based on the above, the distance between the small balls in the double-ball calibration rod can be selected according to the actual situation: when the object is small, the double-ball calibration rod with the ball distance slightly larger than the height of the object is selected, so that the object does not shield the calibration small ball; when the article is large, no additional requirement is placed on the double-ball calibration rod.
Compared with the prior art, the invention has outstanding substantive characteristics and remarkable progress, and particularly provides an interactive industrial CT scanning method based on KCF and GrabCT, which utilizes a KCF target tracking algorithm and a GrabCT image segmentation algorithm to improve the existing calibration mode and scanning method of the industrial cone beam CT, and simultaneously adds a man-machine interactive correction method for the condition that the coordinate of the center of a small sphere projection circle is deviated due to the fact that an object which is complex in structure and is not easy to penetrate shields the calibration small sphere, corrects the coordinate of the center of the deviated small sphere projection circle by manually marking the foreground and the background of a corresponding projection tracking image, and avoids the condition that rescanning is needed due to the deviation of the coordinate of the center of the small sphere projection circle; the invention is compatible with the existing scanning process and software, improves the scanning method of the industrial cone beam CT, synchronously completes the scanning and calibration of the object, reduces the scanning time and improves the scanning efficiency.
Drawings
FIG. 1 is a schematic flow diagram of the present invention.
FIG. 2 is a schematic flow chart of the present invention for correcting the coordinates of the center of the projection circle in a man-machine interaction manner in step 5.
Detailed Description
The technical solution of the present invention is further described in detail by the following embodiments.
For the convenience of the examiner, technical terms and terminology appearing in the present embodiment are explained:
GrabCT image segmentation algorithm
Function prototype: void cv:: gradcut (const Mat & img, mat & mask, rect Rect, mat & bgdModel, mat & fgdModel, int iterCount, int mode)
Description of the function parameters:
img: the source image to be segmented must be an 8-bit 3-channel (CV _8UC 3) image, and cannot be modified in the processing process;
mask: mask image, if using mask to initialize, mask saves the initialized mask information; when the segmentation is executed, the foreground and the background set by the user interaction can be stored in a mask, and then a grabCut function is input; after the process ends, the results are saved in the mask. mask can take only the following four values:
GCD _ BGD (= 0), determine a background image region;
GCD _ FGD (= 1), determine foreground image region;
GCD _ PR _ BGD (= 2), possible background image area;
GCD _ PR _ FGD (= 3), possible foreground image regions.
If GCD _ BGD or GCD _ FGD is not manually marked, then the result will be GCD _ PR _ BGD or GCD _ PR _ FGD only;
rect: for defining the image range to be segmented, only the image part in the rectangular window is processed;
bgdModel: if the background model is null, automatically creating a bgdModel in the function; the bgdModel must be a single-channel floating-point (CV _32FC 1) image, and the number of rows can only be 1, and the number of columns can only be 13x5;
fgdModel: if the foreground model is null, automatically creating an fgdModel inside the function; the fgdModel must be a single-channel floating-point (CV _32FC 1) image with only 1 row and 13x5 columns;
iterCount: the iteration number must be greater than 0;
mode: for indicating what the grabCut function does, optional values are:
GC _ INIT _ WITH _ RECT (= 0), initialize GrabCut WITH a rectangular window;
GC _ INIT _ WITH _ MASK (= 1), initialize GrabCut WITH a MASK image;
GC _ EVAL (= 2), segmentation is performed.
As shown in FIG. 1, the present invention provides an interactive industrial CT scanning method based on KCF and GrabCT, which comprises the following steps:
step 1, because the foam made of polystyrene has the characteristics of light weight, easy penetration of X-rays and firm structure and difficult deformation, in the embodiment, the foam made of polystyrene is preferably selected as the bearing plate; tying an article on the bearing plate, and fixing the double-ball calibration rod on the bearing plate on one side of the article, wherein the ball distance of the double balls is determined according to actual conditions: when the article is small, the distance between the two calibration small balls is slightly larger than the height of the article; when the article is large, the ball distance of the two calibration small balls is set randomly.
And observing by an industrial CT acquisition system to ensure that the object and the double-ball calibration rod cannot go out of a CT view field under the condition of rotating for one circle, and then acquiring a projection image set of the object and the double-ball calibration rod by the industrial CT acquisition system.
And 2, manually framing the two calibration pellets in the first frame of projection image, tracking the two calibration pellets by utilizing a KCF target tracking algorithm, and obtaining projection tracking images of the two calibration pellets in all the frame projection images of the projection image set.
And 3, rapidly dividing the two calibration beads in all the frame projection tracking images by using a GrabCT image division algorithm to obtain binary projection images of the calibration beads, and solving the projection circle center coordinates of the calibration beads in each frame of the binary projection images based on circle detection of Hough transformation.
In the process of fast and automatically segmenting the small balls in the tracking image by using a GrabCont image segmentation algorithm, uniformly expanding the projection tracking images of the two calibration small balls by one time, selecting the expansion images of the projection tracking images of the two calibration small balls as source images to be segmented, and selecting the projection tracking images of the two calibration small balls as possible foreground image areas;
step 4, carrying out ellipse fitting on the projection circle center coordinates of the calibration pellet in all the obtained frame binary projection images based on a least square ellipse fitting algorithm, automatically detecting whether the projection circle center of the calibration pellet deviates, and if detecting that the projection circle center of the calibration pellet deviates, executing step 5; if not, skipping to execute the step 6;
for the two conditions that the calibration small ball is not shielded by the object and the calibration small ball is shielded by the object, the structure of the object is simple and the object is easy to penetrate, the projection circle centers of the corresponding calibration small balls are uniformly stepped in the elliptical image obtained in the step 4; for the condition that the calibration small ball is shielded by the object and the object has a complex structure and is not easy to penetrate through, the projection center of the corresponding calibration small ball deviates in the elliptical image obtained in the step 4;
step 5, determining projection images and projection tracking images corresponding to the projection centers of all the shifted calibration pellets one by one, and correcting the projection center coordinates of the projection tracking images for all the determined projection tracking images in a man-machine interaction mode; specifically, as shown in fig. 2:
marking a determined foreground image area and a determined background image area in each frame of projection tracking image;
according to the determined foreground image area and the determined background image area marked in each frame of projection tracking image, rapidly segmenting two calibration beads in each frame of projection tracking image by using a GrabCT image segmentation algorithm to obtain binary projection images of the calibration beads;
calculating the projection center coordinates of the calibration ball in each frame of binary projection image based on circle detection of Hough transformation;
step 6, generating a calibration parameter file by utilizing calibration parameter generation software according to the circle center coordinate parameters of the calibration small ball projection;
existing scaling parameter generation software may be employed to generate the scaling parameter file, and in particular, since the scaling bead is in space theta,theta + pi andthe centers of the four positions form a square, and theta andline L connecting the centers of the spheres 1 And sum of θ + πLine L connecting the centers of the two points 2 Parallel, then L 1 And L 2 Projection onto a detectorThe focal point of the double-sphere calibration rod is located on a vanishing line, different parallel lines can be obtained by enabling theta = theta + delta, a group of intersection points are obtained, then a vanishing line equation L is determined, the coordinates of the intersection points of the main light beam and the detector can be obtained finally, then a nonlinear optimization model is established based on a geometric artifact correction method of the double-sphere calibration rod, then a Gaussian-Newton iterative algorithm is used for solving, and finally a calibration parameter file is generated.
And 7, reconstructing a three-dimensional reconstruction model containing the calibration beads by using the calibration parameter file and the three-dimensional reconstruction software, wherein the three-dimensional reconstruction software can select the existing three-dimensional reconstruction software based on the FDK algorithm.
And 8, cutting the three-dimensional reconstruction model containing the calibration bead by using three-dimensional visualization software, cutting off the three-dimensional reconstruction model of the calibration bead, and finally obtaining the three-dimensional model of the scanned object.
In this embodiment, the self-developed 130kV industrial cone beam CT system is used to count the durations of the obtained object models for different objects and different collection points, and the results are shown in the following table 1:
TABLE 1 duration comparison results table
As can be seen from the above table, the scanning method in this embodiment has shorter scanning time and higher scanning efficiency than the original scanning method.
Finally, it should be noted that the above examples are only used to illustrate the technical solution of the present invention and not to limit it; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art will understand that: modifications to the specific embodiments of the invention or equivalent substitutions for parts of the technical features may be made; without departing from the spirit of the present invention, it is intended to cover all aspects of the invention as defined by the appended claims.
Claims (4)
1. An interactive industrial CT scanning method based on KCF and GrabCT is characterized by comprising the following steps:
step 1, simultaneously fixing an object and a double-sphere calibration rod on a bearing plate, and collecting a projection image set of the object and the double-sphere calibration rod through an industrial CT collection system, wherein the object and the double-sphere calibration rod do not have a CT view field under the condition of rotating for a circle;
step 2, selecting two calibration pellets in the first frame of projection image, tracking the two calibration pellets by utilizing a KCF target tracking algorithm, and obtaining projection tracking images of the two calibration pellets in all the frame projection images of the projection image set;
step 3, rapidly dividing two calibration pellets in all frame projection tracking images by using a GrabCont image division algorithm to obtain binary projection images of the calibration pellets, and solving the projection center coordinates of the calibration pellets in each frame of binary projection images based on circle detection of Hough transformation;
step 4, carrying out ellipse fitting on the projection circle center coordinates of the calibration pellet in all the obtained frame binary projection images based on a least square ellipse fitting algorithm, automatically detecting whether the projection circle center of the calibration pellet deviates, and if detecting that the projection circle center of the calibration pellet deviates, executing step 5; if not, skipping to execute the step 6;
step 5, determining projection images and projection tracking images corresponding to the projection centers of all the shifted calibration pellets one by one, and correcting the projection center coordinates of the projection tracking images for all the determined projection tracking images in a man-machine interaction mode;
step 6, generating a calibration parameter file by utilizing calibration parameter generation software according to the circle center coordinate parameters of the calibration small ball projection;
step 7, reconstructing a three-dimensional reconstruction model containing the calibration small balls by using the calibration parameter file and the three-dimensional reconstruction software;
and 8, cutting the three-dimensional reconstruction model containing the calibration ball by using three-dimensional visualization software to finally obtain the three-dimensional model of the article.
2. The interactive industrial CT scanning method based on KCF and GrabCT of claim 1, wherein the specific step of correcting the projection circle center coordinates of the projection tracking chart in the step 5 by a man-machine interaction mode is as follows:
marking a determined foreground image area and a determined background image area in each frame of projection tracking image;
according to the determined foreground image area and the determined background image area marked in each frame of projection tracking image, rapidly segmenting two calibration pellets in each frame of projection tracking image by using a GrabCT image segmentation algorithm to obtain binary projection images of the calibration pellets;
and (4) calculating the projection center coordinates of the calibration ball in each frame of binary projection image based on circle detection of Hough transformation.
3. The interactive industrial CT scanning method based on KCF and GrabCut according to claim 1 or 2, characterized in that: in the step 1, the ball distance of two calibration small balls in the double-ball calibration rod is set according to the size of an article: when the article is small, the distance between the two calibration small balls is slightly larger than the height of the article; when the article is large, the ball distance of the two calibration small balls is set randomly.
4. The interactive industrial CT scanning method based on KCF and GrabCut according to claim 1 or 2, characterized in that: in the step 3, in the process of rapidly and automatically segmenting the small balls in the tracking image by utilizing a GrabCT image segmentation algorithm, the projection tracking images of the two calibration small balls are uniformly expanded by one time, the expansion images of the projection tracking images of the two calibration small balls are selected as source images to be segmented, and the projection tracking images of the two calibration small balls are selected as possible foreground image areas.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910645893.8A CN110276832B (en) | 2019-07-17 | 2019-07-17 | Interactive industrial CT scanning method based on KCF and GrabCT |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910645893.8A CN110276832B (en) | 2019-07-17 | 2019-07-17 | Interactive industrial CT scanning method based on KCF and GrabCT |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110276832A CN110276832A (en) | 2019-09-24 |
CN110276832B true CN110276832B (en) | 2022-10-04 |
Family
ID=67964529
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910645893.8A Active CN110276832B (en) | 2019-07-17 | 2019-07-17 | Interactive industrial CT scanning method based on KCF and GrabCT |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110276832B (en) |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101515370B (en) * | 2009-03-06 | 2011-05-18 | 北京航空航天大学 | Calibration method of projection coordinate of ray source focus in three-dimensional microscopic CT scanning system |
WO2018126335A1 (en) * | 2017-01-03 | 2018-07-12 | 苏州海斯菲德信息科技有限公司 | Method for evaluating and correcting geometric parameters of cone-beam ct system based on glomerulus motif |
CN109632844B (en) * | 2019-01-09 | 2021-08-17 | 中国人民解放军战略支援部队信息工程大学 | Dual-energy CT imaging system and method based on linear scanning track |
-
2019
- 2019-07-17 CN CN201910645893.8A patent/CN110276832B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN110276832A (en) | 2019-09-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20220292646A1 (en) | System and method for image reconstruction | |
CN100524358C (en) | Method for removing improved conical bind CT ring shaped false shadow | |
CN106667512B (en) | Geometric correction method of X-ray imaging device and breast tomography device | |
US9858690B2 (en) | Computed tomography (CT) image reconstruction method | |
CN107796834B (en) | Orthogonal electronic linear scanning CL imaging system and method | |
US20130315454A1 (en) | Automated crystal identification achieved via watershed segmentation | |
US20200085404A1 (en) | Geometric calibration for cone beam ct using line fiducials | |
US11504085B2 (en) | Method for calibrating defective channels of a CT device | |
CN103218834B (en) | A kind of Image Reconstruction Method for ICT center positioning method based on point spread function | |
CN102488528B (en) | Correcting method for geometric parameters of tomography | |
CN105319225B (en) | A kind of scan method for realizing plaques high-resolution large-viewing open country CL imaging | |
CN111223159B (en) | Cone beam imaging method, apparatus, computer device and storage medium | |
CN101126723A (en) | Cone-beam CT beam hardening calibration method based on section profile reprojection | |
CN110268441A (en) | The method for obtaining the 3D model data of multiple components of object | |
CN110276832B (en) | Interactive industrial CT scanning method based on KCF and GrabCT | |
CN102062740B (en) | Cone-beam CT (Computed Tomography) scanning imaging method and system | |
CN109919868A (en) | A kind of detecting of cone-beam CT beam hardening curve and projection weighted correction method | |
CN105261048A (en) | Accurate positioning method for central cone-beam projection position of small sphere | |
US9575016B2 (en) | Projection method of three-dimensional imaging | |
CN111583389A (en) | Incomplete scanning CT image reconstruction method based on CAD model | |
CN113729747B (en) | Spherical metal marked cone beam CT metal artifact removal system and removal method | |
CN106419948A (en) | CT data acquisition and scanning process virtual reappearance method | |
CN113409412B (en) | Large-field-of-view CL reconstruction method, device, equipment and medium in offset scanning mode | |
Duan et al. | Knowledge-based self-calibration method of calibration phantom by and for accurate robot-based CT imaging systems | |
Guo et al. | Iterative image reconstruction for limited-angle CT using optimized initial image |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |