AU2013370770A1 - Method for processing and identifying three-dimensional data - Google Patents

Method for processing and identifying three-dimensional data

Info

Publication number
AU2013370770A1
AU2013370770A1 AU2013370770A AU2013370770A AU2013370770A1 AU 2013370770 A1 AU2013370770 A1 AU 2013370770A1 AU 2013370770 A AU2013370770 A AU 2013370770A AU 2013370770 A AU2013370770 A AU 2013370770A AU 2013370770 A1 AU2013370770 A1 AU 2013370770A1
Authority
AU
Australia
Prior art keywords
point
plane
interested target
dimensional data
planes
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.)
Granted
Application number
AU2013370770A
Other versions
AU2013370770B2 (en
Inventor
Zhiqiang Chen
Mingliang Li
Yuanjing Li
Li Zhang
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Nuctech Co Ltd
Original Assignee
Tsinghua University
Nuctech Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Priority claimed from CN201210580652.8A external-priority patent/CN103903297B/en
Application filed by Tsinghua University, Nuctech Co Ltd filed Critical Tsinghua University
Publication of AU2013370770A1 publication Critical patent/AU2013370770A1/en
Application granted granted Critical
Publication of AU2013370770B2 publication Critical patent/AU2013370770B2/en
Ceased legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Description

Three-dimensional data processing and recognizing method
CROSS-REFERENCE TO RELATED APPLICATION
This application claims the benefit of Chinese Patent Application No. 201210580652.8 filed on December 27, 2012 in the State Intellectual Property Office of China, the disclosure of which is incorporated herein by reference.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to three-dimensional data processing and recognizing field, and more particularly, to intelligent recognition of prohibited articles in security inspection apparatuses by CT or to be detected articles in nuclear magnetic resonance systems.
2. Description of the Related Art
Civil aviation transportation industry has already become a primary object that terrorists make terroristic events, since it has characteristics such as big number of passengers, easy attack and large effect afterwards.
After the terrorist attacks occurred at September 11, 2001, U.S. government decides to put additional aviation safety inspections into practice, which demand to inspect explosives for all selected baggage. Known explosive detection systems obtain images of contents in packages by X-ray (radioscopy) or CT techniques. The U.S. government regulates that all the selected baggage must be inspected with respect to explosives by the techniques having been certificated by newly established Transportation Security Agency (TSA), starting from December of 2001. So far, the only one which has been certificated by TSA, is CT technology.
Currently, majority of security inspection CT machine all can obtain three-dimensional CT data. Since passenger flow volume at airport is very large, explosive detection systems for consigned and hand baggage require a high
FP150096AU throughput. There is an urgent need for a solution to intelligently recognize prohibited articles, which can reduce working intensity, decrease anthropogenic factors and increase the throughput.
In view of this, it indeed needs a new three-dimensional data processing and recognizing method used in security inspection CT machine.
SUMMARY OF THE INVENTION
In view of this, an object of the present invention is to alleviate at least one aspect of the above problems and defects in the prior arts.
According to one aspect of the present invention, it provides a new three-dimensional data processing and recognizing method, comprising the following steps of:
scanning and re-constructing objects to be detected so as to obtain three-dimensional data for recognition of the objects to be detected;
extracting data matching to features from the three-dimensional data, so that the extracted three-dimensional data constitute interested targets in order to display and recognize the object to be detected.
The present invention can make an initial recognition of the object to be detected since three-dimensional data thereof can be extracted by a method of matching features. The shapes of the object to be detected can be further recognized through analysis and calculation of the interested targets, for example based on points, lines and/or planes, thereby recognizing shapes of the prohibited articles such as cutting tools, cylinders and cuboids. Any kinds of CT system can adopt such method to recognize the dangerous articles. Other systems generating three-dimensional data, for example, nuclear magnetic resonance system or the like can also adopt such method, to recognize the interested targets. That is, the three-dimensional data are searched, extracted, merged, counted and recognized on terms of shapes, by feature data of the suspect object, for example, dangerous articles or prohibited articles. As image segmentation of the three-dimensional data is still relatively difficult up to now, and accuracy and generality thereof are
FP150096AU not so good, the present invention solves the problems of searching, making statistics and recognizing on terms of shape for the interested targets, starting from another viewpoint.
BRIEF DESCRIPTION OF THE DRAWINGS
These and other aspects as well as advantages of the present invention will become apparent and readily understood from the description of the preferred embodiments taking in conjunction with the accompanying drawings, in which:
Figure 1 is a schematic view of eliminating some data points absent in a surface;
Figure 2 is a schematic view showing an interested target and origin of coordinate selected inside thereof;
Figure 3 is a schematic view showing intersection points of three straight lines with the surface and a plurality of data points adjacent thereto, each of the straight lines in line with one direction of coordinate axis passing through the selected origin of coordinate within the interested target;
Figure 4 is a schematic view for recognizing dimension of the interested target;
Figure 5 is a schematic view for recognizing cutting tools;
Figures 6a and 6b are schematic views of a method to seek a new coordinate axis if the coplanarity occurs;
Figure 7 is a schematic view of a method to recognize a cuboid; and
Figure 8 is a schematic view of a method to recognize a cutting tool.
DETAILED DESCRIPTION OF THE EMBODIMENTS
Technical solutions of the present invention will be described hereinafter in more detail by the way of embodiment with reference to figures 1-8 of the attached drawings, wherein the same or like reference numerals refer to the same or like elements throughout the specification. The explanation to the embodiment
FP150096AU of the present invention with referring to the accompanying drawings is intended to interpret the general inventive concept of the present invention, rather than being construed as a limiting to the present invention.
Taking CT image data obtained by scanning and re-constructing of a security inspection CT system as an example, an embodiment of the present invention explains how to recognize prohibited articles or dangerous articles by a three-dimensional data processing and recognizing method of the present invention. Please be noted that, the skilled person in the art should understand that the present three-dimensional data processing and recognizing method is also applicable to a nuclear magnetic resonance system or the like, so as to recognize the interested targets.
The present general concept is to perform searching, extraction, mergence, statistics and/or shape recognition on the three-dimensional data obtained by scanning and re-constructing of the security inspection CT system by using feature data of the interest targets of interest or suspected targets, for example, dangerous articles or prohibited articles. Taking into consideration of the shortcomings of schemes in the prior art that image segmentation to the threedimensional data is still relatively difficult and the accuracy and universality is not so good, and so on, the present invention proposes out the present threedimensional data processing and recognizing method from a different viewpoint thereto. Further, the present invention solves technical problems to perform searching, statistics and shape recognition with respect to the interested targets.
The present three-dimensional data processing and recognizing method will be described in detail below with reference to figures 1-8 in combination with the following examples.
The present three-dimensional data processing and recognizing method mainly includes: data processing, constructing surface data from the threedimensional data, and recognizing in shape and dimension (or on basis of the shape and dimension of the interested targets). In addition, the present invention
FP150096AU also provides a quick recognizing method for the articles to be inspected having shapes such as cuboids, cylinders and/or cutting tools.
Data Processing
At first, the security inspection CT system scans the object to be detected and re-constructs it to obtain CT image data (i.e., three-dimensional data). And then, prohibited articles (such as cutting tools) may be recognized based on the obtained three-dimensional data.
By using feature data of the prohibited articles, the obtained threedimensional data are searched, merged and the three-dimensional data matching to such feature data are extracted, thereby the interested targets are produced or constituted by the extracted three-dimensional data. The prohibited articles can be common dangerous goods, such as cutting tools, explosives, guns and the like. The feature data can be data about composition materials of the common dangerous articles, for example, iron, copper, or heavy metals. Such data can be any one of attenuation coefficient data, density data and atomic number data (or energy spectrum data of energy spectrum CT), and any combination thereof. Typically or preferably, the data can be density data and/or the atomic number data.
Among the matched three-dimensional data, adjacent data points are merged and classified as one group. Images of these merged three-dimensional data can be displayed individually or highlighted, for inspector’s inspection.
A number of data points among the merged three-dimensional data can be counted to estimate dimension of the prohibited articles. Of course, it is possible at this time to make statistics on mass, position, density and atomic number where the merged three-dimensional data are located, to respectively derive mean value thereof, so that more specific information on position, volume, mass, material species of the interested targets located in the articles to be inspected can be obtained. These would help to further recognize the prohibited articles. It should be understood that whether the articles to be detected or
FP150096AU inspected are drugs or explosives, can be recognized in dependency of their density and atomic number.
It should be noted that the interested targets formed by the extracted three-dimensional data or the interested targets formed by the merged threedimensional data can be used to perform the step of constructing surface data from the three-dimensional data, before a step of recognizing in shape and size.
Constructing Surface data from the three-dimensional data
Herein, taking the interested targets formed by the merged threedimensional data as an example, the step of “constructing surface data from the three-dimensional data” is described in detail. It should be understood that the step of “constructing surface data from the three-dimensional data” can also be performed on basis of the interested targets formed by the extracted threedimensional data.
Specifically, the merged three-dimensional data constitutes or makes up a three-dimensional target (i.e., an interested target). Firstly, one point I is selected within the interested target. Preferably, a method of selecting the point I includes : selecting a maximal value and a minimal value of the data within the merged threedimensional data in directions of x, y, z axes of rectangular coordinate system (herein, the direction of x, y, z axes of rectangular coordinate system can be arbitrarily selected, as long as any two of the three directions are perpendicular to each other, similarly hereinafter); and respectively calculating median of the maximal value and the minimal value in the above described three directions, wherein coordinates of the median can be selected as coordinates of the selected point I.
It is further noted that with respect to some interested targets having abnormal shapes, after finding out the point I, it is possible to check up whether the point I has adjacent data in three dimensional directions (i.e., the threedimensional data of the interested target in forward, backward, left, right, upward, downward directions thereof). If not, it is necessary to again find one point near the
FP150096AU point I as a new point I, which can satisfy the above condition. As such, it can be sure that the point I will always fall within the interested target regardless of its shape.
A spherical coordinate system is established with the point I as origin of coordinate. In the spherical coordinate system, an angle (θ, φ) of each point among the merged solid data (the three-dimensional data) and a distance thereof from the point I are calculated, wherein 0<θ<π, 0<φ<2π(ίη a physical sense, Θ is a zenith angle, and φ is an azimuthal angle). Δθ and είηθΔφ are set and a point within each solid angle βίηθΔφΔθ, which is farthest away from the point I, is selected as one of surface data.
Preferably, some data out of the surface can be eliminated. One preferable means is to obtain an interpolation point of each surface datum(corresponding to each data point) as described above, which is obtained by interpolating surrounding points of the selected point. If a distance of the interpolation point from the point I is larger than a distance of the selected point from the point I, then the selected point which has been considered as a surface data point will be eliminated. Specifically with reference to figure 1 (showing out four points A, B, C, and D), since the distance of the interpolation point of the point B derived by linear interpolation of points A and C from the point I is larger than the distance of point B from the point I, the point B is eliminated.
After the above eliminating process, the left data can be considered as more reliable surface data.
Of course, as for some three-dimensional targets or interested targets having hollow cavity structures, we are not only concerned of outer surface data of such three-dimensional targets, but also concerned of inner surface data thereof. In a manner similar to the above described, when selecting the inner surface data, we choose the point closest to the point I within each solid angle εϊηθΔφΔθ as the inner surface data. Similarly, in the step of eliminating the inner surface data, if the obtained interpolation point of the selected point is closer to the point I than the selected point, then the selected point is eliminated.
FP150096AU
In addition, another preferable method to eliminate the data is described as follows. If a ratio of a distance of the interpolation point of the selected point (for example, obtained with respect to the selected point) from the point I to a distance of the selected point from the point I falls within a predetermined range of threshold (taking as an example, [0.95, 1.05]), then the selected point should be remained; otherwise the selected point should be eliminated. As such, such method can also be applicable to eliminating of the above described surface data and inner surface data.
At this time, the constructed surface data can be displayed to the operator, so as to intuitively and clearly observe shape of the interested target and perform an initial recognition (for example, in shape).
In order to accelerate algorithm, in certain circumstances, for example the interested target having a very simple shape, it is possible to only construct surface data near the three coordinate axes x, y, and z, without constructing all the surface data of the interested target.
Furthermore, another method to construct the surface data is: as for an interested target, each of three-dimensional data is analyzed; if there are additional data adjacent to it in all three dimensional directions (for example, the upward and downward direction, the left and right direction, and the fore-and-aft direction), then the three-dimensional data can be deemed to be non-surface data; otherwise, the remaining three-dimensional data shall be surface data.
Recognition in shape and size
A point I is arbitrarily selected inside the interested target. As described above, with referring to figure 2, it is preferable to select a median of a maximal value and a minimal value of the three-dimensional data in a direction of respective three x, y, and z coordinate axes as the coordinate (x0, y0, z0) of the point I.
Each surface data (x, y, z) among the above constructed surface data is searched to obtain a point P having a minimal value of (x-xo)2+(y-yo)2, that is, as
FP150096AU shown in figure 3, an intersection point of z axis with the upper surface (in the present invention, an intersection point of the straight line with the surface is an approximate intersection point, since the surface consists of a plurality of discrete data points). If it is defined as z>z0or z<z0, one intersection point can be obtained respectively at the upper surface and the lower surface.
As such, a point Q is obtained, having a minimal value of (y-y0)2+(z-z0)2, that is, an intersection point of x axis with the front surface as shown in the figure 3. If it is defined as x>x0or x<x0, one intersection point can be obtained respectively at the front surface and the rear surface.
As such, a point O is obtained, having a minimal value of (z-zo)2+(x-xo)2, that is, an intersection point of y axis with the right surface as shown in the figure 3. If it is defined as y>y0or y<y0, one intersection point can be obtained respectively at the right surface and the left surface.
Herein, for sake of clarity, only intersection points P, Q, O in the positive direction of the x, y, z axes with the corresponding surface are illustrated. Of course, each negative direction of the x, y, z axes will also form an intersection point (not shown) with the respective surface.
Based on the obtained surface data, as shown in figure 4, a plurality of points (for example, four points A', B’, O’, D’) nearest the point P are found out around the point P. It is possible to determine whether the above points co-plane with the point P. If so, a plane equation (where a plane PZ is positioned) is given out.
Likewise, with respect to the point Q, a plane QX is obtained by the same process.
Likewise, with respect to the point O, a plane OY is obtained by the same process.
Hereto, the skilled person in the art should understand from the above solving process that two intersection points are obtained by intersecting a straight line passing through a point within the interested target in line with one direction of the three coordinate axes, with the surfaces of the interested target; each
FP150096AU intersection point and a plurality of data point adjacent thereto constitute a plane, thereby obtaining a pair of planes of the interested target intersecting with one of three coordinate axes directions.
Specifically, if the intersection points at which straight lines passing through the point I within the interested target in line with positive directions (instead, negative directions) of the three coordinate axes respectively intersect with the surfaces of the interested target, are obtained by the above solving process, then the three planes PZ, QX, and OY consisting of each intersection point and a plurality of data points adjacent to it can be acquired.
It should be verified whether the obtained three planes PZ, QX, and OY are coplanar or not. If not, a step of determining whether the three planes are coplanar can be omitted.
If the three planes are coplanar, then it is necessary to find new three coordinate axes by a remedy described below. This process continues until the three planes formed by intersection points of the positive (or negative) directions of the new three coordinate axes with the surfaces of the interested target are not coplanar.
If one of the three planes PZ, QX, and OY is coplanar with one of the remaining two planes, then a straight line passing through the point I and parallel to a normal direction of the coplanar plane is selected as a coordinate axis 1.
A method of solving the other two coordinate axes 2 and 3 being perpendicular to the coordinate axis 1 with each other is one of cases (i) and (ii).
(i) With reference to figure 6a, in case of the remaining plane approximately perpendicular to the coplanar planes, a straight line passing through the point I and being parallel to the normal direction thereof is selected as a straight line L1. A perpendicular line to the coordinate axis 1 is made within a plane N consisting of the coordinate axis 1 and the straight line L1, and is used as a coordinate axis 2. A normal line passing through the point I within the plane N is a coordinate axis 3.
FP150096AU (ii) With reference to figure 6b, in case that the remaining plane is not approximately perpendicular to the coplanar planes, two straight lines perpendicular to each other are selected within a plane M as the coordinate axes 2 and 3, the plane M passing through the point I and perpendicular to the coordinate axis 1.
A straight line passing through the point I within the plane M is arbitrarily selected as the coordinate axis 2, a line perpendicular to the coordinate axis 2 is the coordinate axis 3. An intersection point of the positive (or negative) direction of the new coordinate axes 1, 2 and 3 with the surfaces of the interested target can be solved, and each intersection point and a plurality of points adjacent to it constitutes a plane. It should be determined whether the three planes consisted as described above are coplanar. If two or more of the three planes are coplanar, then the coordinate axis 2 is transformed along one rotation direction within the plane M (for example, to coordinate axes 2’, and 2”), and the coordinate axis 3 is accordingly transformed (for example, to coordinate axes 3’ and 3”). The above steps are repeated until any two of the above three planes are not coplanar or the coordinate axis 2 is rotated along the one rotation direction by 180 degrees.
Herein, a method to solve intersection points of the positive directions of new coordinate axes 1, 2 and 3 with the surfaces of the interested target can be described as follows:
It is assumed that the surface data point of the interested target is a point A, a vector of the point I to the point A is a vector a, a unit vector of the positive direction of one of the new coordinate axes 1, 2 and 3 is n, and a-'h is calculated. If a value of α· η is above zero, then \a\ -(a-ril)2 is continuously calculated. A point at which a least value of |o|2 -(α· η if occurs is an intersection point of the positive direction of the coordinate axis with the surface of the interested target. A method of solving the intersection point of the positive direction of the new coordinate axes 1, 2 and 3 with the surfaces of the interested target is not limited to the application herein, and is also applicable to the process
FP150096AU of directly obtaining the intersection points after obtaining the surface data of the interested target.
The new coordinate axes as mentioned herein do not transform coordinates of the surface data in practical calculation. Of course, the coordinate transformation method can be used as an alternative calculating method or a substitute of the present invention. In view that the skilled person in the art can understand details of the above calculating method, they are not described in detail. The use of new coordinate axes mainly utilizes characteristics of perpendicularity of the three axes of the rectangular coordinate system to each other and the positive directions thereof satisfying right-hand rule. Therefore, it should be understood that any calculating methods using such property or similar to the method described herein will fall within the scope of the present invention.
After the above verifying process, if the directions of the new coordinate axes 1,2 and 3 does not correspond to those of the original coordinate axes x, y, z, then the original coordinate axes are replaced by the new coordinate axes 1, 2, 3. Thereafter, the coordinate axes as shown in figure 4 are considered to be the directions of the new coordinate axes 1, 2, 3.
As shown in figure 4, a method of determining whether a plurality of points are coplanar is as follows: firstly selecting out three non-collinear points (for example, points A’, B’, C’), then writing a plane equation thereof, and finally calculating distances of other points (for example, point D’) from the plane. If the distance is less than a threshold (such as, a value which is obtained by dividing a difference of a maximal value of the coordinate of the surface data in a normal direction of the plane and a minimal value thereof by 100), then they are determined to be coplanar; otherwise they are determined to be non-coplanar.
An intersection line YZ of the plane OY with the plane PZ is calculated, and the data points falling onto the intersection line are found out by analyzing the obtained three-dimensional data. A distance MN between two end points M and N is calculated, and is set as a length in x direction.
FP150096AU
Likewise, an intersection line of the plane PZ with the plane QX is calculated, and the data points falling onto the intersection line are found out by analyzing the obtained three-dimensional data. A distance between two end points thereof is calculated, and is set as a length in y direction.
Likewise, an intersection line of the plane QX with the plane OY is calculated, and the data points falling onto the intersection line are found out by analyzing the obtained three-dimensional data. A distance between two end points thereof is calculated, and is set as a length in z direction.
The method of determining whether the points fall onto the intersection line can be achieved by calculating whether the distances of the points from the intersection line are less than a certain threshold (such as, a value which is obtained by dividing a difference of a maximal value of the coordinate of the surface data and a minimal value thereof on the intersection line by any value between 50 and 100).
In addition, it is possible to calculate an intersection angle between the intersected planes. If all the intersection angles are approximately 90°, then the interested target can be recognized to have a shape of cuboid.
There are two methods to determine whether the two planes are coplanar in the present invention:
1. It is assumed that equations of the two planes are respectively a1x+b1y+c1z+d1=0, and a2x+b2y+c2z+d2=0. If it is determined that a1/a2= b1/b2= c1/c2= d1/d2 is approximately true, then the two planes are approximately coplanar; otherwise they are non-coplanar.
2. Because in the present invention the plane equation is solved by a plurality of points within the plane, the distances of a plurality of points on one of the two planes from the other of the two planes can be calculated; and if the distances of all points on one plane from the other plane are less than a predetermined threshold, then the two planes are determined to be coplanar, otherwise non-coplanar.
FP150096AU
If the interested target has a shape of cuboid, then it has six planes as shown in figure 5 (a front surface, a rear surface, a left surface, a right surface, a upper surface and a lower surface). If a density (or atomic number, or attenuation coefficient) of the interested target corresponds to a feature of cutting tools, then a function of recognizing the cutting tools are turned on. There are 12 intersection lines through the planes intersecting with each other, and the lengths of the intersection lines are calculated (i.e., lengths on x, y, z directions as described above). If the lengths of two intersection lines are much less than those of other (for example 8) intersection lines (such as a ratio thereof is less than 1/10), then the interested target can be determined as a knife-shaped object or cutting-shaped object. Preferably and further, if the intersection line has a least length (for example a thickness d) less than a threshold (for example, 2 mm), then it can be precisely determined as a dangerous cutting tool.
Based on the above descriptions, the present invention also provides a quick recognition for objects having shapes of cuboid, cylinder, and cutting tools, or the like.
Cuboid recognition
In a recognition solution of cuboid, it still employs the steps of selecting point I, and obtaining planes of the surfaces by intersection points of coordinate axes with surface data, as set out in the parts of “recognition in shape and dimension”. As long as a pair of parallel or approximately parallel planes, formed by intersecting of one coordinate axis (one of x, y, z coordinate axes) with the surfaces of the interested target, are found out, it can turn to the following steps.
With reference to figure 7, it is assumed that an intersection point of a positive direction of one of three coordinate axes intersecting with a pair of parallel or substantially parallel planes of the interested target is point G, and an intersection point of a negative direction of the coordinate axis intersecting with the pair of parallel or substantially parallel planes is point G’.
FP150096AU
A plane R is made, which passes through the point I and is parallel to the pair of parallel or substantially parallel planes. The points which fall onto the plane R, within the surface data of the interested target, should be analyzed, and if these points constitute a rectangular shape, then the following steps will continue.
A point K is selected on a connection line of the point G to the point I, and lies between the point G and the point I. In the present embodiment, a ratio of the distance of the point K from the point G to the distance of the point K from the point I is 1:a (for example, a =10). A plane S is made which passes through the point K and is parallel to the plane R. The points which fall onto the plane S, within the surface data of the interested target, should be analyzed, and if these points constitute a rectangular shape having the same length and width as that of the above described rectangular shape, then the following steps will continue.
A point K’ is selected on a connection line of the point G' to the point I, and lies between the point G’ and the point I. In the present embodiment, a ratio of the distance of the point K’ from the point G’ to the distance of the point K’ from the point I is 1:a (for example, a =10). A plane S’ is made, which passes through the point K’ and is parallel to the plane R. The points which fall onto the plane S’, within the surface data of the interested target, should be analyzed, and if these points constitute a rectangular shape having the same length and width as that of the above described rectangular shape, then the interested target is determined to have a shape of cuboid.
Dimensions of the cuboid on three directions can be estimated by the length and width of the rectangular shape and the distance between the point G and the point G’ (the dimension of the third direction equals to this distance multiplying by an absolute value of cosine of the intersection angle of the coordinate axis with the normal direction of the plane R).
The above arithmetic can be relaxed or simplified. For example, it is possible to calculate an intersection point of only a single direction (positive or negative direction) of one coordinate axis with the surface of the interested target, and to carry out the subsequent calculating operations similar to those as
FP150096AU described above. In this way, as compared with the above steps, almost half amount of calculation can be saved.
Cylinder recognition
The cylinder recognition is similar to the solution of the above cuboid recognition, but for the following differences: a shape obtained by intersecting the interested target with a plane parallel to the pair of parallel or substantially parallel planes is approximately a circle, and then the interested target can be determined to have a cylindrical shape. Finally, dimension of the cylinder can be estimated by a radius of the circle and the distance of the point G’ to the point G.
Cutting tool/ sheet-shaped object recognition
In a recognition solution of cutting tool/ sheet-shaped object, it still employs the steps of selecting point I, and obtaining planes of the surfaces by intersection points of coordinate axes (for example x coordinate axis) with surface data, as set out in the parts of “recognition in shape and dimension”. As long as a pair of parallel or approximately parallel planes, formed by intersecting a certain coordinate axis (one of x, y, z coordinate axes) with the surfaces of the interested target, are found out, it can turn to the following steps.
With reference to figure 8, when a pair of parallel or substantially parallel planes of one coordinate axis with the surfaces of the interested target is found out, the intersection angle of the two planes can be calculated. If the intersection angle is less than a threshold (for example, 6°, in case of recognizing the cutting tool like axe, the threshold can be adjusted to be larger), then it can directly turn to the following steps.
If the above conditions are not satisfied, the coordinate axis can be replaced by for example, y axis, or even z axis, to continue the above calculation. Provided that after calculating with the three coordinate axes, it always does not find out a pair of parallel or substantially parallel planes having an intersection
FP150096AU angle less than a threshold, the interested target can be almost determined not to be a cutting tool.
In the case of an intersection angle of the pair of parallel or substantially parallel planes less than the threshold, the following steps are carried out: calculating an intersection line L2 of the pair of parallel or substantially parallel planes; and finding out a point V on the intersection line L2.
One preferable method to do this is, to calculate a projection point V of the point I onto the intersection line L2; and to obtain a plane a passing through the point V and with the line L2 as a normal vector. The surface data of the interested target are searched to find out which points fall onto the plane a. If these points constitute trapezoid or substantially trapezoid, and a shorter bottom side of the trapezoid is less than a threshold (for example, 2mm), then the interested target can be determined as a dangerous cutting tool.
Of course, before recognizing the interested target as a cutting tool, its composition can be analyzed to determine whether it corresponds to the feature of the cutting tool.
The recognition of two-dimensional shapes, for example, rectangular, circle, trapezoid, can be done with neural network method or by analyzing feature.
Concerning the above, the present invention extracts three-dimensional data by matching features, and recognizes shapes of the object by analyzing and calculating based on points, lines and planes, thereby recognizing prohibited articles such as cutting tools. Various CT systems can employ such method, to recognize the dangerous cargos. Other systems generating the three-dimensional data, such as nuclear magnetic resonance system or the like, can also utilize such method, to recognize the interested target. That is, the three-dimensional data of the interested target can be searched, extracted, merged, counted, and recognized in shape, by the feature data of the interested target. Since image segmentation of the three-dimensional data is yet relatively difficult, and accuracy and generality is not so good, the present method solves problems of searching, counting and recognizing in shape of the interested target from a different viewpoint.
FP150096AU
It should be noted that a method of the present invention is applicable to systems generating three-dimensional data such as CT and nuclear magnetic resonance system, and so on. Although in the present invention, a rectangular coordinate system and a spherical coordinate system are taken as an example to 5 describe the three-dimensional data processing and recognizing method of the present invention, it should be understood that using other coordinate systems based on the present invention or simple development of the present invention all fall within the scope of the present invention.
Although some embodiments of the general inventive concept are illustrated and explained, it would be appreciated by those skilled in the art that modifications and variations may be made in these embodiments without departing from the principles and spirit of the general inventive concept of the disclosure, the scope of which is defined in the appended claims and their equivalents.

Claims (33)

  1. What the claims is:
    1. A three-dimensional data processing and recognizing method, characterized in that, it comprises following steps of:
    scanning and re-constructing objects to be detected so as to obtain three-dimensional data for recognition of the objects to be detected;
    extracting data matching to features from the three-dimensional data, so that the extracted data constitutes an interested target in order to display and recognize the object to be detected.
  2. 2. The three-dimensional data processing and recognizing method of claim 1, characterized in that, the data matching to features are obtained by searching the threedimensional data in conformity with existing data of common dangerous articles.
  3. 3. The three-dimensional data processing and recognizing method of claim 2, characterized in that, the existing data is any one of attenuation coefficient data, density data and atomic number data or a combination thereof.
  4. 4. The three-dimensional data processing and recognizing method of claim 1, characterized in that, with respect to the data matching to features, adjacent data points are merged and classified as one group, to form an image of the merged interested target, for an inspector to inspect and recognize it.
  5. 5. The three-dimensional data processing and recognizing method of claim 4, characterized in that, with respect to the merged data, a number of data points therein is used to estimate size of the object to be detected.
    FP150096AU
  6. 6. The three-dimensional data processing and recognizing method of claim 5, characterized in that, an average value is obtained by statistics of any one of positions, mass, attenuation coefficient, density, and atomic number, and/or a combination thereof, of an area in which the merged data are located.
  7. 7. The three-dimensional data processing and recognizing method of any one of claims 1-6, characterized in that, analyzing each three-dimensional data point of the interested target, if the three-dimensional data point has other adjacent three-dimensional data points in three-dimensional directions, then it will be deemed as non-surface data; otherwise, if not, it will be deemed as a surface data.
  8. 8. The three-dimensional data processing and recognizing method of any one of claims 1-6, characterized in that, the interested target has surface data, which is obtained by an analyzing method of spatial angle distribution, the method comprising the following steps:
    selecting an arbitrary point I within the interested target as an origin of coordinate, and then establishing a corresponding spherical coordinate system;
    calculating a zenith angle θ and an azimuthal angle φ of each point of the three-dimensional data as well as a distance of the point from the original of coordinate I under the spherical coordinates;
    setting up Δθ and είηθΔφ, and selecting a point having a farthest distance from the origin of coordinate within each solid angle είηθΔφΔθ as a surface datum.
  9. 9. The three-dimensional data processing and recognizing method of claim 8, characterized in that, with respect to each surface data point obtained thereby, an interpolation point of the selected point is obtained by interpolating calculation of
    FP150096AU its surrounding surface data points, if a distance of the interpolation point from the point I is larger than a distance of the selected point from the point I, then the selected point is eliminated.
  10. 10. The three-dimensional data processing and recognizing method of claim 8, characterized in that, with respect to each surface data point obtained thereby, an interpolation point of the selected point is obtained by interpolating calculation of its surrounding surface data points, if a ratio of a distance of the interpolation point from the point I to a distance of the selected surface data point from the point I falls within a predetermined range, then the selected surface data point is remained; otherwise, it is deleted.
  11. 11. The three-dimensional data processing and recognizing method of claim 8, characterized in that, in a circumstance of the interested target having a hollow structure, inner surface data of the interested target are recognized by the above analyzing method of spatial angle distribution, a point having a closest distance from the origin of coordinate within each solid angle βΐηθΔφΔθ is selected as one of the inner surface data.
  12. 12. The three-dimensional data processing and recognizing method of claim 11, characterized in that, with respect to each inner surface data point obtained thereby, an interpolation point of the selected point is obtained by interpolating calculation of its surrounding surface data points, if a distance of the interpolation point from the point I is less than a distance of the selected point from the point I, then the selected surface data point is deleted.
    FP150096AU
  13. 13. The three-dimensional data processing and recognizing method of any one of claims 7-10, characterized in that, solving two intersection points of a straight line passing through a point within the interested target in line with one direction of three coordinate axes directions, with surfaces of the interested target, each intersection point and adjacent data points thereof constitutes a plane, thereby obtaining a pair of planes passing through the two intersection points;
    the solving method comprises the following steps:
    setting the point I within the interested target to have a coordinate (x0, y0, z0);
    searching each surface data (x, y, z) of all the obtained threedimensional data, to obtain a point at which (x-xo)2+(y-yo)2 has a minimal value, the point being an intersection point of a straight line in line with z coordinate axis with the corresponding surface;
    likewise, obtaining another point at which (y-y0)2+(z-z0)2 has a minimal value, this point being an intersection point of a straight line in line with x coordinate axis with the corresponding surface; and likewise, obtaining a further point at which (z-z0)2+(x-x0)2 has a minimal value, this point being an intersection point of a straight line in line with y coordinate axis with the corresponding surface.
  14. 14. The three-dimensional data processing and recognizing method of claim 13, characterized in that, an intersection point of a positive direction or a negative direction of the straight line along three coordinate axes passing through the point I within the interested target with the surfaces of the interested target are obtained by the above solving method, and then each intersection point and a plurality of data points adjacent to it constitute three planes, and any two of three planes are not co-planar.
    FP150096AU
  15. 15. The three-dimensional data processing and recognizing method of claim 14, characterized in that, if two or more of the three planes are coplanar, then it is necessary to find out new three coordinate axes by a remedy as follows, until any two of three planes obtained by the intersection points of the positive or negative direction of the new three coordinate axes with the surfaces of the interested target are not coplanar:
    provided that one plane of the three planes is co-planar with another plane of the other two planes, a straight line passing through the point I and in line with a normal direction of their co-planar plane is selected as a coordinate axis 1;
    a method of solving another two coordinate axes 2 and 3 perpendicular to the coordinate axis 1 with each other is one of cases (i) and (ii);
    (i) if a remaining plane is approximately perpendicular to the co-planar plane, then a straight line passing through the point I and in line with a normal direction of the remaining plane is selected as a straight line L1, a line perpendicular to the coordinate axis 1 is obtained within a plane N formed by the coordinate axis 1 and the straight line L1, wherein the perpendicular line is used as a coordinate axis 2 and a normal line of the plane N passing through the point I is used as the coordinate axis 3;
    (ii) if a remaining plane is not perpendicular to the coplanar plane, then two straight lines perpendicular to each other are selected out within a plane M which passes through the point I and is perpendicular to the coordinate axis 1, as coordinate axes 2 and 3; after that, an intersection point of a positive or negative direction of the new coordinate axes 1, 2 and 3 with the surfaces of the interested target are solved, and each intersection point and a plurality of data points adjacent to it form a plane, in order to determine whether the newly obtained three planes are coplanar or not, if two or more of the three planes are coplanar, the coordinate axis 2 is transformed along a rotation direction within the plane M, and the coordinate axis 3 is accordingly transformed;
    FP150096AU repeating the above steps until any two of the three planes are not coplanar or the coordinate axis is rotated along one direction within the plane M by a half round.
  16. 16. The three-dimensional data processing and recognizing method of claim 15, characterized in that, a method of solving the intersection points of the positive or negative direction of the new coordinate axes 1, 2 and 3 with the surface of the interested target comprises the following steps:
    a surface data point of the interested target is supposed to be a point A, a vector from the point I to the point A is a, a unit vector of the positive or negative direction of one among the new coordinate axes 1,2 and 3 is h, and 5 ’ ” is calculated, if a value of 5·« js above zero, then (fl-ηί) js further calculated, and a point at which _ ' 0 pas a |east value is the intersection point of the positive or negative direction of the coordinate axis with the surface of the interested target.
  17. 17. The three-dimensional data processing and recognizing method of claim 13, characterized in that, upon determining whether the intersection point is coplanar with a plurality of adjacent data points, three non-collinear points are selected firstly, and their plane equation is written out, and distances of other data points to this plane is calculated;
    if such distance is less than a first predetermined threshold, the point is determined to be in the plane; and otherwise, the point is determined to be out of the plane.
  18. 18. The three-dimensional data processing and recognizing method of claim 14 or 15, characterized in that,
    FP150096AU a method of determining whether two planes are coplanar comprise the following steps of:
    (a) assuming equations of the two planes are respectively a1x+b1y+c1z+d1=0, and a2x+b2y+c2z+d2=0, if it is determined that a1/a2= b1/b2= c1/c2= d1/d2 is approximately true, then the two planes are approximately coplanar; otherwise they are non-coplanar; or (b) the distances of a plurality of points on one of the two planes from the other of the two planes are calculated; and if the distances of all points on one plane from the other plane are less than a second predetermined threshold, then the two planes are determined to be coplanar, otherwise non-coplanar.
  19. 19. The three-dimensional data processing and recognizing method of claim 14, characterized in that, if a distance of one data point from an intersection line between two planes is less than a third predetermined threshold, then the point is determined to lie onto the intersection line, otherwise, not onto the intersection line.
  20. 20. The three-dimensional data processing and recognizing method of claim 19, characterized in that, a distance between two outermost end points onto each intersection line is calculated, to obtain a dimension in each direction of the interested target.
  21. 21. The three-dimensional data processing and recognizing method of claim 20, characterized in that,
    If all of intersection angles between the intersecting planes are approximately 90°, then the interested target is determined to be cuboid.
  22. 22. The three-dimensional data processing and recognizing method of claim 21, characterized in that,
    FP150096AU the cuboid or approximate cuboid has 12 intersection lines in total due to intersections of any two planes therein, and if it is obtained after calculation that two of the intersection lines have much smaller lengths than eight of others, then the interested target is recognized as a cutting-shaped article.
  23. 23. The three-dimensional data processing and recognizing method of claim 22, characterized in that, the interested target will be recognized as a cutting tool, if a thickness of the interested target is much less than a length and a width thereof, and any one of the thickness, density, atomic number and attenuation coefficient or a combination thereof is matched with corresponding features of the cutting tool.
  24. 24. The three-dimensional data processing and recognizing method of claim 13, characterized in that, assuming a respective intersection points G and G’ of a positive or negative direction of one of the three coordinate axes with a pair of parallel or approximately parallel planes of the interested target;
    drawing a plane R passing through the point I and parallel to surfaces of the pair of parallel or approximately parallel surfaces, and analyzing the surface data points of the interested target which fall onto the plane R, if these surface data points constitute a rectangular shape, then the following steps (c1) and/ or (c2) are continued;
    step (c1): a point K is chosen in a connecting line between the point G and the point I, and lies between the point G and the point I, a plane S is drawn out to pass through the point K and to be parallel with the plane R, the surface data points of the interested target which fall onto the plane S are analyzed, if these surface data points constitute a rectangular shape having same or substantially same length and width as that of the previous described rectangular shape, then the interested target is determined to have a cuboid shape;
    FP150096AU step (c2): a point K’ is chosen in a connecting line between the point G’ and the point I, and lies between the point G’ and the point I, a plane S’ is drawn out to pass through the point K’ and to be parallel with the plane R, the surface data points of the interested target which fall onto the plane S’ are analyzed, if these surface data points constitute a rectangular shape having same or substantially same length and width as that of the previous described rectangular shape, then the interested target is determined to have a cuboid shape.
  25. 25. The three-dimensional data processing and recognizing method of claim 24, characterized in that, dimensions of the interested target on three directions are estimated by the length and width of the rectangular shape and the distance between the point G and the point G’.
  26. 26. The three-dimensional data processing and recognizing method of claim 13, characterized in that, assuming respective intersection points G and G’ of a positive or negative direction of one of the three coordinate axes with a pair of parallel or approximately parallel planes of the interested target;
    drawing a plane R passing through the point I and parallel to the pair of parallel or approximately parallel surfaces, and analyzing the surface data points of the interested target which fall onto the plane R, if these surface data points constitute a circle shape, then the following steps (d1) and/ or (d2) are continued;
    step (d1): a point K is chosen in a connecting line between the point G and the point I, and lies between the point G and the point I, a plane S is drawn out to pass through the point K and to be parallel with the plane R, the surface data points of the interested target which fall onto the plane S are analyzed, if these surface data points constitute a circle shape having same or substantially same radius as the previous described circle, then the interested target is determined to have a cylindrical shape;
    FP150096AU step (d2): a point K’ is chosen in a connecting line between the point G’ and the point I, and lies between the point G’ and the point I, a plane S' is drawn out to pass through the point K’ and to be parallel with the plane R, the surface data points of the interested target which fall onto the plane S’ are analyzed, if these surface data points constitute a circle shape having same or substantially same radius as the previous described circle, then the interested target is determined to have a cylindrical shape.
  27. 27. The three-dimensional data processing and recognizing method of claim 26, characterized in that, dimension of the interested target is estimated by a radius of the circle and the distance between the point G and the point G’.
  28. 28. The three-dimensional data processing and recognizing method of claim 13, characterized in that, calculating an intersecting angle between a pair of parallel or approximately parallel planes, to determine whether this intersecting angle is less than a fourth predetermined threshold, if this intersecting angle is less than a fourth predetermined threshold, an intersecting line L2 of the pair of parallel or approximately parallel planes is calculated;
    finding out a point V onto the intersecting line L2;
    obtaining a plane a passing through the point V and with the line L2 as a normal vector;
    searching the surface data points falling onto the plane a, if these surface data points constitute a trapezoid or an approximate trapezoid, and shorter bottom side of the trapezoid has a length which is less than a fifth predetermined threshold, then the interested target is determined to be a dangerous cutting tool.
    FP150096AU
  29. 29. The three-dimensional data processing and recognizing method of claim 28, characterized in that, if the intersecting angle is equal to or larger than the fourth predetermined threshold, then another pair of parallel or approximately parallel planes are calculated and obtained wherein the pair of planes passing through intersection points are formed by intersecting directions of at least one of the other two coordinate axes with the surfaces of the interested target, an intersecting angle between the another pair of parallel or approximately parallel planes is again calculated, in order to determine whether the intersecting angle is less than the fourth predetermined threshold;
    if the intersecting angle is less than the fourth predetermined threshold, the interested target is determined to be a cutting tool, otherwise not a cutting tool.
  30. 30. The three-dimensional data processing and recognizing method of claim 29, characterized in that, a projection point of the point I onto the intersecting line L2 is calculated as the point V.
  31. 31. The three-dimensional data processing and recognizing method of claim 28, characterized in that, before making a recognition of the cutting tool, composition of the interested target is analyzed to confirm whether it corresponds to the features of the cutting tool.
  32. 32. The three-dimensional data processing and recognizing method of claim 1, characterized in that, the three-dimensional data are obtained by scanning and re-constructing the object to be detected by a security inspection CT system or a nuclear magnetic resonance system.
    FP150096AU
  33. 33. The three-dimensional data processing and recognizing method of claim 8, characterized in that, a median of a maximal value and a minimal value of the threedimensional data on directions of x, y and z coordinate axes is chosen as
    5 coordinate of the origin of coordinate within the interested target.
AU2013370770A 2012-12-27 2013-07-05 Method for processing and identifying three-dimensional data Ceased AU2013370770B2 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
CN201210580652.8 2012-12-27
CN201210580652.8A CN103903297B (en) 2012-12-27 2012-12-27 Three-dimensional data processes and recognition methods
PCT/CN2013/078877 WO2014101386A1 (en) 2012-12-27 2013-07-05 Method for processing and identifying three-dimensional data

Publications (2)

Publication Number Publication Date
AU2013370770A1 true AU2013370770A1 (en) 2015-06-25
AU2013370770B2 AU2013370770B2 (en) 2017-04-27

Family

ID=50115668

Family Applications (1)

Application Number Title Priority Date Filing Date
AU2013370770A Ceased AU2013370770B2 (en) 2012-12-27 2013-07-05 Method for processing and identifying three-dimensional data

Country Status (7)

Country Link
US (1) US9465975B2 (en)
EP (1) EP2750079A3 (en)
JP (1) JP6195938B2 (en)
CN (1) CN103903297B (en)
AU (1) AU2013370770B2 (en)
BR (1) BR112015011618B1 (en)
WO (1) WO2014101386A1 (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103900503B (en) 2012-12-27 2016-12-28 清华大学 Extract the method for shape facility, safety detection method and equipment
CN103903298B (en) 2012-12-27 2017-03-01 同方威视技术股份有限公司 Three-dimensional data is processed and recognition methodss
CN104123724B (en) * 2014-07-09 2017-01-18 华北电力大学 Three-dimensional point cloud quick detection method
JP6409513B2 (en) * 2014-11-07 2018-10-24 大日本印刷株式会社 Data output restriction device for 3D object modeling
CN105809091B (en) * 2014-12-30 2020-10-16 同方威视技术股份有限公司 Inspection method and system
CN105527654B (en) 2015-12-29 2019-05-03 中检科威（北京）科技有限公司 A kind of inspection and quarantine check device
CN106932414A (en) 2015-12-29 2017-07-07 同方威视技术股份有限公司 Inspection and quarantine inspection system and its method
CN108389213B (en) * 2018-01-24 2021-06-11 上海工程技术大学 Adhesion circumference measuring method based on tangent plane point cloud
CN108898129A (en) * 2018-07-25 2018-11-27 四川理工学院 A kind of express delivery bar code automatic recognition system merging machine vision and laser scanning
CN109993827B (en) * 2019-03-29 2023-09-12 宁波睿峰信息科技有限公司 Elevation view identification method for converting building drawing into three-dimensional BIM model
CN110516531B (en) * 2019-07-11 2023-04-11 广东工业大学 Identification method of dangerous goods mark based on template matching
CN113866196B (en) * 2021-06-16 2024-06-21 北京艾尚燕食品科技有限公司 Bird's nest composition detecting system

Family Cites Families (41)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4687933A (en) 1976-12-14 1987-08-18 Telecommunications Radioelectriques Et Telephoniques, Trt Optical and mechanical scanning device
US5114662A (en) 1987-05-26 1992-05-19 Science Applications International Corporation Explosive detection system
US5006299A (en) 1987-05-26 1991-04-09 Science Applications International Corporation Explosive detection system
JP3378401B2 (en) 1994-08-30 2003-02-17 株式会社日立メディコ X-ray equipment
US6148095A (en) 1997-09-08 2000-11-14 University Of Iowa Research Foundation Apparatus and method for determining three-dimensional representations of tortuous vessels
CN1305610A (en) * 1998-02-11 2001-07-25 模拟技术公司 Computed tomography X-ray photography apparatus and method for classifying objects
US6026171A (en) 1998-02-11 2000-02-15 Analogic Corporation Apparatus and method for detection of liquids in computed tomography data
US6345113B1 (en) * 1999-01-12 2002-02-05 Analogic Corporation Apparatus and method for processing object data in computed tomography data using object projections
US6539330B2 (en) 2000-07-19 2003-03-25 Pentax Corporation Method and apparatus for measuring 3-D information
US7538764B2 (en) * 2001-01-05 2009-05-26 Interuniversitair Micro-Elektronica Centrum (Imec) System and method to obtain surface structures of multi-dimensional objects, and to represent those surface structures for animation, transmission and display
US7856081B2 (en) * 2003-09-15 2010-12-21 Rapiscan Systems, Inc. Methods and systems for rapid detection of concealed objects using fluorescence
US7302083B2 (en) * 2004-07-01 2007-11-27 Analogic Corporation Method of and system for sharp object detection using computed tomography images
US7653176B2 (en) * 2005-06-14 2010-01-26 L-3 Communications Security and Detection Systems Inc. Inspection system with material identification
JP4632891B2 (en) 2005-07-22 2011-02-16 株式会社モリタ製作所 X-ray CT imaging apparatus and X-ray CT imaging method
JP2009519436A (en) * 2005-11-09 2009-05-14 キネティック リミテッド Passive detection device
JP2009519471A (en) 2005-12-12 2009-05-14 リビール イメージング テクノロジーズ CT examination with shifted radiation
US7358733B2 (en) * 2006-02-28 2008-04-15 Ge Security, Inc. High performance security inspection system with physically isolated detection sensors
US7925065B2 (en) * 2006-08-22 2011-04-12 Siemens Medical Solutions Usa, Inc. Finding blob-like structures using diverging gradient field response
US7548606B2 (en) * 2006-08-31 2009-06-16 Ge Homeland Protection, Inc. System and method for integrating explosive detection systems
US7539283B2 (en) * 2007-01-17 2009-05-26 Ge Homeland Protection, Inc. Combined computed tomography and nuclear resonance fluorescence cargo inspection system and method
CN101617247A (en) * 2007-02-22 2009-12-30 通用电气公司 In-line high-throughput contraband detection system
CN101403711B (en) 2007-10-05 2013-06-19 清华大学 Liquid article examination method and equipment
WO2009094042A1 (en) * 2008-01-25 2009-07-30 Analogic Corporation Image combining
CN101303317B (en) * 2008-03-05 2010-11-17 中国科学院合肥物质科学研究院 Explosive substance testing system apparatus and testing method thereof
CN101271526B (en) * 2008-04-22 2010-05-12 深圳先进技术研究院 Method for object automatic recognition and three-dimensional reconstruction in image processing
US8600149B2 (en) * 2008-08-25 2013-12-03 Telesecurity Sciences, Inc. Method and system for electronic inspection of baggage and cargo
WO2010025539A1 (en) 2008-09-05 2010-03-11 Optosecurity Inc. Method and system for performing x-ray inspection of a liquid product at a security checkpoint
CA2737075A1 (en) 2008-09-15 2010-03-18 Optosecurity Inc. Method and apparatus for assessing properties of liquids by using x-rays
JP5536669B2 (en) 2008-12-05 2014-07-02 株式会社日立メディコ Medical image display device and medical image display method
CN101807244B (en) * 2009-02-13 2012-02-08 北京邮电大学 machine recognition and reconstruction method
US8180139B2 (en) * 2009-03-26 2012-05-15 Morpho Detection, Inc. Method and system for inspection of containers
CN201514387U (en) * 2009-07-03 2010-06-23 公安部第一研究所 System for utilizing multi-view X ray to automatically detect explosive in luggage
CN101592579B (en) * 2009-07-03 2011-07-06 公安部第一研究所 Method and device for automatically detecting explosive substances in luggage by using multi-view X rays
US8503750B2 (en) * 2009-10-06 2013-08-06 General Electric Company Method and apparatus for reduction of metal artifacts in CT images
US8254656B2 (en) * 2009-10-13 2012-08-28 Morpho Detection, Inc. Methods and system for selective resolution improvement in computed tomography
JP5480914B2 (en) * 2009-12-11 2014-04-23 株式会社トプコン Point cloud data processing device, point cloud data processing method, and point cloud data processing program
US20110304619A1 (en) * 2010-06-10 2011-12-15 Autodesk, Inc. Primitive quadric surface extraction from unorganized point cloud data
CN101943761B (en) * 2010-09-12 2012-09-05 上海英迈吉东影图像设备有限公司 Detection method of X-ray
EP2676128B1 (en) * 2011-02-18 2019-08-21 Smiths Heimann GmbH System and method for multi-scanner x-ray inspection
CN103903298B (en) 2012-12-27 2017-03-01 同方威视技术股份有限公司 Three-dimensional data is processed and recognition methodss
CN103900503B (en) 2012-12-27 2016-12-28 清华大学 Extract the method for shape facility, safety detection method and equipment

Similar Documents

Publication Publication Date Title
US9465975B2 (en) Three-dimensional data processing and recognizing method involving inspecting an object and extracting data matching to features
AU2013370770A1 (en) Method for processing and identifying three-dimensional data
Zhang et al. Automated method for extracting and analysing the rock discontinuities from point clouds based on digital surface model of rock mass
Liu et al. Automated matching of multiple terrestrial laser scans for stem mapping without the use of artificial references
JP6013621B2 (en) 3D data processing and identification method
US20070172129A1 (en) Method of registration in a contraband detection system
JP2009165615A (en) Tumor region size measurement method, apparatus, and program
CA2716880A1 (en) Registration of 3d point cloud data by creation of filtered density images
Takai et al. Accurate registration of MMS point clouds of urban areas using trajectory
Varley et al. Estimating depth from line drawing
Potó et al. Laser scanned point clouds to support autonomous vehicles
Zhang et al. Compensated convexity and Hausdorff stable extraction of intersections for smooth manifolds
Tu et al. Detecting facade damage on moderate damaged type from high-resolution oblique aerial images
Hu et al. Planar object detection from 3D point clouds based on pyramid voxel representation
Dou et al. Research on the measurement method of crankshaft coaxiality error based on three-dimensional point cloud
Etienne et al. State of the art in patterns for point cluster analysis
Zhao et al. Calculating vanishing points in dual space
Zhang et al. High resolution remote sensing image change detection based on law of cosines with box-whisker plot
بشير ميلاد الخراز et al. Framework for Feature Matching in Synthetic Multi-View X-ray Imaging Using SIFT
Lu et al. ICP registration with SHOT descriptor for arresters point clouds
Wang et al. Optimization of Line Extraction Based on an Affine Camera Model
Etienne et al. Oriented spatial box plot, a new pattern for points clusters
Lee et al. Object recognition and pose estimation based on 3D planar closed loop boundaries
Tran et al. 3D Geometric Primitive Alignment Revisited.
Zhang et al. Multi-Scale and Irregularly Distributed Circular Hole Feature Extraction from Engine Cylinder Point Clouds