CN103230283B - Method for optimizing ultrasonic probe imaging plane space position calibration - Google Patents

Abstract

The invention discloses a method for optimizing ultrasonic probe imaging plane space position calibration and belongs to the field of medical ultrasonic three-dimensional imaging and ultrasonic image fusion. The method is an optimizing method for calibrating the transformation relation of the space position of an ultrasonic probe imaging plane and the space position of a space position sensor, namely the space transformation matrix. Under the assisting of a three-dimensional positioning system, the method includes steps of: 1, constructing a three-dimensional positioning system; 2, manufacturing an N-line model; 3, building a coordinate system set of the positioning system; 4, calibrating a placing space position of a water tank model; 5, collecting ultrasonic images under different positions; extracting N-line marks and calculating the ultrasonic plane position; 7, mapping N-line mark points on the ultrasonic images onto a three-dimensional space plane; 8, performing registering on three-dimensional coordinate points on the imaging plane and an N line on the water tank model; 9, performing optimization calculation; and 10, calculating the optimum transformation between an imaging plane and the position sensor. The method has the advantages of reducing errors in space position calibration of the imaging plane.

Description

The optimization method that a kind of ultrasonic probe imaging plane locus is demarcated

Technical field:

The invention belongs to medical ultrasonic three-dimensional imaging and ultrasonoscopy and merge field, concrete application comprises 3 D medical ultra sonic imaging, and the spatial registration fusion of imaging of medical ultrasonic image and other pattern image (as CT, MRI).

Background technology:

Medical ultrasonic image, as a kind of harmless, real-time, inexpensive imaging examination, is widely used in medical science, and at present clinical widely used ultrasonic examination adopts Type B imaging pattern, " B ultrasonic " being commonly called as.What Type B imaging obtained is the two-dimentional fault image along with time real-time change, and doctor relies on the understanding to human anatomic structure, in brains, forms about the whether normal judgement of inside of human body three dimensional structure tissue.Development along with Medical Imaging Technology, 3-D supersonic imaging more and more receives publicity, there are at present two kinds of modes that realize three-dimensional imaging, a kind of is directly to adopt three dimensional ultrasound probe, by carrying out supersonic sounding in the sampling volume rectangular pyramid shape, obtain the ultrasonic echo in this three-dimensional sampling volume, and the image volume data of reconstruction of three-dimensional; Another kind of mode is directly to adopt traditional Type B imaging pattern, and the collection of two-dimentional tomography ultrasonoscopy and the splicing of space half-tone information by a plurality of known location, reconstruct three-dimensional volume data, completes three-dimensional imaging.Above-mentioned two kinds of methods cut both ways, the three dimensional ultrasound probe that the former needs specialized designs to manufacture, and its cost is high, probe size is large, uses inconvenience, and needs special supporting ultrasonic device; The latter can directly process to increase three-dimensional imaging function by computer on existing ultrasonic device, and good, easy-to-use with the compatibility of ultrasonic device in use, cost is lower.

When adopting traditional Type B imaging pattern to carry out the splicing reconstruct of 3-dimensional image, the problem of core is the relative position relation that need to accurately know between each two-dimentional tomography ultrasonoscopy.Conventionally the method adopting is at present on ultrasonic probe, to fix a position tracing sensor, the locus of accurate tracking ultrasonic probe when gathering each two-dimensional ultrasonic image, thus carry out three-dimensional splicing reconstruct.Under this pattern, need know in advance the locus sensor and the supersonic sounding imaging that are cemented on ultrasonic probe relative tertiary location relation between plane, just can carry out the three-dimensional splicing reconstruct of data.Therefore need to design a kind of method, the spatial alternation relation measurement between probe imaging plane and locus sensor can be demarcated out, this process is called probe the locus of imaging plane is demarcated.

In clinical medicine, utilizing the interior guiding of surveying of body such as ultrasonoscopy punctures is common ultrasonoscopy application.But because the resolution of ultrasonoscopy is relatively low, therefore often need to be by the fusion of other pattern images and ultrasonoscopy, to give full play to, ultrasonoscopy is real-time, safety, the high-resolution of radiology image (CT, MRI), high-precision advantage separately, for clinician provides more abundant and anatomical information accurately, help them to complete better the interventional procedures such as aspiration biopsy.In this case, in order to realize the fusion of two or more images, its Key technique problem is also the accurate alignment that will realize two kinds of image space positions, thereby makes the gradation of image message complementary sense of two kinds of mode.Need to follow the tracks of the locus of real-time ultrasonic image equally in this case, thereby know that ultrasonic probe is in where three-dimensional, and merge demonstration with the CT of correspondence position or the 3-dimensional image data of MRI.Therefore, also need high-precision the locus of probe imaging plane is demarcated.

In above-mentioned technical development process, developed the method that multiple probe imaging plane locus is demarcated, mainly comprise the kinds such as point model, three line models, areal model, two-dimensional alignment model.At the thesis for the doctorate of Hsu P-W, see Hsu P-W.Freehand three-dimensional ultrasound calibration, University of Cambridge; 2007 and Mercier see Mercier L, t, Lindseth F, et al.A review of calibration techniques for freehand3-D ultrasound systems.Ultrasound in medicine & biology.2005,31 (2): the people such as 143-165 conclude the work in this field.

Comeau RM is shown in by N line (also claiming Z line) model, Fenster A, Peters TM.Integrated MR and ultrasound imaging for improved image guidance in neurosurgery.Paper presented at:Medical Imaging1998:Image Processing1998; Comeau RM, Sadikot AF, Fenster A, et al.Intraoperative ultrasound for guidance and tissue shift correction in image-guided neurosurgery.Medical Physics.2000,27:787-800 is the most frequently used a kind of two-dimensional alignment model.This model adopts the thin-line-shaped target of pre-designed position, and every three fine rules form the shape of an English alphabet " N ".When several N shape line groups being put into a water tank of demarcating for ultrasonic probe, spatial relation between these N shape lines be just fixed up (seeing Fig. 2), we are by water tank and the fine rule as ultra sonic imaging target of fixing above, as a block mold, be called " water mould ".In practicality can by water mould manufacture and design guarantee wherein each group accurate locus of N line, or know that by measurements each organizes the accurate locus at N line place in water mould.When the fine rule having poured in the water mould of clear water or ultrasonic coupling liquid being carried out to imaging with ultrasonic probe, the N shape fine rule of three group will present three speck target E, F, G (seeing Fig. 3 a, Fig. 3 b) on ultrasonoscopy.From Fig. 2 and Fig. 3, by what be identified in that a certain group of speck target in ultra sonic imaging plane present, it is which the group N line on water mould, just can be according to the structural design scheme of water mould, know two straight line AB and CD corresponding coordinate figure (seeing Fig. 2) in water tank model coordinate systems of N line.Meanwhile, as shown in Figure 3, the three-dimensional space position that ABCD is ordered on water mould is known, and measures behind E, F, G position by ultrasonoscopy, can be from similar triangles Δ EBF and Δ GCF, and the ratio by limit EF with GF, obtains the ratio of limit BF and CF.By similar triangle theory, can be by the ratio of distance between any two of three speck E, F, G on ultrasonoscopy, obtain the oblique line point wherein Z coordinate in three dimensions, thereby obtain the three-dimensional space position that F is ordered, so just obtained on a ultrasonoscopy " n line index point".Many groups " N line index points " by imaging simultaneously on a ultrasound image plane, can simulate the three-dimensional plane of water tank model (from the principle of 3 definite planes, need to be greater than 3 N line index points), i.e. the plane of ultra sonic imaging.So, according to the three-dimensional coordinate of these N line index points in water mould, just can calculate this ultra sonic imaging plane with respect to the locus of water mould coordinate system.While gathering this ultrasonoscopy by inquiry again, the fixed locus sensor position in reference frame on ultrasonic probe, and in conjunction with water mould coordinate system the locus in reference frame, just can calculate the imaging plane of ultrasonoscopy of required demarcation and the spatial alternation being cemented between the locus sensor on probe and be related to T _s-usee Comeau RM, Fenster A, Peters TM.Integrated MR and ultrasound imaging for improved image guidance in neurosurgery.Paper presented at:Medical Imaging1998:Image Processing1998; Comeau RM, Sadikot AF, Fenster A, et al.Intraoperative ultrasound for guidance and tissue shift correction in image-guided neurosurgery.Medical Physics.2000,27:787-800.

Yet, by searching document and the existing scaling method of analysis, can see, effective imaging sound field in actual calibration process during due to ultra sonic imaging has certain thickness (as shown in Figure 3 a), simultaneously due to the poor problem of ultra sonic imaging lateral resolution, in resulting ultrasonoscopy, sign fine rule is laterally being diffused as long oval speck (as shown in Figure 3 b) in (X-direction), causes on ultrasonoscopy, determining exactly the position of fine rule.This not only has influence on the accuracy of choosing index point E, F, G position in ultrasonoscopy, and, because the calculating of F point three-dimensional coordinate under water mould coordinate system is that ratio by EF and FG length determines, the error of above-mentioned E, F, G position all can cause the calculating of F point three-dimensional coordinate inaccurate.And, in actual water mould design, in order to guarantee to cover abundant N shape line group in a width ultrasonoscopy, can not be too wide between AB line and CD line, this just causes the θ angle in Fig. 3 smaller, thereby makes to indicate on image that clicking the site error of getting is exaggerated in Z direction.To a plurality of independent extractions in imaging n line index point; their accuracy of three-dimensional position and coplanarities of these index points under water mould coordinate system have so just been lost; thereby cause the position of the imaging plane that calculates inaccurate, therefore the imaging plane of the resulting ultrasonoscopy uncertain error larger with the existence of the spatial alternation relation between the sensor of locus.

By analyzing the result of study of other researcher reports, can see in traditional N line model demarcation processing procedure, data discrete at the upper translational component of imaging plane thickness direction (with the perpendicular direction of ultrasound image plane) is all larger, and this illustrates that traditional N collimation method is lower to the estimated accuracy of this component.Pagoulatos etc. are shown in Pagoulatos N, Haynor DR, Kim Y.A fast calibration method for3-D tracking of ultrasound images using a spatial localizer.Ultrasound in medicine & biology.2001,27 (9): although do not emphasize this error in the research paper of 1219-1229, its data clearly demonstrate the existence of this problem.Hsu P-W is shown in Hsu PW, Prager RW, Gee AH, et al.Real-time freehand3D ultrasound calibration.Ultrasound in medicine & biology.2008, 34 (2): 239-251 and Chen etc. is shown in Chen T, Thurston A, Moghari M, et al.A real-time ultrasound calibration system with automatic accuracy control and incorporation of ultrasound section thickness.Proceedings of SPIE.2008, 6918:2A1-11 and Chen TK, Thurston AD, Ellis RE, et al.A real-time freehand ultrasound calibration system with automatic accuracy feedback and control.Ultrasound in medicine & biology.2009, 35 (1): the research of 79-93 is also inquired into this problem, and claim this problem for " thickness error" (elevational error).The main source of this error is that ultrasonic wave beam thickness and resolution problem causes surveyed fine rule on ultrasonoscopy, to be diffused as larger speck, causes the error to its location estimation, and this error is ubiquitous in traditional scaling method.

For overcoming this defect, see in the literature Zhu Liren, Li Wenjun, fourth brightness, Wang Guangzhi, a kind of calculation method that improves N line model probe stated accuracy, Chinese biological engineering in medicine journal, 31 (3): 337-343,2012, proposed a kind ofly by " coplanar constraint ", to improve n line index point extraction accuracy, thereby the method for raising probe imaging plane location position precision, experimental data shows that the method can improve precision property.But the method be by rigidly by extracted N line index o'clock to projection on a virtual plane manually simulating, because simulated plane itself just exists error, by projection process, be merely able to partly improve not coplanar the caused impact of each N line index point, but cannot guarantee the accuracy of this plane.

The people's such as Hsu P-W research advises introducing the image of a plurality of imaging planes, its processing method is after routine demarcation processing finishes, an additional independent step is refined the estimation of thickness direction translational movement is shown in to Hsu PW, Prager RW, Gee AH, et al.Real-time freehand3D ultrasound calibration.Ultrasound in medicine & biology.2008,34 (2): 239-251.They studies have shown that, under tradition is resolved framework, the information of introducing a plurality of synusia contributes to improve the repeatability of demarcating conversion.Yet, such method need to increase extra treatment step, and need to use improved N line model just can see Hsu PW, Prager RW, Gee AH, et al.Real-time freehand3D ultrasound calibration.Ultrasound in medicine & biology.2008,34 (2): 239-251; In addition, the method first supposed conventional demarcation afterwards other location components except imaging thickness direction be all accurately, and this hypothesis may not necessarily be set up in most practical applications.

Consider the deficiency of above-mentioned scaling method, the present invention probe imaging plane locus scaling method based on N line model is entered gone improvement.For traditional N line scaling method, relying on single width two-dimensional ultrasonic image completely extracts impact point position and resolves three-dimensional position put, make the larger problem of imaging thickness direction error, we propose to obtain many by locus sensor fixed on ultrasonic probe individual relative position is fixed and known imaging plane, by the extraction to impact point on these imaging planes, by locus sensing device information is rebuild the three-dimensional space position of the two-dimentional index point extracting in each plane, so by these objective points be imaged water mould in registration between fine rule and optimize and calculate, reach the accurate calibration of probe imaging plane locus.because this method can effectively be utilized high accuracy locus sensor and rebuilds the three-dimensional relative position information of two-dimentional index point, make to extract in traditional ultrasonoscopy the thickness error of index point, be distributed in each ultra sonic imaging plane, on spatial orientation, there is randomness, can reduce the deviation of the imaging plane locus demarcation causing due to loss coplanarity in traditional method.

Summary of the invention:

the object of the invention is to provide a kind of scaling method and handling process of new ultrasonic probe imaging plane position, thereby subtract few due to the intrinsic effective sound field thickness of ultra sonic imaging and lower the caused calibrated error of imaging resolution, improve the essence of demarcating degree.designed method and flow process go for following the tracks of the three-dimension ultrasonic imaging system of the locus of ultrasonic probe with optics or electromagnetic space position sensor, or utilize these locus sensors to follow the tracks of the system that ultrasonic probe carries out multi-mode 3 D medical visual fusion, become in a kind of manufacture at this class ultrasonic instrument and use the method and apparatus of accurately effective calibration system Probe Ultrasonic Searching plane of delineation locus.The present invention can overcome existing N line model timing signal, rely on merely on two-dimensional ultrasonic image and carry out to spreading also fuzzy impact point " thickness error " that position extraction causes, thereby make impact point depart from the shortcoming of coplanarity, make the measurement of ultra sonic imaging plane and demarcate more scientific and reasonable, simple and easy to do, designed method and flow process practical has clear and definite use value in ultrasonic instrument.

We notice under study for action, when adopting traditional N line model to demarcate processing, because the impact point on each width ultrasonoscopy is to extract separately and estimation, the direction of these index point error maximums all appears at the direction of vertical ultra sonic imaging plane, therefore be difficult to reduce this error by the method for multiple image error equalization, even as done in some documents, by gathering several two-dimensional ultrasonic images, demarcate, error is also always more outstanding in the direction ratio vertical with probe imaging plane.Therefore we wish, by processing method of the present invention and strategy, to overcome the orientation of this error, reduce the error of demarcating.

We notice again simultaneously, use N line model to carry out the process of ultra sonic imaging plane space location position, in fact be exactly by the position of impact point in a width of N line model or the two-dimensional slice image of a few width particular pose, rebuild and estimate the position of these sections in three dimensions and the process of attitude information.Therefore, the essence of problem of calibrating is carrying on two-dimentional ultrasound tomographic image the two dimension target point of getting, carries out the problem of registration with three-dimensional a collection of fine rule object.The core of this technology is the registration that on a kind of bidimensional image, impact point is cut into slices to threedimensional model.Theoretically, as all monoplanes bidimensional image arrives three-dimensional registration technique, in the direction perpendicular to bidimensional image, its error is always maximum, and by registration problems being expanded to three dimensions, likely reduce the error that specific direction is larger, thereby the synthesis precision of demarcating is increased.We consider one group of relative position is known a plurality of orientationsimpact point on two-dimensional ultrasonic image, by the information of visiting headroom position sensor, be mapped on several imaging planes of the non-parallel distribution in space, that is: the landmark space of some groups of two dimensions is distributed, revert to the spatial distribution of one group of three-dimensional, and then calculate to three-dimensional registration by three dimensions, index point is registrated on known three-dimensional fine rule model, thereby improves the precision of demarcating.Compare with traditional scaling method, the inventive method is likely passed through homogenize and the redistribution of error in all directions, obtains more accurate location position result.

According to this thinking, we have set up and a kind ofly by index point on several two-dimensional ultrasonic images and three-dimensional N line model, have carried out the processing policy of registration optimization, and have designed feasible processing method and handling process.Under the framework of this two dimension and three-dimensional registration problem, by the optics on ultrasonic probe or electromagnetic transducer, introduce the information of different angles and position, can significantly improve accuracy and the precision of final calibration result.According to our literature survey, find no researcher from thisly first carrying out the position reconstruction of 2 d-to-3 d, carry out again the problem of calibrating that the angle of three dimensions registration and optimization goes to set forth ultra sonic imaging plane, so processing policy and processing method that we propose have novelty and originality.

The method of inventing system form and data acquisition on, utilize N line model to demarcate needed hardware unit to be duplicate, to have kept the advantage of N line model demarcation with traditional.When gathering each ultra sonic imaging plane, synchronously gather space coordinates and the direction of locus sensor fixed on ultrasonic probe, so just can know the relative position relation between gathered some ultra sonic imaging planes.After having gathered needed one group of ultrasonoscopy, no longer according to traditional method, by extracting the two-dimentional index point on ultrasonoscopy, and the fine rule three dimensional local information designing by water mould, calculate one by one the three-dimensional position of N line index point in water mould coordinate system, then pass through the locus of the Fitting Calculation imaging plane, thereby calculate the transformation matrix between position sensor and imaging plane, but enter the designed following core processing flow process of the present invention.

(1) first in ultra sonic imaging plane, extract the position (two-dimensional coordinate in each ultra sonic imaging face) of fine rule index point, corresponding each ultrasonoscopy, extracts the two-dimensional position coordinate that all N shape sign fine rules present speck on ultrasonoscopy;

(2) by the relative position relation between each ultra sonic imaging plane being obtained by locus sensor when the acquiring ultrasound image, two-dimentional ultra sonic imaging plane is carried out to spatial organization, formation is some two-dimensional imaging planes of certain three-dimensional spatial distribution, and by the binding positions between these planes, form one group of imaging plane in three-dimensional spatial distribution.And then according to extracting the corresponding imaging plane of each index point two-dimensional position, the a certain width ultrasound image plane of take is reference, locus sensor information during by every width image acquisition, successively the two-dimentional index point extracting on each imaging plane is mapped in corresponding 3-D supersonic imaging plane, so just formed a collection of index point in three-dimensional spatial distribution, their correspondences the intersection point of imaging plane and fine rule when from different directions fine rule model is cut;

(3) carry out the initialization of ultra sonic imaging plane and water mould location matches position, utilize a certain width ultrasound image plane gathering, according to traditional N line scaling method, calculate this imaging plane with respect to the initial three-dimensional space position of water mould coordinate system, obtain this imaging plane to the initial space transformation matrix of water mould coordinate system, then by the three-dimensional symbol point on all imaging planes of previous step binding positions, according to this spatial alternation, be mapped to the three-dimensional space position that water mould is corresponding, form a collection of thick registration point (initial registration point) of three-dimensional spatial distribution;

(4) do not having error ideally, above-mentioned three-dimensional symbol point should accurately drop on corresponding fine rule.Owing to having certain site error between the two-dimensional ultrasonic image plane of utilizing above-mentioned a certain width ultrasonoscopy to calculate and three-dimensional water mould, will inevitably make also to have certain departing between index point on other each ultrasound image planes and the fine rule of distributed in three dimensions, our target is to wish to find a spatial alternation, make the index point subject to registration of above-mentioned all three-dimensional spatial distribution and the distance of corresponding fine rule on three-dimensional water mould reach minimum, thereby obtain an optimum transformation relation.Because these imaging planes are not parallel each other, can overcome the intrinsic maximum error due to index point of traditional method and all there is " thickness error " that close orientation causes.In processing, we adopt optimum solving method to carry out iterative processing, by changing locus and the orientation of whole three-dimensional symbol point group subject to registration, find this optimal transformation;

(5), according to above-mentioned consideration, setting is optimized object functionthat the index point subject to registration distributing in each ultra sonic imaging plane reaches minimum with the distance of corresponding fine rule on three-dimensional water mould; Space initial positionit is the initial transformation matrix that the 3rd step calculates above; adjustable variablesthat imaging plane position in water tank model is in translation and the rotation in 6DOF space; So can carry out optimized iterative computation based on above condition.By setting error threshold or the greatest iteration step number of optimizing, control the termination of iteration, obtain the most at last a spatial alternation matrix of having optimized, it has represented a plurality of imaging planes of gathering and an optimum matching of water mould, and the imaging plane and the conversion between sensor that obtain are optimum;

(6) during according to each ultrasonoscopy of collection, be cemented in the locus of the locus sensor on probe, one group of correspondence position of cutting into slices with optimum matching obtained above, calculates the imaging plane of ultrasonoscopy and is cemented in the optimal spatial transformation relation between the locus sensor on probe the target that namely the present invention will find.

The invention provides a kind of method of ultrasonic probe imaging plane locus stated accuracy and complete handling process of improving, can reduce because the intrinsic sound field thickness of ultra sonic imaging is large and low the caused calibrated error of resolution.The method that the present invention is designed and flow process, can become in the manufacture and use of this class ultrasonic instrument, accurately and efficiently the method for calibration system Probe Ultrasonic Searching plane of delineation locus.The present invention has overcome and utilizes conventional N line model timing signal on ultrasonoscopy, to extract the impact point position " thickness error " that cause of diffusion, the processing that makes measurement and demarcate is more scientific and reasonable, simple and easy to do, designed method and flow process practical, has good practical value.

Accompanying drawing explanation

Fig. 1 is the schematic diagram of relation between the formation of whole measurement calibration system and each subsystem coordinate system;

1 personal computer

2 three-dimensional localization measuring instruments

3 location transmitters

4 three-dimensional probe pens

5 ultrasonic probes and alignment sensor

6 water tank models

7 ultrasonic imaging devices

Fig. 2 is how much signals of N line model and the definition of model coordinate systems;

1 location pit

Fig. 3 is geometrical principle and sound field thickness and the image blurring schematic diagram that extracts error that causes that N line index point extracts;

Imaging sound field and N line that Fig. 3 a has thickness intersect signal

The diffusion hot spot of fine rule on the actual ultrasonoscopy of Fig. 3 b

Fig. 3 c sees the geometrical principle that solves F point three-dimensional coordinate from tank top

Fig. 4 adopts the inventive method to carry out the synchronous geometrical principle schematic diagram of demarcating of many imaging planes;

The geometrical relationship schematic diagram of the many imaging planes of Fig. 5 and the preliminary registration of water mould;

Fig. 6 adopts many ultrasound image planes to be optimized the flow chart of calculating;

The comparison diagram of the traditional method that Fig. 7 Practical Calculation obtains and the inventive method calibrated error;

Fig. 8 program flow diagram of the present invention.

The specific embodiment

For realizing calculation method proposed by the invention, and verify its effectiveness, our actual test macro of having built, and carried out performance test contrast experiment, on the ultrasonoscopy of same set of collection and locus sensing data, the calculation accuracy of traditional N line scaling method and the inventive method and the repeatable error of demarcation have been compared, and error is with the situation of the ultrasonoscopy number decline that participates in demarcating, experiment has confirmed that method of the present invention is when having several ultrasonoscopys, can make quickly calibrated error decline, stated accuracy when having larger " thickness error " is better, thereby can make the calibration process of probe more easily with more reliable.In Fig. 8, provide program flow diagram of the present invention, the example of the test comparison of specific implementation is as described in following steps:

The 1st step, builds a three-dimensional localization measuring system of demarcating for ultrasonic probe.The three-dimensional localization measuring system that we demarcate for ultrasonic probe, comprises four capital equipments: three-dimensional localization measuring instrument, for demarcating water tank model, Medical Ultrasonic Imaging System and the personal computer of imaging.Three-dimensional localization measuring instrument wherein, water tank model, ultrasonic image-forming system are jointly for the ultrasonic probe of being demarcated is measured, and personal computer is for collection and the date processing of measurement data.Three-dimensional localization measuring instrument comprises again following four critical pieces: be fixed on the control box that 6DOF locus sensor, high accuracy three-dimensional on ultrasonic probe are located probe pen, used wireless launcher and transmit for measuring system Control & data acquisition as the location of georeferencing coordinate system.Ultrasonic image-forming system comprises ultrasonic probe and the imaging system main frame two large divisions who is demarcated, for generation of the ultrasonoscopy of water tank model to be collected, wherein demarcated fixed above-mentioned 6DOF locus sensor on ultrasonic probe.Picture catching card has been installed in personal computer, for gathering from the ultrasonoscopy of described ultrasonic image-forming system output, meanwhile, personal computer is exported from described three-dimensional localization measuring instrument control box by USB interface collection the locator data of 6DOF locus sensor and the locator data at three-dimensional localization probe pen tip.The formation of the three-dimensional localization measuring system of demarcating for ultrasonic probe as shown in Figure 1;

The 2nd step: make N line model.Utilize according to the following steps poly (methyl methacrylate) plate and thin nylon wire to make a N line peg model:

(1) first, with poly (methyl methacrylate) plate, make a rectangular water tank 1, it is of a size of 240*180*120mm ³, tank wall thickness is 5mm.According to N line number and the space layout of design, be of a size of exact position corresponding on former and later two relative tank walls of 240*180mm, having got out coaxially a collection of diameter is the threading aperture of 0.2mm, for the thin nylon wire that is 0.08mm through diameter, forms a collection of N shape line pair in water tank.Meanwhile, four angular distance casing outer rim 5mm places on the tank wall of front and back, respectively get out accurate taper location, 4 positions pit.The degree of depth of the upper conical location of the water tank pit that we manufacture is 0.2mm, and cone-apex angle is 90 °.The rectangular distribution in tank wall plane of four location pits, its horizontal direction distance is 230mm, vertical direction distance is 170mm.The relative position of location groove position and above-mentioned threading aperture by digital control processing to guarantee its accuracy, for measure water case and fine rule in the position of space positioning system,

(2) after water tank completes, as the partial enlarged drawing of Fig. 2 and Fig. 3 a is illustrated, the thin nylon wire that is 0.08mm through diameter in certain threading aperture A on tank wall, and guide to coaxial aperture B corresponding on side box wall, after nylon wire is strained and fixed, from aperture B, go back to the identical adjacent aperture C of level height same sidewall paneling, form the oblique line of N shape; By going back to the threading aperture D offside wallboard from C again after nylon wire tension, nylon wire is strained and fixed, make it to form a N shape line pair in water tank.So repeat, can in water tank, build some groups of horizontally disposed N shape lines pair, form the object that ultra sonic imaging is demarcated.In Fig. 1, illustrated three layers of horizontally disposed N line, maximum layer comprises 3 N shapes, and all the other are two-layer respectively comprises 5 N shapes.Three layers of N line height in water tank model coordinate systems is respectively 65mm, 95mm and 125mm, and the N line in the horizontal direction distance of two parallel edges is 30mm, so the θ angle in Fig. 3 c is about 15 °.The position of the threading aperture of each N line in water tank all accurately guarantees by digital control processing.In actual use, for improving precision and guaranteeing, in imaging viewing field, there is abundant N line group, can be according to visual field size and the ultrasonoscopy resolution of ultrasonic probe, 6～8 layers of N line of choice arrangement, the N demand pairs of every layer also can be arranged 3～6, thereby make ultrasonoscopy can collect more fine rule target, and make target be full of as far as possible the visual field of ultra sonic imaging.After the whole threading work of water tank completes, utilize silica gel to seal all threading apertures, make water tank water-tight.Then in water tank, pour into pure water, and more than making it to exceed top fine rule 30mm, by standing 1 day of water tank, the bubble in water is overflowed.Because all fine rules are all immersed in the water, can be probeed into the ultrasonic imaging probe imaging of the water surface.So just completed the making of the water tank model for demarcating;

The 3rd step, set up each relative coordinate system of described three-dimensional localization measuring system of demarcating for ultrasonic probe and the transformation relation between coordinate system:

As shown in Figure 1, in calibration experiment, relate to four coordinate systems.Coordinate system comprises the defined reference frame of high-precision three-dimensional positioning measurment system, uses X _wy _wz _wrepresent, be located at described location with on wireless launcher, zero is at described discharger center; Water tank model coordinate systems, uses X _my _mz _mrepresent X _mwater tank length direction, Z _mwater tank width, Y _mbe water tank short transverse, the initial point of coordinate system is located on the location pit in the water tank front face upper left corner, and the direction of each coordinate axes is followed right-hand rule; Locus sensor coordinate system, uses X _sy _sz _srepresent, be located on the sensor of described 6DOF locus, zero is at the center of described locus sensor; Ultra sonic imaging plane coordinate system, uses X _uy _urepresent, be located on ultrasonoscopy, zero is located at ultrasonoscopy mid point topmost, X _udirection is image level direction, Y _udirection is ultra sonic imaging depth direction.For setting up the transformation relation between each coordinate system, set the spatial alternation matrix T of 4*4 _w-mfrom described water tank model coordinate systems X _my _mz _mto described reference frame X _wy _wz _wspatial alternation matrix, T _w-sfrom described locus sensor coordinate system X _sy _sz _sto described reference frame X _wy _wz _wspatial alternation matrix, T _s-ufrom described ultra sonic imaging plane coordinate system X _uy _ulocus sensor coordinate system X to described location use _sy _sz _sspatial alternation matrix, T _u-mfrom described water tank model coordinate systems X _my _mz _mto described ultra sonic imaging plane coordinate system X _uy _uspatial alternation matrix;

The 4th step, demarcates the locus of water mould in reference frame

We,, according to the position of location pit, have determined water tank model coordinate systems X _my _mz _m, be about to the location pit conduct in the water tank antetheca upper left corner water tank model coordinate systems initial point, the straight line that connects location, angle pit on left and right two is defined as X _maxle, the straight line that connects two, left side location pit downward direction is defined as Y _maxle, is defined as Z by the direction of water tank wall location pit before and after connecting _maxle.Like this in water tank model coordinate systems each N shape fine rule group all with X _m-Z _mplane parallel, the straight line of each N shape all with Z _maxle is parallel, and its position coordinates in water tank model coordinate systems is all known, its concrete layout as schematically shown in Figure 2.In demarcation, first to, by measuring, mark the transformation matrix T that is tied to reference frame from water tank model coordinate _w-m.We adopt the high-precision three-dimensional electromagnetic positioning system of Canadian NDI company to measure, and this system comprises high-precision three-dimensional location probe pen that a 6DOF locus sensor being attached on ultrasonic probe, most advanced and sophisticated positioning precision are 0.2mm and with the navigation system discharger of reference location.As shown in Figure 1, regioselective system discharger is fixing reference framex _wy _wz _waccording to the operation instruction of navigation system, water tank is placed in the spatial dimension of the most applicable location discharger detection, adopt the tip of the high-precision three-dimensional location probe pen of navigation system outfit, survey one by one the three-dimensional space position of each location pit of arranging in water tank front and back walls, and by its three-dimensional space position P in reference frame of computer journal _wi=(x _wi, y _wi, z _wi), i=1 wherein, 2 ..., 8, for locating the sequence number of pit on water tank model.From the definition of water tank model coordinate systems, can obtain the coordinate P of these location pits in water tank model coordinate systems _mi=(x _mi, y _mi, z _mi) be respectively: P _m1=(0,0,0), P _m2=(230,0,0), P _m3=(0,170,0), P _m4=(230,170,0), P _m5=(0,0,120), P _m6=(230,0,120), P _m7=(0,170,120), P _m8=(230,170,120), its unit is mm,

Note water tank model coordinate systems X _my _mz _mto reference frame X _wy _wz _wspatial alternation matrix T _w-mfor:

$T_{w-n}=\left[\begin{array}{cc}R& 0\\ t& 1\end{array}\right]=\left[\begin{array}{cccc}{r}_{00}& r_{01}& r_{02}& 0\\ r_{10}& r_{11}& r_{12}& 0\\ r_{20}& r_{21}& r_{22}& 0\\ t_x& t_y& t_z& 1\end{array}\right]---\left(1\right)$

Wherein the 3*3 piecemeal R in the upper left corner is rotation transformation matrix, and the 1*3 piecemeal t in the lower left corner is translation transformation.Locate the three-dimensional space position P of pit in reference frame for 8 _wi=( x _wi , y _wi , z _wi ) and its respective coordinates P in water tank model coordinate systems _mi=(x _mi, y _mi, z _mi), meet following spatial alternation equation:

$\left[\begin{array}{cccc}{x}_{w1}& y_{w1}& z_{w1}& 1\\ x_{w2}& y_{w2}& z_{w2}& 1\\ x_{w3}& y_{w3}& z_{w3}& 1\\ x_{w4}& y_{w4}& z_{w4}& 1\\ x_{w5}& y_{w5}& z_{w5}& 1\\ x_{w6}& y_{w6}& z_{w6}& 1\\ x_{w7}& y_{w7}& z_{w7}& 1\\ x_{w8}& y_{w8}& z_{w8}& 1\end{array}\right]=\left[\begin{array}{cccc}{x}_{m1}& y_{m1}& z_{m1}& 1\\ x_{m2}& y_{m2}& z_{m2}& 1\\ x_{m3}& y_{m3}& z_{m3}& 1\\ x_{m4}& y_{m4}& z_{m4}& 1\\ x_{m5}& y_{m5}& z_{m5}& 1\\ x_{m6}& y_{m6}& z_{m6}& 1\\ x_{m7}& y_{m7}& z_{m7}& 1\\ x_{m8}& y_{m8}& z_{m8}& 1\end{array}\right]\·T_{w-m}---\left(2\right)$

Therefore, can solve from water tank model coordinate and be tied to the spatial alternation matrix T between reference frame from equation group (2) _w-m.

The 5th step, several ultrasonoscopys to the N line collection diverse location in water mould

As shown in fig. 1, before experiment, first on ultrasonic probe, fixes a 6DOF position sensor, make it and ultrasonic probe between firmly link irremovable.Utilization is carried out imaging with the ultrasonic probe of position sensor to the fine rule in water tank model, selects the artifact-free image space of imaging clearly, and stable firmly ultrasonic probe, carries out the collection of image and corresponding locus.Ultrasonoscopy is wherein exported by the image output interface of medical supersonic instrument, by the video frequency collection card (adopting OSPREY100 capure card in experiment) being arranged in PC computer, caught, the position and attitude matrix of locus sensor in reference frame can directly be read by high-precision three-dimensional positioning measurment system, by USB interface, passes in computer.The ultrasonoscopy gathering is successively inputted after computer, is stored as Img (j), j=1, and 2 ..., J, the ultrasonoscopy number of J for gathering, the spatial alternation matrix T to reference frame of simultaneously storing the locus sensor that this ultrasonoscopy is corresponding _w-s(j).Diverse location in water mould, by different probe attitudes, repeated acquisition is stored the ultrasonoscopy of some width Different Plane, and stores the spatial alternation matrix T of corresponding locus sensor _w-s(j).The ultrasonoscopy of the above diverse location of 7 width and corresponding locator data in our experiment, have all been gathered at every turn.

The 6th step, extracts index point position and the position of calculating this ultra sonic imaging plane on ultrasonoscopy

First define ultra sonic imaging plane coordinate system X _uy _u, as shown in Figure 1.Under this coordinate system by manual on every width ultrasonoscopy to the clicking of N shape fine rule index point speck, extract each sign speck at ultrasonoscopy coordinate system X _uy _uin two-dimensional coordinate.Particularly, at ultrasonoscopy Img (j), go up according to first along X _mincrease progressively, then press Y _mthe order increasing progressively, the manual two-dimensional coordinate position that clicks all N lines in the ultrasonoscopy visual field and the crossing formed triplet speck of ultrasound image plane center successively, and the order being clicked according to every group of N line is write down its sequence number k.Three limits of one group of N line of choosing speck on ultrasonoscopy of setting up an office is respectively E (j, k), F (j, k), G (j, k), k=1 wherein, and 2 ..., K _j, K _jfor the sum at the upper N line index point obtaining of this width ultrasonoscopy Img (j).The two-dimensional coordinate of the speck that note clicks is: $X_u^{\left(E\right)}(j,k)=[x_u^{\left(E\right)}(j,k),y_u^{\left(E\right)}(j,k)],$ $X_u^{\left(F\right)}(j,k)=[x_u^{\left(F\right)}(j,k),y_u^{\left(F\right)}(j,k)],$ $X_u^{\left(G\right)}(j,k)=[x_u^{\left(G\right)}(j,k),y_u^{\left(G\right)}(j,k)],$ Can obtain the distance between three clicked points according to image resolution ratio | EF| _jkwith | EG| _jk:

${\left|\mathrm{EF}\right|}_{\mathrm{jk}}=\sqrt{{(x_u^{\left(E\right)}(j,k)-x_u^{\left(F\right)}(j,k))}^2+{(y_u^{\left(E\right)}(j,k)-y_u^{\left(F\right)}(j,k))}^2};---\left(3\right)$

${\left|\mathrm{EG}\right|}_{\mathrm{jk}}=\sqrt{{(x_u^{\left(E\right)}(j,k)-x_u^{\left(G\right)}(j,k))}^2+{(y_u^{\left(E\right)}(j,k)-y_u^{\left(G\right)}(j,k))}^2}.---\left(4\right)$

According to the method described above the N line index point position on the ultrasonoscopy of all collections is extracted, and store according to sequence number.

According to columns and the number of plies at handled N line place, obtain the through wires hole A of k group _kb _kc _kd _kx on water mould _mwith Y _mcoordinate figure:

$X_m^{\left(A\right)}(j,k)=[x_m^{\left(A\right)}(j,k),y_m^{\left(A\right)}(j,k),z_m^{\left(A\right)}(j,k)]---\left(5\right)$

$X_m^{\left(B\right)}(j,k)=[x_m^{\left(B\right)}(j,k),y_m^{\left(B\right)}(j,k),z_m^{\left(B\right)}(j,k)]---\left(6\right)$

$X_m^{\left(C\right)}(j,k)=[x_m^{\left(C\right)}(j,k),y_m^{\left(C\right)}(j,k),z_m^{\left(C\right)}(j,k)]---\left(7\right)$

$X_m^{\left(D\right)}(j,k)=[x_m^{\left(D\right)}(j,k),y_m^{\left(D\right)}(j,k),z_m^{\left(D\right)}(j,k)]---\left(8\right)$

Wherein, A hole is coaxial with B hole, and C hole is coaxial with D hole, that is: $x_m^{\left(A\right)}(j,k)=x_m^{\left(B\right)}(j,k),$ $y_m^{\left(A\right)}(j,k)=y_m^{\left(B\right)}(j,k),$ $x_m^{\left(C\right)}(j,k)=x_m^{\left(D\right)}(j,k),$ $y_m^{\left(C\right)}(j,k)=y_m^{\left(D\right)}(j,k),$ And contour in each Kong Shui mould coordinate system in N line group, that is: $y_m^{\left(A\right)}(j,k)=y_m^{\left(B\right)}(j,k)=y_m^{\left(C\right)}(j,k)=y_m^{\left(D\right)}(j,k).$ By ultrasonoscopy, measure after distance between E (j, k), F (j, k), G (j, k) (referring to Fig. 3 b, for for purpose of brevity, omitted the subscript of each symbol here, lower with) again, can be from similar triangles EBF and GCF, by | EF| _jkwith | GF| _jkratio obtain | BF| _jkwith | CF| _jkratio, thereby obtain the Z of F point in water tank model coordinate systems _mcoordinate (seeing Fig. 3 c), thus obtain that F orders coordinate, like this, a N line that is k to sequence number in ultrasonoscopy Img (j), to (ternary speck), can obtain the three-dimensional coordinate of a N line index point F (j, k) in water tank model coordinate systems.Suppose that the three-dimensional coordinate that A is ordered is under the water tank model coordinate systems shown in Fig. 2:

$X_m^{\left(A\right)}=[x_m^{\left(A\right)}(j,k),y_m^{\left(A\right)}(j,k),z_m^{\left(A\right)}(j,k)],$ The three-dimensional coordinate of F point in water tank model coordinate systems

$X_m^{\left(F\right)}=[x_m^{\left(F\right)}(j,k),y_m^{\left(F\right)}(j,k),z_m^{\left(F\right)}(j,k)]$ Can be expressed as:

$\left\{\begin{array}{c}{x}_m^{\left(F\right)}(j,k)=x_m^{\left(A\right)}(j,k)+\frac{{\left|\mathrm{EF}\right|}_{\mathrm{jk}}}{{\left|\mathrm{EG}\right|}_{\mathrm{jk}}}\·{\left|\mathrm{BD}\right|}_{\mathrm{jk}}\\ y_m^{\left(F\right)}(j,k)=y_m^{\left(A\right)}(j,k)\\ z_m^{\left(F\right)}(j,k)=z_m^{\left(A\right)}(j,k)+\frac{{\left|\mathrm{EF}\right|}_{\mathrm{jk}}}{{\left|\mathrm{EG}\right|}_{\mathrm{jk}}}\·{\left|\mathrm{AB}\right|}_{\mathrm{jk}}\end{array}\right.---\left(9\right)$

Wherein, j is the sequence number of the ultrasonoscopy Img (j) for demarcating, k=1, and 2 ..., K _j, K _jit is the number of the index point that extracts on this width image.And: ${\left|\mathrm{BD}\right|}_{\mathrm{jk}}=\sqrt{{(x_m^{\left(D\right)}(j,k)-x_m^{\left(B\right)}(j,k))}^2},$ ${\left|\mathrm{AB}\right|}_{\mathrm{jk}}=\sqrt{{(z_m^{\left(B\right)}(j,k)-z_m^{\left(A\right)}(j,k))}^2}.$ In above formula, at X _mwith Z _mthe coordinate figure of direction is according to the similar triangles relation of △ BEF and △ CGF, and at the location positioning of the datum mark (being A point) that water is molded to be defined while making herein, and F point is at Y _mthe coordinate figure of direction is obtained by the position of this N line place layer.

All N line groups that can obtain are all processed like this to the two-dimentional N line index point coordinates of each sequence number k that imaging obtains in image I mg (j) there is a three-dimensional coordinate point under model coordinate systems the point of formation order correspondence is right with it.Ideally, these N line index points, in water tank three-dimensional system of coordinate, will form a plane.We can, to Img (j), utilize equation (4) to solve the coordinate Mapping being tied between this ultrasound image plane coordinate system from water tank model coordinate and be related to T _u-m(j):

$\left[\begin{array}{cccc}{x}_u^{\left(F\right)}(j,1)& y_u^{\left(F\right)}(j,1)& 0& 1\\ x_u^{\left(F\right)}(j,2)& y_u^{\left(F\right)}(j,2)& 0& 1\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ x_u^{\left(F\right)}(j,K_j)& y_u^{\left(F\right)}(j,K_j)& 0& 1\end{array}\right]=\left[\begin{array}{cccc}{x}_m^{\left(F\right)}(j,1)& y_m^{\left(F\right)}(j,1)& z_m^{\left(F\right)}(j,1)& 1\\ x_m^{\left(F\right)}(j,2)& y_m^{\left(F\right)}(j,2)& z_m^{\left(F\right)}(j,2)& 1\\ .& .& .& \\ .& .& .& \\ .& .& .& \\ x_m^{\left(F\right)}(j,K_j)& y_m^{\left(F\right)}(j,K_j)& z_m^{\left(F\right)}(j,K_j)& 1\end{array}\right]\·T_{u-m}\left(j\right)---\left(10\right)$

Or write as:

$X_u^{2D}(j,k)=X_m(j,k)\·T_{u-m}\left(j\right)---\left(11\right)$

J=1 wherein, 2 ..., J, J is the number of gathered ultrasonoscopy, k=1,2 ..., K _j, K _jit is the number at the upper N line index point extracting of this image I mg (j).

The 7th step, is mapped to the N line index point on each width two-dimensional ultrasonic image in the multiframe imaging plane of three-dimensional spatial distribution

As above-mentioned, when gathering each width ultrasonoscopy, ultrasonic probe is certain distribution at three dimensions, has formed fine rule on water mould in the cutting of different spatial.We can be according to the position T that gathers each width ultrasonoscopy time space position sensor _w-s(j), in reference frame, several two-dimentional ultra sonic imaging planes are got up in three dimensions virtual portfolio, form the some width imaging planes that are certain three-dimensional spatial distribution.In upper any the two-dimentional index point position of extracting of each ultra sonic imaging planar I mg (j) it is the intersection point of this ultrasound image plane specific fine rule in particular spatial location and water mould, therefore from ultra sonic imaging angle, these intersection points are the two-dimentional index points in the virtual image plane dropping on separately, from water modular angle degree, be that these fine rules are cut by the virtual image plane of this collection of spatial distribution and a collection of plane distributing at space three-dimensional that forms and the intersection point of line.Our target is just to locate: where these cutting dough sheets are during in, and the index point extracting from the two-dimensional ultrasonic image of several known location, can have best mating with the fine rule in water mould.In Practical Calculation, do not need really to obtain the locus of these virtual image planes, and only need to be when gathering each ultrasonoscopy Img (j) the position T of position sensor _w-s(j), by the two-dimentional index point position of extracting on each imaging plane by the position and attitude that gathers this image time space position sensor, be mapped in three dimensions, and combine, just form a collection of registration point in three dimensions, this batch of three dimensions point can with water mould in fine rule carry out three-dimensional registration, the algorithm of specific implementation is shown in the 8th step.

With the example that is demarcated as of sector scanning ultrasonic probe, by the positional information T of locus sensor _w-s(j) geometrical concept that combines each virtual image plane as shown in Figure 4, in figure 5 fan-shaped be 5 virtual image planes, on every fan-shaped two-dimensional ultrasonic image, can extract some index points.

Therefore the upper index point of each ultrasonoscopy Img (j) position that, the 6th step is extracted can be when gathering each width ultrasonoscopy Img (j) the evolution matrix T of position sensor _w-s(j), successively the two-dimentional index point position on each imaging plane is mapped in three-dimensional j corresponding virtual image plane, so just formed a collection of index point in three-dimensional spatial distribution, their correspondences the intersection point of a plurality of imaging planes and fine rule while from different directions fine rule model being carried out to imaging.

For the purpose of standard, the index point that our unification is above extracted ultra sonic imaging planar I mg (j) two-dimensional coordinate is written as: l=1 wherein, 2 ..., 3 * K _j, K _jit is the number of the N line index point that extracts on image I mg (j); Will the three-dimensional coordinate being mapped on each imaging plane of three dimensions is designated as $X_u^{3D}(j,l)=(x_u^{3D}(j,l),y_u^{3D}(j,l),z_u^{3D}(j,l)),$ :

$X_u^{3D}(j,l)=X_u^{2D}(j,l)\·T_{w-s}\left(j\right)---\left(12\right)$

Wherein, T _w-s(j) be the evolution matrix of the position sensor of corresponding Img (j) imaging plane that obtains from described personal computer from three-dimensional localization measuring instrument, j=1,2 ..., J, J is the ultrasonoscopy number gathering, l=1,2 ..., 3 * K _j, K _jit is the number of the N line index point that extracts on image I mg (j).Being mapped to spatial point in each virtual ultra sonic imaging plane is illustrated as the round dot on each covering of the fan in Fig. 4.

The 8th step: carry out the initialization of space matched position

Utilize the equation (10) of the 6th step can be according to the three-dimensional coordinate of each N line index point position of extracting from a certain width ultrasound image plane Img (j) and the two-dimensional coordinate of corresponding speck on this ultrasonoscopy solve the coordinate Mapping being tied between this ultrasound image plane coordinate system from water tank model coordinate and be related to T _u-m(j).This conversion has been set up ultra sonic imaging plane index point to the transformation relation fine rule in water mould and imaging plane intersection point.Due in the 7th step by all index points that extract the orientation during according to its imaging carried out three dimensions restructuring, obtain the point subject to registration of a collection of three-dimensional spatial distribution when we select a width ultrasonoscopy, obtain this image after the transformation relation of water intermode, can utilize this conversion by the index point in all three-dimensional imagings space be mapped to initial position more approaching with corresponding fine rule in water mould.Suppose to select n width ultrasonoscopy Img (n) to initialize registration, can obtain transformation matrix T according to equation (13) _u-m(n), that is:

$X_u^{\left(F\right)}(n,k)=X_m^{\left(F\right)}(n,k)\·T_{u-m}\left(n\right)---\left(13\right)$

Then, by all index points in other each virtual image plane obtaining (j=1,2 ..., n-1, n+1 ... J, l=1,2 ..., 3 * K _j), along with planar I mg (n) carries out spatial translation and rotation, transform in water tank model coordinate systems, so just formed a collection of three-dimensional coordinate point subject to registration in model coordinate systems

$X_m^{3D}(j,k)=X_u^{3D}(j,k)\·T_{w-s}^{-1}\left(n\right)\·T_{u-m}^{-1}\left(n\right);j=\mathrm{1,2},...,J---\left(14\right)$

$=X_u^{2D}(j,k)\·T_{w-s}\left(j\right)\·T_{w-s}^{-1}\left(n\right)\·T_{u-m}^{-1}\left(n\right)$

So far, the three dimensions index point in constructed virtual ultra sonic imaging plane, has transformed in water tank model coordinate systems, and together with fine rule preliminary and on water mould is registered in, index point will be very approaching with the corresponding fine rule in water mould, as shown in schematic diagram 5.

The 9th step: set object function and the initial value optimized, and be optimized calculating

By the processing of the 8th step, all three-dimensional coordinate points subject to registration have all been mapped in water mould coordinate system, ideally these points should overlap with fine rule corresponding on water mould (apart from being zero), but owing to extracting index point on ultrasonoscopy location error, therefore to corresponding fine rule, also may there is certain distance.We choose optimization aimfunction is the index point of all three-dimensional spatial distribution with on three-dimensional water mould, the average distance of fine rule is minimum recently.Notice that the N shape line in the design of water mould is to be distributed on the horizontal plane of differing heights level, therefore the y of every layer of fine rule _mcoordinate is fixed, and for the parallel edges of N shape line, its x _mcoordinate is also fixed, and supposes total R layer N shape line in water mould, and every layer has S bar parallel lines, can be to these fine rule serial numbers, and the linear equation of these parallel fine rules can be written as:

$\left\{\begin{array}{c}{x}_m\left(l\right)=L_x(r,s);\\ y_m\left(l\right)=L_y(r,s);\end{array}\right.$ Wherein $\left\{\begin{array}{c}r=\mathrm{1,2},...,R\\ s=\mathrm{1,2},...,S\end{array}\right.---\left(15\right)$

L wherein _x(r, s) and L _y(r, s) be respectively define in water mould when design the coordinate of the thin parallel lines of r layer s bar in water tank model coordinate systems, be given constant, l is the sequence number of each fine rule.

And the hypotenuse of N shape line is to be distributed in the horizontal plane of water mould coordinate system, can define its three dimensions linear equation by the get lines crossed position of aperture (being also the aperture of wearing parallel lines) of the starting point of oblique line and terminal simultaneously:

$\left\{\begin{array}{c}\frac{x-x_m\left(l\right)}{x_m(l+1)-x_m\left(l\right)}=\frac{z-z_m\left(l\right)}{z_m(l+1)-z_m\left(l\right)};\\ y_m\left(l\right)=L_y(r,s);\end{array}\right.---\left(16\right)$

Therefore, in water mould the locus of all fine rules all by strict difinition.For trying to achieve space indicate point to the distance of its line correspondence, after above-mentioned initialization process, can certain index point of direct solution to the distance of all straight lines, for parallel fine rule, only need by in x component and the respective components of y component and fine rule subtract each other, and according to Pythagorean theorem computed range, for oblique line, due to they horizontal distribution, can first solve x, distance on z component, more synthetic with y component.Obtaining after the distance of all fine rules, choose one of them minimum distance D (j, l) as spatial point distance with nearest fine rule corresponding on three-dimensional water mould, can obtain index point like this average distance with fine rule nearest on three-dimensional water mould:

$D_{\mathrm{avg}}=\frac{1}{J\·K_j}{\mathrm{\Σ}}_{l=1}^{3\×K_j}D(j,l)---\left(17\right)$

Wherein, the ultrasonoscopy number of J for participating in calculating, K _jfor the index point number that can extract in j width ultrasonoscopy.

Above-mentioned optimization is calculated initial space positiont _u-m(n) be that the 8th step is utilized imaging plane that certain width ultrasonoscopy Img (n) the calculates locus in water mould coordinate system above.Can adjust T on this basis _u-mvariable be optimized iterative computation, by optimization, calculate and make average distance D _avgminimization. optimised variableit is the spatial alternation matrix T between ultra sonic imaging coordinate system and water mould coordinate system _u-m(comprising three translations and three rotations, altogether 6DOF).By setting average distance D _avgerror threshold (being taken as 2mm) and greatest iteration step number (500 step), control the end of Optimized Iterative process, finally obtained the transformation matrix of an optimization, it has represented an optimum matching conversion of fine rule target in many imaging planes of gathering and water mould

In this example we adopted Matlab software with optimization toolbox in the optimization computational tool of sequential quadratic programming (sequential quadratic programming algorithm, SQP) carry out optimizing.Can certainly adopt other mathematical calculation instrument to be optimized calculating.

The 10th step: calculate the optimal spatial conversion between imaging plane and sensor during according to collection each ultrasonoscopy Img (j), be cemented in the locus T of the locus sensor on probe _w-s(j) spatial alternation of the best section correspondence position, obtaining with the 9th step by the relation between defined each coordinate system in Fig. 1, known: to the point set in ultra sonic imaging coordinate system or can pass through spatial alternation transform in water mould coordinate system, obtain corresponding point set in Practical Calculation we by formula (12) in the 7th step by the index point extracting on each ultrasonoscopy transform in three dimensions virtual image plane, obtain point set for solving optimum ultrasonoscopy to the conversion between sensor the flow process of calculating is as follows.First the index point extracting on ultrasonoscopy is transformed to virtual image plane, obtains three-dimensional symbol point position:

$X_u^{3D}(j,l)=X_u^{2D}(j,l)\·T_{w-s}\left(j\right);j=\mathrm{1,2},..,J;l=\mathrm{1,2},...,3\×K_J---\left(18\right)$

Again by the point set in virtual image plane transform in water mould coordinate system:

$X_m^{3D}(j,l)=X_u^{3D}(j,l)\·{\tilde{T}}_{u-m}^{-1};i=\mathrm{1,2},..,J;l=\mathrm{1,2},...,3\×K_j---\left(19\right)$

Then, then by the point set in water mould coordinate system transform in sensor coordinate system:

$X_s^{3D}(j,l)=X_m^{3D}(j,l)\·T_{w-m}\·T_{w-s}^{-1}\left(j\right);j=\mathrm{1,2},..,J;l=\mathrm{1,2},...,3\×K_j---\left(20\right)$

Wherein, T _w-s(j) be the position of position sensor in reference frame while gathering ultrasonoscopy Img (j), can read from navigation system, and T _w-mbe the 4th pacing amount and solve the water mould coordinate obtaining and be tied to the spatial alternation matrix between reference frame.

So just obtained ultrasonoscopy coordinate system and a collection of paired index point data in corresponding position sensor coordinate system: with j=1 wherein, 2 ..., J, J is the number of gathered ultrasonoscopy, l=1,2 ..., 3 * K _j, K _jit is the number of correspondence image Img (j) the N line index point that can extract.Utilize these data, can calculate ultrasonoscopy coordinate to be solved according to equation (21) and be tied to the global optimum's conversion between the locus sensor on ultrasonic probe

$X_s^{3D}(j,l)=X_u^{3D}(j,l)\·{\tilde{T}}_{s-u}---\left(21\right)$

In fact, a flow process that also can be other, obtains same result.We can, according to the order that gathers image, utilize formula (22) to come to calculate ultrasonoscopy to the optimal transformation between sensor by width to arbitrary width image I mg (j):

$X_s^{3D}(j,k)=X_u^{3D}(j,k)\·{\tilde{T}}_{s-u}\left(j\right);k=\mathrm{1,2},...,K_J---\left(22\right)$

The benefit of doing is like this from equation (11), and (12) and (14) can direct derivation go out ultrasonoscopy to be solved and to the optimal transformation between sensor are:

${\tilde{T}}_{s-u}\left(j\right)={\tilde{T}}_{u-m}^{-1}\·T_{w-m}\·T_{w-s}^{-1}\left(j\right)---\left(23\right)$

Finally, then the optimal spatial transformation matrix that several ultrasonoscopys are calibrated averages, and obtains optimum calibration result

${\tilde{T}}_{s-u}=\frac{1}{J}{\mathrm{\Σ}}_{j=1}^J{\tilde{T}}_{s-u}\left(j\right)---\left(24\right)$

Wherein J is the ultrasonoscopy number of collection.Utilize equation (21) or formula (24) to calculate be exactly that the target ultrasonoscopy coordinate that will find of the present invention is tied to the optimal transformation between the locus sensor coordinate system on ultrasonic probe.

In sum, whole handling process of the present invention as shown in Figure 6.

For the inventive method and traditional N line method are contrasted, we extract and have calculated each " N line index point " three-dimensional space position in water tank model coordinate systems on each width ultrasonoscopy obtaining also according to N line scaling method and by these position the Fitting Calculation, obtain the position of the ultra sonic imaging plane that gathers.By the space position solution of corresponding position sensor, according to traditional N line scaling method, the spatial alternation calculating between ultra sonic imaging planar I mg (j) and position sensor is related to T simultaneously _s-u(j).

For the spatial alternation solving between ultra sonic imaging planar I mg (j) and position sensor is related to T _s-u(j), only need to find a collection of in sensor coordinate system with ultra sonic imaging plane coordinate system in paired spatial point, demarcate.We remember the N line index point of extraction and the three-dimensional coordinate of the intersection point F of Img (j) imaging plane under water tank model coordinate systems by the relation between defined each coordinate system in Fig. 1 these spatial point can, by following equation (25) transformed mappings in the sensor reference coordinate system of position, obtain the three-dimensional coordinate under position sensor reference frame $X_s^{\left(F\right)}(j,k)=$ $[x_s(j,k),y_s(j,k),z_s(j,k)]:$

$X_s^{\left(F\right)}(j,k)=X_m^{\left(F\right)}(j,k)\·T_{w-m}\·T_{w-s}^{-1}\left(j\right)---\left(25\right)$

Wherein, T _w-s(j) while being Img (j) imaging, the locus transformation matrix of position sensor in reference frame, can read in real time from navigation system, and T _w-mthat the water tank model coordinate that the 4th step solves is tied to the spatial alternation matrix between reference frame.

Notice the point in the sensor space the match point of fastening in image coordinate is exactly the two-dimensional coordinate to bend at the upper N line extracting of ultrasonoscopy Img (j) so just formed a collection of from image coordinate be tied between sensor reference coordinate system, put one to one right with by being write as homogeneous coordinates, just can calculate the conversion T between them _s-u, as equation (26):

$X_s^{\left(F\right)}(j,k)=X_u^{\left(F\right)}(j,k)\·T_{s-u}\left(j\right)---\left(26\right)$

What equation (26) obtained is exactly the spatial alternation matrix T of utilizing a width ultrasonoscopy Img (j) of traditional N line method demarcation _s-u(j), for improving its precision, the spatial alternation matrix that several ultrasonoscopys can be calibrated averages, and obtains carrying out with several ultrasonoscopys the result that traditional N line is demarcated, and is designated as T ^t _s-u.

$T_{s-u}^T=\frac{1}{J}{\mathrm{\Σ}}_{j=1}^J{T}_{s-u}\left(j\right)---\left(27\right)$

Wherein J is the ultrasonoscopy number of collection.

Compare with conventional process flow, can see traditional N line model scaling method, directly utilize water mould design data and water mould locating information, utilize two-dimentional index point position and the similar triangles relation on ultrasonoscopy, extracted to resolve plane of ultrasound and water mould fine rule intersection point, and utilize the pairing point set extracting, utilize equation group (26) to calculate and solve ultrasonoscopy to the spatial alternation T of sensor _s-uand our invention, the relative position information of several ultrasonoscopys when by imaging, space correlation relation between multiple image is used, the optimization that forms distance between gang's two dimensional image target and three-dimensional fine rule in water mould coordinate system is calculated, reason, more effectively utilized the information obtaining, therefore can improve the precision of demarcation.

We adopt above-mentioned handling process to carry out actual experiment test.Tested altogether 8 sets of data, every cover all comprises the water mould ultrasonoscopy of 7 groups of diverse locations collections and corresponding locus sensing data.In processing, we choose 5 groups of ultrasonoscopys in every sets of data and locator data for the demarcation of imaging plane and sensor locus, and the checking for precision by other two groups of data.Therefore every group of data one have C ₅ ⁷=21 kinds of combinations, the mean accuracy that we demarcate these 21 groups combinations is as the precision of overall calculating.In experiment, adopt respectively traditional N line model computational methods and method of the present invention to process, the stated accuracy obtaining is compared.Fig. 7 provided one group typically with the data of thickness error in computational process, the curve that calibrated error declines with the ultrasonoscopy number that participates in calculating.Can see, when selecting a width ultrasonoscopy to carry out timing signal, adopt our method to be optimized, can reduce to a certain extent error, but calibrated error is still larger, and after introducing several ultrasonoscopys, the calibrated error of the inventive method more promptly declines than traditional method, when adopting 3 width with epigraph, can reach smaller error, and adopt 5 width images to carry out timing signal, obtained good demarcation effect.Therefore processing method of the present invention can obtain higher stated accuracy on the basis that gathers a small amount of ultrasonoscopy and locator data.

Inventor's situation

Inventor: Wang Guangzhi, fourth brightness, Zhu Liren

Unit: medical college biomedical engineering system of Tsing-Hua University

Telephone number: (010) 62783631

Contact person: fourth brightness, E-mail: dinghuitsinghua.edu.cn

Plan is applied for a patent type: patent of invention

Claims (1)

1. the optimization method that a ultrasonic probe imaging plane locus is demarcated, it is characterized in that, this optimization method is the method for transformation relation between a kind of locus of demarcating ultrasonic probe imaging plane and sensor locus, locus, be the optimization method of spatial alternation matrix, this optimization method is to realize according to the following steps successively under a three-dimensional localization measuring system auxiliary:

Step 1, three-dimensional localization measuring system of demarcating for ultrasonic probe of structure:

Described three-dimensional localization measuring system of demarcating for ultrasonic probe, comprise: three-dimensional localization measuring instrument, water tank model, Medical Ultrasonic Imaging System and personal computer, three-dimensional localization measuring instrument wherein, water tank model, ultrasonic image-forming system is jointly for measuring the ultrasonic probe of being demarcated, and personal computer is for collection and the date processing of measurement data, three-dimensional localization measuring instrument wherein comprises: be fixed on the 6DOF locus sensor on ultrasonic probe, the three-dimensional localization probe pen that most advanced and sophisticated positioning precision is 0.2mm, location wireless launcher as georeferencing coordinate system, ultrasonic image-forming system comprises ultrasonic probe and the imaging system main frame of being demarcated, ultrasonoscopy for generation of water tank model to be collected, wherein demarcated fixed above-mentioned 6DOF locus sensor on ultrasonic probe, picture catching card has been installed in personal computer, for gathering from the ultrasonoscopy of described ultrasonic image-forming system output, simultaneously, personal computer is the locator data of 6DOF locus sensor and the locator data at three-dimensional localization probe pen tip from described three-dimensional localization measuring instrument output by USB interface collection,

Step 2, making N line model: described N line model is the fine rule of lining up " N " word shape by many groups, layering also successively waits highland to be fixed in the cuboid water tank of three direction sizes of length setting in advance, each N font forms one " N line target " described N line model, it is the coaxial aperture running through by making on the front and rear wall at cuboid water tank, penetrate therein thin nylon wire and form the N line in water tank, described N line model is according to visual field size and the image resolution of described ultrasonic probe, from top to bottom, arrange 3 layers of N line, maximum layer comprises 3 N lines, all the other are two-layer respectively comprises 5 N shapes, so that ultrasonic image can gather more N line target, and make described each N line target be full of as far as possible the visual field of ultra sonic imaging, each is organized height and the position of N line in water tank model coordinate systems and sets in advance, and guarantee its positional precision by the manufacture of water tank model, the pit-mark of respectively arranging four location surveys at corner location on two walls before and after water tank, each forms a rectangle, the position of putting for defining water tank coordinate system measure water case,

Step 3, the coordinate system set of setting up described three-dimensional localization measuring system of demarcating for ultrasonic probe: comprise reference frame, water tank model coordinate systems, locus sensor coordinate system, ultra sonic imaging plane coordinate system, reference frame X _wy _wz _wrepresent, be located at described location with on wireless launcher, zero is at described discharger center; Water tank model coordinate systems X _my _mz _mrepresent X _mwater tank length direction, Z _mwater tank width, Y _mbe water tank short transverse, the initial point of coordinate system is located on the location pit in the water tank front face upper left corner, and the direction of each coordinate axes is followed right-hand rule; Locus sensor coordinate system X _sy _sz _srepresent, be located on the sensor of described 6DOF locus, zero is at the center of described locus sensor; Ultra sonic imaging plane coordinate system X _uy _urepresent, be located on ultrasonoscopy, zero is located at ultrasonoscopy mid point topmost, X _udirection is image level direction, Y _udirection is ultra sonic imaging depth direction,

For setting up the transformation relation between each coordinate system, set the spatial alternation matrix T of 4*4 _w-mfrom described water tank model coordinate systems X _my _mz _mto described reference frame X _wy _wz _wspatial alternation matrix, T _w-sfrom described locus sensor coordinate system X _sy _sz _sto described reference frame X _wy _wz _wspatial alternation matrix, T _s-ufrom described ultra sonic imaging plane coordinate system X _uy _ulocus sensor coordinate system X to described location use _sy _sz _sspatial alternation matrix, T _u-mfrom described water tank model coordinate systems X _my _mz _mto described ultra sonic imaging plane coordinate system X _uy _uspatial alternation matrix;

Step 4, demarcate the locus that described water tank model is put, obtain water tank model coordinate systems X _my _mz _mto reference frame X _wy _wz _wspatial alternation matrix:

Step 4.1: adopt the tip of the three-dimensional localization probe pen of described three-dimensional localization measuring system configuration, survey be one by one positioned at described location with each four the location pits of arranging at corner location on two walls before and after the water tank model of the effective measurement zone of wireless launcher at described reference frame X _wy _wz _win three-dimensional space position P _wi=(x _wi, y _wi, z _wi), i=1 wherein, 2 ..., 8, be the serial number of location pit,

Step 4.2: respectively locate the coordinate P of pit in water tank model coordinate systems described in same sequential query or mensuration water tank design and manufacture of casting mould _mi=(x _mi, y _mi, z _mi), i=1 wherein, 2 ..., 8, be the sequence number of location pit,

Step 4.3 is tied to the spatial alternation matrix T of described reference frame by water tank model coordinate described in solving equations below _w-m

$\left[\begin{array}{cccc}{x}_{w1}& y_{w1}& z_{w1}& 1\\ x_{w2}& y_{w2}& z_{w2}& 1\\ x_{w3}& y_{w3}& z_{w3}& 1\\ x_{w4}& y_{w4}& z_{w4}& 1\\ x_{w5}& y_{w5}& z_{w5}& 1\\ x_{w6}& y_{w6}& z_{w6}& 1\\ x_{w7}& y_{w7}& z_{w7}& 1\\ x_{w8}& y_{w8}& z_{w8}& 1\end{array}\right]=\left[\begin{array}{cccc}{x}_{m1}& y_{m1}& z_{m1}& 1\\ x_{m2}& y_{m2}& z_{m2}& 1\\ x_{m3}& y_{m3}& z_{m3}& 1\\ x_{m4}& y_{m4}& z_{m4}& 1\\ x_{m5}& y_{m5}& z_{m5}& 1\\ x_{m6}& y_{m6}& z_{m6}& 1\\ x_{m7}& y_{m7}& z_{m7}& 1\\ x_{m8}& y_{m8}& z_{m8}& 1\end{array}\right]\·T_{w-m};$

Step 5, gather several ultrasonoscopys under diverse location in water tank model: what described in using, have 6DOF locus sensor is demarcated ultrasonic probe, many groups N line fine rule in water tank is carried out to imaging, described ultrasonic probe is arranged in to the top of described water tank model, by mobile ultrasonic probe, in non-overlapping or parallel different imaging planes, gather ultrasonoscopys more than 7 width, and by described picture catching card, the ultrasonoscopy of collection is successively inputted to described personal computer, be stored as Img (j), j=1, 2, J, the ultrasonoscopy number of J for gathering, simultaneously, pass through USB interface, described in reading while gathering every width ultrasonoscopy Img (j), locus sensor is at described reference frame X _wy _wz _win locus and attitude, and calculate locus sensor coordinate system X _sy _sz _sto described reference frame X _wy _wz _wspatial alternation matrix T _w-s(j), the ultrasonoscopy of inputting can show on the screen of described personal computer, and the cross section of each N line fine rule is shown as speck on ultrasonoscopy,

Step 6, extract the position of N line index point on each width ultrasonoscopy and the position of calculating this plane of ultrasound, step is as follows:

The ultrasonoscopy Img (j) that step 6.1 shows on described personal computer goes up according to first along X _mincrease progressively, then press Y _mthe order increasing progressively, the manual two-dimensional coordinate position that clicks each speck in all N lines in the ultrasonoscopy visual field and crossing formed one group of three speck of ultrasound image plane successively, and the order being clicked according to every group of N line writes down its sequence number k, the speck of three limits of one group of N line of the choosing of setting up an office on ultrasonoscopy is respectively E (j, k), F (j, k), G (j, k), k=1 wherein, 2,, K _j, K _jfor the sum at the upper N line index point obtaining of this width ultrasonoscopy Img (j), the two-dimensional coordinate of the speck that note clicks $X_u^{\left(E\right)}(j,k)=[x_u^{\left(E\right)}(j,k),y_u^{\left(E\right)}(j,k)],$ $X_u^{\left(F\right)}(j,k)=[x_u^{\left(F\right)}(j,k),y_u^{\left(F\right)}(j,k)],$ according to image resolution ratio, obtain the distance between three clicked points | EF| _jkwith | EG| _jk:

${\left|\mathrm{EF}\right|}_{\mathrm{jk}}=\sqrt{{(x_u^{\left(E\right)}(j,k)-x_u^{\left(F\right)}(j,k))}^2+{(y_u^{\left(E\right)}(j,k)-y_u^{\left(F\right)}(j,k))}^2};$

${\left|\mathrm{EG}\right|}_{\mathrm{jk}}=\sqrt{{(x_u^{\left(E\right)}(j,k)-x_u^{\left(G\right)}(j,k))}^2+{(y_u^{\left(E\right)}(j,k)-y_u^{\left(G\right)}(j,k))}^2};$

According to the method for step 6.1, the N line index point position on the ultrasonoscopy of all collections is extracted, and according to sequence number storage,

Step 6.2 is obtained according to the following steps for any width ultrasonoscopy Img (j), from X _my _mz _mcoordinate is tied to ultra sonic imaging plane coordinate system X _u, Y _uspatial alternation matrix T _u-m(j):

The sequence number k of the N line that step 6.2.1 clicks on ultrasonoscopy according to step 6.1, select the N line group of correspondence position in water tank model, known four threading aperture A at water tank certain group N line that sequence number is k on inwall just, below, B, C, D are at water tank model coordinate systems X _my _mz _min three-dimensional coordinate be respectively

$X_m^{\left(A\right)}(j,k)=[x_m^{\left(A\right)}(j,k),y_m^{\left(A\right)}(j,k),z_m^{\left(A\right)}(j,k)],$

$X_m^{\left(B\right)}(j,k)=[x_m^{\left(B\right)}(j,k),y_m^{\left(B\right)}(j,k),z_m^{\left(B\right)}(j,k)],$

$X_m^{\left(C\right)}(j,k)=[x_m^{\left(C\right)}(j,k),y_m^{\left(C\right)}(j,k),z_m^{\left(C\right)}(j,k)],$

$X_m^{\left(D\right)}(j,k)=[x_m^{\left(D\right)}(j,k),y_m^{\left(D\right)}(j,k),z_m^{\left(D\right)}(j,k)],$

Wherein A hole is coaxial with B hole, and C hole is coaxial with D hole, that is: $y_m^{\left(A\right)}(j,k)=y_m^{\left(B\right)}(j,k),$ $x_m^{\left(C\right)}(j,k)=x_m^{\left(D\right)}(j,k),$ $y_m^{\left(C\right)}(j,k)=y_m^{\left(D\right)}(j,k),$ Therefore, two fine rules that link A, B hole and link C, D hole have formed the straight line of N line index, the fine rule that links B, C hole has formed the hypotenuse of N line index, if one group of N line that ultrasonoscopy Img (j) and sequence number that sequence number is j are k cuts mutually, fine rule AB in this width ultrasonoscopy and N line intersects at E, BC intersects at F with fine rule, and CD intersects at G with fine rule, and definition E, F, the three-dimensional coordinate of G point in water tank model are:

$X_m^{\left(E\right)}(j,k)=[x_m^{\left(E\right)}(j,k),y_m^{\left(E\right)}(j,k),z_m^{\left(E\right)}(j,k)],$

$X_m^{\left(F\right)}(j,k)=[x_m^{\left(F\right)}(j,k),y_m^{\left(F\right)}(j,k),z_m^{\left(F\right)}(j,k)],$

$X_m^{\left(G\right)}(j,k)=[x_m^{\left(G\right)}(j,k),y_m^{\left(G\right)}(j,k),z_m^{\left(G\right)}(j,k)]$

Step 6.2.2 obtains in the N line that this group sequence number is k F point at water tank model coordinate systems X by following formula _my _mz _min three-dimensional coordinate $X_m^{\left(F\right)}(j,k)=[x_m^{\left(F\right)}(j,k),y_m^{\left(F\right)}(j,k),z_m^{\left(F\right)}(j,k)]:$

$\left\{\begin{array}{c}{x}_m^{\left(F\right)}(j,k)=x_m^{\left(B\right)}(j,k)+\frac{{\left|\mathrm{EF}\right|}_{\mathrm{jk}}}{{\left|\mathrm{EG}\right|}_{\mathrm{jk}}}\·{\left|\mathrm{BD}\right|}_{\mathrm{jk}}\\ y_m^{\left(F\right)}(j,k)=y_m^{\left(B\right)}(j,k)\\ z_m^{\left(F\right)}(j,k)=z_m^{\left(B\right)}(j,k)+\frac{{\left|\mathrm{EF}\right|}_{\mathrm{jk}}}{{\left|\mathrm{EG}\right|}_{\mathrm{jk}}}\·{\left|\mathrm{AB}\right|}_{\mathrm{jk}}\end{array}\right.$

Wherein: ${\left|\mathrm{BD}\right|}_{\mathrm{jk}}=\sqrt{{(x_m^{\left(B\right)}(j,k)-x_m^{\left(D\right)}(j,k))}^2},$ ${\left|\mathrm{AB}\right|}_{\mathrm{jk}}=\sqrt{{(z_m^{\left(B\right)}(j,k)-z_m^{\left(D\right)}(j,k))}^2}$

Step 6.2.3 personal computer, by the measurement to the upper all complete N line speck obtaining of this width ultrasonoscopy Img (j), is respectively organized N line index point F at X described in obtaining _my _mz _mthree-dimensional coordinate in coordinate system k=1,2 ..., K _j, K _jfor the sum at the upper N line index point obtaining of this width ultrasonoscopy Img (j),

The corresponding one group of N line index point F of a width ultrasonoscopy j that step 6.2.4 obtains according to step (6.2.3) and step (6.2.2), at X _my _mz _mthree-dimensional coordinate under coordinate system is expressed as: , the coordinate figure of this point is solved and obtains by step 6.2.2, the F index point of the upper different sequence number k that extract of same width ultrasonoscopy Img (j) in water tank three-dimensional system of coordinate, will form a plane, and these points are at ultrasonoscopy coordinate system X _uy _uin respective coordinates be to Img (j), utilize below Solving Equations to solve the coordinate Mapping being tied between this ultrasound image plane coordinate system from water tank model coordinate and be related to T _u-m(j):

$\left[\begin{array}{cccc}{x}_u^{\left(F\right)}(j,1)& y_u^{\left(F\right)}(j,1)& 0& 1\\ x_u^{\left(F\right)}(j,2)& y_u^{\left(F\right)}(j,2)& 0& 1\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ x_u^{\left(F\right)}(j,K_j)& y_u^{\left(F\right)}(j,K_j)& 0& 1\end{array}\right]=\left[\begin{array}{cccc}{x}_m^{\left(F\right)}(j,1)& y_m^{\left(F\right)}(j,1)& z_m^{\left(F\right)}(j,1)& 1\\ x_m^{\left(F\right)}(j,2)& y_m^{\left(F\right)}(j,2)& z_m^{\left(F\right)}(j,2)& 1\\ .& .& .& \\ .& .& .& \\ .& .& .& \\ x_m^{\left(F\right)}(j,K_j)& y_m^{\left(F\right)}(j,K_j)& z_m^{\left(F\right)}(j,K_j)& 1\end{array}\right]\·T_{u-m}\left(j\right)$

Or write as: $X_u^{2D}(j,k)=X_m(j,k)\·T_{u-m}\left(j\right)$

J=1 wherein, 2 ..., J, J is the number of gathered ultrasonoscopy, k=1,2 ..., K _j, K _jit is the number of the N line index point F that can extract on this image I mg (j);

Step 7, the N line index point on described two-dimensional ultrasonic image is mapped on many imaging planes of three-dimensional spatial distribution, described many imaging planes refer to by described locus sensor at X _wy _wz _wthe locus matrix of the corresponding ultrasonoscopy Img (j) obtaining in coordinate system each two-dimensional ultrasonic imaging plane at described X _wy _wz _win coordinate system, combine, formation is a plurality of imaging planes of certain distributed in three dimensions, by locus sensor information, the N line index point extracting on each two-dimensional ultrasonic image is mapped on three-dimensional imaging plane to the N line index point on the ultrasonoscopy then step 6.1 being extracted by the position and attitude that gathers this image time space position sensor, combine, form a collection of registration point in three dimensions, the unified two-dimensional coordinate by the upper N line index point extracting of ultrasonoscopy Img (j) is written as: l=1 wherein, 2 ..., 3*K _j, K _jit is the number of the N line index point that extracts on image I mg (j); the three-dimensional coordinate being mapped on each imaging plane of three dimensions is $X_u^{3D}(j,l)=(x_u^{3D}(j,l),y_u^{3D}(j,l),z_u^{3D}(j,l)),$ :

$X_u^{3D}(j,l)=X_u^{2D}(j,l)\·T_{w-s}\left(j\right)$

Wherein, T _w-s(j) be the evolution matrix of the position sensor of corresponding Img (j) imaging plane that obtains from described personal computer from three-dimensional localization measuring instrument, j=1,2 ..., J, J is the ultrasonoscopy number gathering, l=1,2 ..., 3*K _j, K _jit is the number of the N line index point that extracts on image I mg (j);

Step 8, according to the following steps to a collection of three-dimensional coordinate point that is mapped to many imaging planes initialize registration with the N line on water tank model:

Step 8.1, utilizes N line index point on the ultrasonoscopy that step 6.1 and step 6.2.3 obtain with coordinate in corresponding water tank model the ultrasonoscopy that an optional width sequence number is n, utilizes equation group below to calculate from described X _my _mz _mcoordinate is tied to X _uy _uspatial alternation matrix T between coordinate system _u-m(n):

$X_u^{\left(F\right)}(n,k)=X_m^{\left(F\right)}(n,k)\·T_{u-m}\left(n\right)$

Step 8.2, the initial space transformation matrix T that utilizes step 8.1 to obtain _u-m(n) three-dimensional coordinate of all index points that, step 7 obtained j=1,2 ..., n-1, n+1 ... J, l=1,2 ..., 3*K _jbe mapped in water tank model coordinate systems, calculate a collection of three-dimensional coordinate point subject to registration in water tank model coordinate systems

$X_m^{3D}(j,l)=X_u^{3D}(j,l)\·T_{w-s}^{-1}\left(n\right)\·T_{u-m}^{-1}\left(n\right);$

Thereby tentatively the N line index point in each plane of ultrasound together with N line on water tank model matches,

Object function and the initial value that step 9, setting optimization are calculated is also optimized calculating:

Optimization aim function is by X described in set-up procedure 8.1 _my _mz _mcoordinate is tied to X _uy _uspatial alternation matrix T between coordinate system _u-m, make the index point subject to registration of all three-dimensional spatial distribution that step 8.2 obtains it is minimum with the average distance of fine rule corresponding on water tank,

Therefore optimised variable is transformation matrix T _u-mvariables Sequence, the initial position optimize calculating is the transformation matrix T that utilizes n width ultrasonoscopy to calculate _u-m(n), establish total J tomographic image and participate in calculating, the index point number of every layer is K _jindividual, therefore always have J*K _jindividual three-dimensional point is for calculating the distance of water tank model fine rule, by certain index point of direct solution to the distance D of all straight lines in water tank, and choose one of them minimum distance D (j, l) as spatial point distance with nearest fine rule corresponding on 3 d water tank model, obtains index point average distance with fine rule nearest on 3 d water tank model:

$D_{\mathrm{avg}}=\frac{1}{J\·K_j}\underset{j=1}{\overset{J}{\mathrm{\Σ}}}\underset{l=1}{\overset{3*K_j}{\mathrm{\Σ}}}D(j,l)$

Choose average distance D _avgerror threshold be 2mm, greatest iteration step number is 500 steps, by optimization, is calculated and is made average distance D _avgminimization, obtains an optimum matching conversion of fine rule target in many imaging planes and water tank model, is designated as the optimization tool adopting is the sequential quadratic programming function in Matlab optimization toolbox, and English name is sequential quadratic programming algorithm, is called for short SQP;

Step 10, the optimal spatial of calculating between imaging plane and locus sensor convert

Step 10.1, by the two-dimentional N line index point coordinates extracting on each ultrasonoscopy locus sensor transformation matrix T during by this image of collection _w-s(j), transform to three dimensions, obtain the three-dimensional symbol point position on a plurality of imaging planes

$X_u^{3D}(j,l)=X_u^{2D}(j,l)\·T_{w-s}\left(j\right);j=\mathrm{1,2},..,J;l=\mathrm{1,2},...,3*K_j$

Step 10.2, by three-dimensional point set the optimum matching conversion obtaining by step 9 be mapped to water tank model coordinate systems X _my _mz _min:

$X_m^{3D}(j,l)=X_u^{3D}(j,l)\·{\tilde{T}}_{u-m}^{-1};j=\mathrm{1,2},..,J;l=\mathrm{1,2},...,3*K_j,$

Step 10.3, by water tank model coordinate systems X _my _mz _min point set transform to sensor coordinate system X _sy _sz _sin:

$X_s^{3D}(j,l)=X_m^{3D}(j,l)\·T_{w-m}\·T_{w-s}^{-1}\left(j\right);j=\mathrm{1,2},..,J;l=\mathrm{1,2},...,3*K_j,$

Step 10.4, utilizes the corresponding point set obtaining in step 10.3 and step 10.1 with utilize solving equations ultrasonoscopy below to convert to the global optimum between sensor

$X_s^{3D}(j,l)=X_u^{3D}(j,l)\·{\tilde{T}}_{s-u};j=\mathrm{1,2},..,J;l=\mathrm{1,2},...,3*K_j,$

Solve and calculate the ultrasonoscopy coordinate that will demarcate is exactly tied to the optimal transformation between the locus sensor coordinate system on ultrasonic probe.