CN103584888A - Ultrasonic target motion tracking method - Google Patents

Abstract

The invention belongs to the field of medical treatment and image processing, and provides an ultrasonic target motion tracking method. The ultrasonic target motion tracking method comprises the steps of identifying a target image selected by a tracker on a first frame, defining the target image as target information to be tracked, defining the search range of the next frame around the target information, calculating coefficients of similarity between the frame and the target information in a sliding window within the search range, and using the position with the maximum similarity coefficient as the tracked position of the frame. In order to not lose a target, a method of adjusting the search range of the next frame in a self-adaptation mode according to the maximum similarity coefficient is provided. In order to reduce the frequency of the jitter phenomenon, a method for calculating the similarity coefficient in a central weighting mode is provided. According to the ultrasonic target motion tracking method, a structural similarity coefficient model is used in cooperation, tracking is carried out on the target by adjusting the search range in the self-adaptation mode on the basis of the target information and the central weighting similarity coefficient of an image in the sliding window, the problem of jitter is solved, and meanwhile the target can be precisely tracked.

Description

Ultrasonic target travel method for tracing

Technical field

The invention belongs to computer vision field, relate in particular to a kind of ultrasonic target travel method for tracing.

Background technology

In ultrasonic contrast diagnosis, often to dynamically carry out quantitative analysis to the tangent plane of the cross section of certain root blood vessel or certain tumor, and these target areas or because of the breathing campaign by diagnosis person, or be difficult to be fixed on a certain position of screen because of the movement of diagnosis person and probe, this has caused the input of quantitative analysis to mix the information that nontarget area information does not have target area even at all, and these all greatly reduce the credibility of quantitative analytical data.This problem definition: the target area of quantitative analysis should not be fixed on a certain position of screen, and should move along with the motion of target.

On the other hand, ultrasonoscopy target is often understood deformation, and shape can along with the difference of scanning section, difference even disappears, and this is the main aspect that is different from natural image target tracking, is also the difficult point place of ultrasonic target tracking.

Existing method for tracing generally adopts Minimum Mean Square Error, least absolute value and, maximum cross correlation coefficient and rectangular histogram the simple relatively method of two width image similarity degree such as mate most.These methods are all relatively not consider the structural information of target by Pixel Information, and it is obviously inappropriate being therefore used in the ultrasonic image of a lot of speckle noises, is difficult to accurately follow the trail of the objective.Minimum Mean Square Error and least absolute value and method for tracing be meeting lose objects when following the trail of the objective slightly deformation, can only follow the trail of the objective the short time, and the method for tracing mating most based on rectangular histogram is only considered the intensity profile information of target, can produce serious jitter phenomenon, target is not often in region central authorities, simultaneously easily with losing target.

On the other hand, except maximum cross correlation coefficient value has certain limit, the similarity degree measure numerical value of other method or without the upper bound or without bound, therefore we cannot determine what kind of the similarity degree of this two width image is to get a numerical value, therefore, further, cannot come self adaptation adjustment aim hunting zone according to image similarity degree, lack this adjustment mechanism and can cause when displacement of targets lose objects and follow the trail of failure when large.

In realizing the scheme of prior art, find that prior art exists following technical problem:

Existing technical scheme cannot follow the trail of the objective accurately, and inaccurate problem causes following the trail of the objective.

Summary of the invention

The object of the embodiment of the present invention is to provide a kind of ultrasonic target travel method for tracing, and it solves the problem that cannot follow the trail of the objective accurately of prior art.

The embodiment of the present invention is achieved in that one side, and a kind of ultrasonic target travel method for tracing is provided, and described method comprises the steps:

The target of using depth-first Flood Fill method to obtain before starting tracking follower in the first frame B ultrasonic image identifies, and obtains target area A ₁; And the image information of obtaining this region is as target information (be and treat tracked information to be designated as x), this information remains unchanged in tracing process.

Obtain the region of search B of the second frame ₂, target area A wherein ₁be positioned at region of search B ₂，Qie region of search, center B ₂length be target area A ₁length+2*SW; Region of search B ₂height be target area A ₁height+2*SH; Wherein, SW is hunting zone varying width, and SH is hunting zone change in elevation;

With B ₂the point in the upper left corner be starting point, at B ₂obtain A ₁the information of the sampling window of size; By setpoint distance, move sampling window, every movement once, is obtained the information of a sampling window, i.e. sampled signal y;

Use structural similarity Modulus Model SSIM to calculate the SSIM value of echo signal x and sampled signal y;

SSIM(x,y)＝[l(x,y)] ^α[c(x,y)] ^β[s(x,y)] ^γ

Wherein

$l(x,y)=\frac{{2\mathrm{\&mu;}}_x{\mathrm{\&mu;}}_y+{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="HDA0000428408210000021.png,HDA0000428408210000042.png"\; state="\{\{state\}\}">C</figure-callout>}}_1}{{\mathrm{\&mu;}}_x^2{\mathrm{\&mu;}}_{\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">y</figure-callout>}}^2+{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="HDA0000428408210000021.png,HDA0000428408210000042.png"\; state="\{\{state\}\}">C</figure-callout>}}_1},$

$c(x,y)=\frac{{2\mathrm{\&sigma;}}_x{\mathrm{\&sigma;}}_y+{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_2}{{\mathrm{\&sigma;}}_x^2{\mathrm{\&sigma;}}_y^2+{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_2},$

$s(x,y)=\frac{{\mathrm{\&sigma;}}_{\mathrm{xy}}+{\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_3}{{\mathrm{\&sigma;}}_x{\mathrm{\&sigma;}}_y+{\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_3}.$

Wherein, l (x, y) is the relatively brightness of x and y, C (x, y) is for comparing the contrast of x and y, S (x, y) for comparing the structure of x and y, α > 0, β > 0, γ > 0, and α, β, γ are respectively and adjust l (x, y), C (x, y), S (x, y) parameter of relative importance, μ _xand μ _ybe respectively the meansigma methods of x and y; σ _xand σ _ybe respectively the standard deviation of x and y, σ _xyfor the co-variation heteromerism of x and y, C ₁, C ₂, C ₃be all constant, in order to maintain the stable of l (x, y), C (x, y), S (x, y);

Choosing sampling window corresponding to maximum SSIM value is that final trace location is the target area A of next frame _n.

Optionally, described method also comprises after being final trace location choosing sampling window that maximum SSIM is corresponding:

Obtain the region of search B of next frame _n; Target area A wherein _n-1be positioned at region of search B _n，Qie region of search, center B _nlength be target area A _n-1length+2*SW (search width adjustment amount); Region of search B _nheight be target area A _n-1height+2*SH (search height adjustment amount); Wherein, n is the frame number of B ultrasonic image; A _n-1it is the target area of n-1 frame B ultrasonic image;

With B _nthe point in the upper left corner in region is starting point, at B _nregion obtains A _n-1the information of the sampling window of size; By setpoint distance, move sampling window, every movement once, is obtained the information of a sampling window, i.e. sampled signal y _ni; Wherein i is the i time mobile logo.

Use structural similarity Modulus Model to calculate echo signal x and sampled signal y _nisSIM value;

Choosing sampling window corresponding to maximum SSIM value is that final trace location is the target area A of next frame _n; N is greater than 2 integer.

Optionally, in the method for described tracking, described target area becomes regular figure from irregular figure.

Optionally, described use structural similarity Modulus Model SSIM zoning A ₁echo signal x and the SSIM value of sampled signal y specifically comprise:

A ₁echo signal x and A ₁sampled signal y be divided into the fritter of n*n, calculate the SSIM of corresponding blocks, the SSIM of all corresponding blocks is added up and is total ssim, be i.e. the SSIM value of echo signal x and sampled signal y;

$\mathrm{total}\_\mathrm{ssim}=\underset{i,j<n}{\mathrm{\&Sigma;}}\mathrm{SSIM}({\mathrm{<figure-callout\; id="1"\; label="I"\; filenames="HDA0000428408210000021.png,HDA0000428408210000042.png"\; state="\{\{state\}\}">I</figure-callout>}1}_{\mathrm{ij}},{\mathrm{<figure-callout\; id="2"\; label="I"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">I</figure-callout>}2}_{\mathrm{ij}})*W_{\mathrm{ij}};$

Wherein, I1 _ijrepresent the corresponding fritter of echo signal x, I2 _ijrepresent the corresponding fritter of sampled signal y, w represents weighted value.

Described method also comprises after choosing maximum SSIM value:

When maximum SSIM is larger, when SSIM is greater than first threshold, reduce the value of SW and SH; When maximum SSIM is less, be greater than Second Threshold, while being less than first threshold, increase the value of SW and SH; When SSIM is very little, while being less than Second Threshold, enter alarm program, stop following the trail of.

In embodiments of the present invention, technical scheme provided by the invention has accurately and follows the trail of the objective, and solves the advantage of jitter phenomenon.

Accompanying drawing explanation

Fig. 1 is the flow chart of a kind of ultrasonic target travel method for tracing provided by the invention;

Fig. 2 is the basic module figure of a kind of ultrasonic target travel method for tracing provided by the invention;

Fig. 3 is the detail flowchart of a kind of ultrasonic target travel method for tracing provided by the invention;

Fig. 4 is the area schematic of the second two field picture provided by the invention;

Fig. 5 is the sign schematic diagram that follows the trail of the objective in example provided by the invention;

Fig. 6 is irregular in example provided by the invention and regular targets sign schematic diagram;

Fig. 7 is the tracking result schematic diagram of the 350th frame and the 568th frame in example provided by the invention;

Fig. 8 is the schematic diagram of the weighted calculation SSIM that provides of embodiments of the invention;

Fig. 9 is the weighting windows schematic diagram of weighted value provided by the invention;

Figure 10 is the exemplary weights window schematic diagram of weighted value provided by the invention;

Figure 11 is more new module of target search scope provided by the invention.

The specific embodiment

In order to make object of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.

The specific embodiment of the invention provides a kind of ultrasonic target travel method for tracing, and the method as described in Figure 1, comprises the steps:

1001, the target of using depth-first Flood Fill method to obtain before starting tracking follower in the first frame B ultrasonic image identifies, and obtains target area A ₁; And the image information of obtaining this region is as target information (be and treat tracked information to be designated as x), this information remains unchanged in tracing process.

1002, obtain the region of search B of the second frame ₂, target area A wherein ₁be positioned at region of search B ₂，Qie region of search, center B ₂length be target area A ₁length+2*SW (search width); Region of search B ₂height be target area A ₁height+2*SH (search height);

1003, with B ₂the point in the upper left corner in region is starting point, at B ₂region obtains A ₁the information of the sampling window of size; By setpoint distance, move sampling window, every movement once, is obtained the information of a sampling window, i.e. sampled signal y;

1004, use structural similarity Modulus Model (English full name: structural similarity index model, English abbreviation: SSIM) calculate A ₁echo signal x and the SSIM value of sampled signal y;

SSIM(x,y)＝[l(x,y)] ^α[c(x,y)] ^β[s(x,y)] ^γ

Wherein

$l(x,y)=\frac{{2\mathrm{\&mu;}}_x{\mathrm{\&mu;}}_y+{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="HDA0000428408210000021.png,HDA0000428408210000042.png"\; state="\{\{state\}\}">C</figure-callout>}}_1}{{\mathrm{\&mu;}}_x^2{\mathrm{\&mu;}}_{\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">y</figure-callout>}}^2+{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="HDA0000428408210000021.png,HDA0000428408210000042.png"\; state="\{\{state\}\}">C</figure-callout>}}_1},$

$c(x,y)=\frac{{2\mathrm{\&sigma;}}_x{\mathrm{\&sigma;}}_y+{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_2}{{\mathrm{\&sigma;}}_x^2{\mathrm{\&sigma;}}_y^2+{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_2},$

$s(x,y)=\frac{{\mathrm{\&sigma;}}_{\mathrm{xy}}+{\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_3}{{\mathrm{\&sigma;}}_x{\mathrm{\&sigma;}}_y+{\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="HDA0000428408210000021.png"\; state="\{\{state\}\}">C</figure-callout>}}_3}.$

Wherein, l (x, y) is the relatively brightness of x and y, C (x, y) is for comparing the contrast of x and y, S (x, y) for comparing the structure of x and y, α > 0, β > 0, γ > 0 and α, β, γ are respectively and adjust l (x, y), C (x, y), the parameter of S (x, y) relative importance, μ _xand μ _ybe respectively the meansigma methods of x and y; σ _xand σ _ybe respectively the standard deviation of x and y, σ _xyfor the co-variation heteromerism of x and y, C ₁, C ₂, C ₃be all constant, in order to maintain the stable of l (x, y), C (x, y), S (x, y); .The value of SSIM, in [11] scope, more approaches 1 and represents that the similarity degree of two signals is higher.

1005, choosing sampling window corresponding to maximum SSIM value is that final trace location is the target area A of next frame ₂.

Optionally, in the method for described tracking, described target area becomes regular figure (concrete variation as shown in Figure 6) from irregular figure.

Optionally, said method 1005 choose maximum SSIM after, can also comprise:

When maximum SSIM larger, while being greater than first threshold (as being greater than 0.7) illustrate target information and final position sample information similarity degree higher, within the scope of current search, final position almost overlaps with target's center, be that displacement of targets is little, now can suitably reduce hunting zone (reducing SW and SH according to motion vector) and improve tracking speed; On the contrary, when maximum SSIM is less, be greater than Second Threshold, while being less than first threshold (as 0.4 to 0.7), explanation final position and target's center within the scope of current search have and depart from, i.e. displacement is larger, now should increase hunting zone (increasing SW and SH according to motion vector); When SSIM is very little, while being less than Second Threshold (as being less than 0.4), explanation follows the trail of the objective and at present frame, disappears, and now should enter alarm program, stops following the trail of (as Figure 11).

Optionally, said method can also comprise after 1005:

Obtain the region of search B of next frame _n; Target area A wherein _n-1be positioned at region of search B _n，Qie region of search, center B _nlength be target area A _n-1length+2*SW (search width); Region of search B _nheight be target area A _n-1height+2*SH (search height); Wherein, n is the frame number of B ultrasonic image; A _n-1it is the target area of n-1 frame B ultrasonic image;

With B _nthe point in the upper left corner in region is starting point, at B _nregion obtains A _n-1the information of the sampling window of size; By setpoint distance, move sampling window, every movement once, is obtained the information of a sampling window, i.e. sampled signal y _n;

Use structural similarity Modulus Model to calculate A _n-1echo signal x and sampled signal y _nsSIM;

Choosing the sampling window that maximum SSIM is corresponding is that final trace location is the target area A of next frame _n; N is greater than 2 integer.

Optionally, the distance that the setpoint distance in above-mentioned 1003 is single pixel or the distance of a plurality of pixels, but the minimum distance that does not surpass 5 pixels of these a plurality of pixels, if setpoint distance is oversize, shake is more serious.This kind of mode can reduce amount of calculation.

The basic module of said method as shown in Figure 2, specifically comprises: target image information collection 1, image sampling 2, the comparison 3 of target sample information and definite trace location 4.As shown in Figure 3, in said method, as shown in Figure 4, wherein in Fig. 4, B is region of search to the image of the second frame to the detail flowchart of this method, and A is the target area of the first frame; SW is search width, and SH is search height, and SW and SH all can be set voluntarily by user; C represents the viewing area of B ultrasonic image;

Above-mentioned 1004 implementation method is specifically as follows: A ₁echo signal x(as the image 1 in Fig. 8) and A ₁sampled signal y(as the image 2 in Fig. 8) (example n is 3 to be divided into the fritter of n*n, certainly n can be also other numerical value), calculate the SSIM of corresponding blocks, finally they are weighted to summation with the large weighting windows of central weights (as exemplary weights window) and draw image 1 and the final structural similarity coefficient total_ssim of image 2;

$\mathrm{total}\_\mathrm{ssim}=\underset{i,j<n}{\mathrm{\&Sigma;}}\mathrm{SSIM}({I1}_{\mathrm{ij}},{I2}_{\mathrm{ij}})*W_{\mathrm{ij}};$

Wherein, I1i _jrepresent the corresponding fritter of echo signal x, I2i _jrepresent the corresponding fritter of sampled signal y, w represents weighted value (as shown in Figure 9), and wherein W value can be set (as shown in figure 10) by user self.

This weighting scheme is based on such fact, follower is when choosing following the trail of the objective at first, generally target can be retouched in central authorities, be that sampling window center should comprise maximum target informations, the weight that strengthens target information meets such fact undoubtedly, this scheme can make trace location more accurate, further reduces shake.

Example

Fig. 5 to Fig. 7 is an example of the present invention, follower identifies and waits to follow the trail of the objective in a liver B ultrasonic image of Fig. 5, it is a vascular cross-section herein, Fig. 6 is the irregular and regular targets sign that the present invention generates, Fig. 7 follows the trail of result figure through two of 600 two field pictures, and left figure follows the trail of result when having compared with large deformation, and right figure is the tracking result while having larger displacement, can find out that the present invention can accurately follow the trail of ，Qie target's center in identified areas center non-jitter phenomenon.

It should be noted that in above-described embodiment, included unit is just divided according to function logic, but is not limited to above-mentioned division, as long as can realize corresponding function; In addition, the concrete title of each functional unit also, just for the ease of mutual differentiation, is not limited to protection scope of the present invention.

In addition, one of ordinary skill in the art will appreciate that all or part of step realizing in the various embodiments described above method is to come the hardware that instruction is relevant to complete by program, corresponding program can be stored in a computer read/write memory medium, described storage medium, as ROM/RAM, disk or CD etc.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any modifications of doing within the spirit and principles in the present invention, be equal to and replace and improvement etc., within all should being included in protection scope of the present invention.

Claims (5)

1. a ultrasonic target travel method for tracing, is characterized in that, described method comprises the steps:

The target of using depth-first Flood Fill method to obtain before starting tracking follower in the first frame B ultrasonic image identifies, and obtains target area A ₁;

Obtain the region of search B of the second frame ₂, target area A wherein ₁be positioned at region of search B ₂，Qie region of search, center B ₂length be target area A ₁length+2*SW; Region of search B ₂height be target area A ₁height+2*SH; Wherein, SW is initial ranging width, and SH is initial ranging height;

With B ₂the point in the upper left corner be starting point, at B ₂obtain A ₁the information of the sampling window of size; By setpoint distance, move sampling window, every movement once, is obtained the information of a sampling window, i.e. sampled signal y;

Use structural similarity Modulus Model SSIM zoning A ₁echo signal x and the SSIM value of sampled signal y;

SSIM(x,y)＝[l(x,y)] ^α[c(x,y)] ^β[s(x,y)] ^γ

Wherein

$l(x,y)=\frac{{2\mathrm{\&mu;}}_x{\mathrm{\&mu;}}_y+C_1}{{\mathrm{\&mu;}}_x^2{\mathrm{\&mu;}}_y^2+C_1},$

$c(x,y)=\frac{{2\mathrm{\&sigma;}}_x{\mathrm{\&sigma;}}_y+C_2}{{\mathrm{\&sigma;}}_x^2{\mathrm{\&sigma;}}_y^2+C_2},$

$s(x,y)=\frac{{\mathrm{\&sigma;}}_{\mathrm{xy}}+C_3}{{\mathrm{\&sigma;}}_x{\mathrm{\&sigma;}}_y+C_3}.$

Wherein, l (x, y) is the relatively brightness of x and y, C (x, y) is for comparing the contrast of x and y, S (x, y) for comparing the structure of x and y, α > 0, β > 0, γ > 0, and α, β, γ are respectively and adjust l (x, y), C (x, y), S (x, y) parameter of relative importance, μ _xand μ _ybe respectively the meansigma methods of x and y; σ _xand σ _ybe respectively the standard deviation of x and y, σ _xyfor the co-variation heteromerism of x and y, C ₁, C ₂, C ₃be all constant, in order to maintain the stable of l (x, y), C (x, y), S (x, y);

Choosing sampling window corresponding to maximum SSIM value is that final trace location is the target area A of next frame _n.

2. method according to claim 1, is characterized in that, described method also comprises after being final trace location choosing sampling window that maximum SSIM is corresponding:

Obtain the region of search B of next frame _n; The target area A that wherein former frame tracks _n-1be positioned at region of search B _n，Qie region of search, center B _nlength be target area A _n-1length+2*SW (search width); Region of search B _nheight be target area A _n-1height+2*SH (search height); Wherein, n is the frame number of B ultrasonic image; A _n-1it is the target area of n-1 frame B ultrasonic image;

With B _nthe point in the upper left corner in region is starting point, at B _nregion obtains A _n-1the information of the sampling window of size; By setpoint distance, move sampling window, every movement once, is obtained the information of a sampling window, i.e. sampled signal y _n;

Use structural similarity Modulus Model to calculate echo signal x and sampled signal y _nsSIM value;

Choosing sampling window corresponding to maximum SSIM value is that final trace location is the target area A of next frame _n; N is greater than 2 integer.

3. method according to claim 1, is characterized in that, in the method for described tracking, described target area becomes regular figure from irregular figure.

4. method according to claim 1, is characterized in that, described use structural similarity Modulus Model SSIM zoning A ₁echo signal x and the SSIM value of sampled signal y specifically comprise:

A ₁echo signal x and A ₁sampled signal y be divided into the fritter of n*n, calculate the SSIM of corresponding blocks, the SSIM of all corresponding blocks is added up and is total ssim, be i.e. the SSIM value of echo signal x and sampled signal y;

$\mathrm{total}\_\mathrm{ssim}=\underset{i,j<n}{\mathrm{\&Sigma;}}\mathrm{SSIM}({I1}_{\mathrm{ij}},{I2}_{\mathrm{ij}})*W_{\mathrm{ij}};$

Wherein, I1 _ijrepresent the corresponding fritter of echo signal x, I2 _ijrepresent the corresponding fritter of sampled signal y, w represents weighted value.

5. method according to claim 1, is characterized in that, described method also comprises the hunting zone of adaptive updates next frame after choosing maximum SSIM value:

When maximum SSIM is larger, when SSIM is greater than first threshold, reduce the value of SW and SH; When maximum SSIM is less, be greater than Second Threshold, while being less than first threshold, increase the value of SW and SH; When SSIM is very little, while being less than Second Threshold, enter alarm program, stop following the trail of.

Images