Summary of the invention
The purpose of this invention is to provide and a kind ofly utilize ultrasound wave to measure Meromelia bone and skin to form two dimensional image and carry out feature extracting methods, the methods such as Gypsum Fibrosum complex method precision is low, CT scan that solved have radiating weak point to human body, can measure the deformed limb profile and the inner skeleton of skin simultaneously.
The technical solution adopted in the present invention is: adopt ultrasonic measuring device, B ultrasonic diagnostic apparatus, ultrasonic image work station and graphic system to form ultrasonic measurement system, probe on the ultrasonic measuring device is connected with the B ultrasonic diagnostic apparatus, the B ultrasonic diagnostic apparatus is connected with graphic system with the ultrasonic image work station by data wire, it is characterized in that this method may further comprise the steps:
A. image acquisition
Probe moves in a circle around the center of rotation of the residual body interval with per 0 °<α °≤90 °, simultaneously ultrasonic measurement figure is gathered, stores by the ultrasonic image work station, obtain k=360 °/α frame B ultrasonic image, probe vertically moves down then, the center of rotation of residual body also moves down thereupon, probe is done wheel measuring around new center of rotation with the interval of per 0 °<α °≤90 ° again, obtain k=360 °/α frame B ultrasonic image, by that analogy, promptly obtain the multiframe two-dimensional section faultage image of tested residual each horizontal plane of body;
B. image 2-d reconstruction
The first step is determined the center of rotation of residual body earlier, determines its home position by the measurement standard column, if the center of circle overlaps with probe focusing center, then this center of circle promptly is a center of rotation;
Second step rotated to new coordinate system with the anglec of rotation α of k two field picture when measuring separately around center of rotation, and according to the rotation of coordinate principle, its conversion expression formula is:
In the following formula, (x y) is the horizontal vertical coordinate point of original image coordinate system, (u v) is the horizontal vertical coordinate point of postrotational image coordinate system, again following formula is write as matrix form and is:
Here, R
aBe the rotation transformation matrix, (x y) is the coordinate system of single-frame images, and (u v) is the coordinate system of combination picture;
The 3rd step superposeed to the k two field picture that rotates to reposition,
If the single frames measurement image is A
k, combination picture is B
K+1,
A wherein
Ij (l)Expression m * n matrix A
l, m=768, n=576,1≤i≤m, 1≤j≤n, 1≤l≤k+1;
The 4th step rounded pixel value, was expressed as C
K+1={ B
K+1, wherein { x} represents to surpass the smallest positive integral of x;
The 5th step denoising, threshold value is made as 128,
C
K+1Expression to the image pixel after compound round with denoising after image pixel matrix, c
IjExpression is through compound, after the denoising image pixel capable at i, the gray value that j lists;
C. image is handled, extracted the edge feature of skeleton and skin
With function f (then the grey scale change gradient of this point is for x, y) any gray scale of any in the presentation video:
is a gradient operator, and T represents transposition, and Gx and Gy are illustrated in the intensity of variation of the gray value on x and the y direction, and gradient is a vector, and its amplitude is
The edge strength of representing certain dot image, satisfy following three conditions and both can find the edge of image point:
With this point be edge strength maximum in 3 * 3 neighborhoods at center less than certain threshold values,
2. this edge strength is greater than the edge strength along two neighbor pixels of this gradient direction,
3. the phase value of this point
Value is less than 45 °,
Wherein, (x y), discrete change into f (i j), is designated as f each some f of image array
I, j, gradient G
x, G
yDiscrete turning to:
G
x=[f
i-1,j-1+2f
i,j-1+f
i+1,j-1]-[f
i-1,j+1+2
fi,j+1+f
i-1,j+1]
G
y=[f
i+1,j-1+2f
i+1,j+f
i+1,j+1]-[f
i-1,j-1+2f
i-1,j+f
i-1,j+1]
The template of operator is
The above step of process is just finished the feature extraction to the skeleton and the skin of image.
The invention has the beneficial effects as follows: the present invention utilizes the ultrasonic imaging principle, skeleton and skin to deformed limb carry out ultrasonic measurement, can measure skin outline and inner skeleton simultaneously, image acquisition wherein, both can obtain the image of different aspects, and on same aspect, can obtain multi-angle images, improved the degree of accuracy of measuring, compare with similar technology, greatly reduced cost, reduced radiation, the feature extraction of the reconstruction of image and skeleton and skin is that the foundation of the three-dimensional CAD of deformed limb is laid a good foundation.
The specific embodiment
Below in conjunction with accompanying drawing the present invention is elaborated.
Referring to Fig. 1, adopt ultrasonic measuring device 1, B ultrasonic diagnostic apparatus 4, ultrasonic image work station 5 and graphic system 6 to form ultrasonic measurement system, probe 2 on the ultrasonic measuring device 1 is connected with B ultrasonic diagnostic apparatus 4, B ultrasonic diagnostic apparatus 4 is connected with graphic system 6 with ultrasonic image work station 5 by data wire, it is characterized in that this method may further comprise the steps:
A. image acquisition
Probe 2 moves in a circle around the center of rotation of the residual body interval with per 0 °<α °≤90 °, simultaneously gather, store by 5 couples of ultrasonic measurement figure of ultrasonic image work station, obtain k=360 °/α frame B ultrasonic image, pop one's head in then and 2 vertically move down, the center of rotation of residual body also moves down thereupon, probe 2 is done wheel measuring around new center of rotation with the interval of per 0 °<α °≤90 ° again, obtain k=360 °/α frame B ultrasonic image, by that analogy, promptly obtain the multiframe two-dimensional section faultage image of tested residual each horizontal plane of body, referring to Fig. 2;
B. image 2-d reconstruction
The first step is determined the center of rotation of residual body earlier, determines its home position by the measurement standard column, if the center of circle overlaps with probe focusing center, then this center of circle promptly is a center of rotation;
Second step rotated to new coordinate system with the anglec of rotation α of k two field picture when measuring separately around center of rotation, and according to the rotation of coordinate principle, its conversion expression formula is:
In the following formula, (x y) is the horizontal vertical coordinate point of original image coordinate system, (u v) is the horizontal vertical coordinate point of postrotational image coordinate system, again following formula is write as matrix form and is:
Wherein
Here, R
aBe the rotation transformation matrix, (x y) is the coordinate system of single-frame images, and (u v) is the coordinate system of combination picture;
The 3rd step superposeed to the k two field picture that rotates to reposition,
If the single frames measurement image is A
k, combination picture is B
K+1,
A wherein
Ij (l)Expression m * n matrix A
l, m=768, n=576,1≤i≤m, 1≤j≤n, 1≤l≤k+1;
The 4th step rounded pixel value, was expressed as C
K+1={ B
K+1, wherein { x} represents to surpass the smallest positive integral of x;
The 5th step denoising, threshold value is made as 128,
C
K+1Expression to the image pixel after compound round with denoising after image pixel matrix, c
IjExpression is through compound, after the denoising image pixel capable at i, the gray value that j lists;
C. image is handled, extracted the edge feature of skeleton and skin
With function f (then the grey scale change gradient of this point is for x, y) any gray scale of any in the presentation video:
is a gradient operator, and T represents transposition, and Gx and Gy are illustrated in the intensity of variation of the gray value on x and the y direction, and gradient is a vector, and its amplitude is
The edge strength of representing certain dot image, satisfy following three conditions and both can find the edge of image point:
With this point be edge strength maximum in 3 * 3 neighborhoods at center less than certain threshold values,
2. this edge strength is greater than the edge strength along two neighbor pixels of this gradient direction,
3. the phase value of this point
Value is less than 45 °,
Wherein, (x y), discrete change into f (i j), is designated as f each some f of image array
I, j, gradient G
x, G
yDiscrete turning to:
G
x=[f
i-1,j-1+2f
i,j-1+f
i+1,j-1]-[f
i-1,j+1+2f
i,j+1+f
i-1,j+1]
G
y=[f
i+1,j-1+2f
i+1,j+f
i+1,j+1]-[f
i-1,j-1+2f
i-1,j+f
i-1,j+1]
The template of operator is
The above step of process is just finished the feature extraction to the skeleton and the skin of image.
Check the precision of ultrasonic measurement and the precision of image characteristics extraction by the measurement standard column
1) interior contour accuracy analysis
With diameter of Carnis Sus domestica embedding fluorothene cylindrical rod that is 30mm, be placed in the liquid storing barrel 3, the setting anglec of rotation is 9 ° it is done ultrasonic measurement, adopt the method for above-mentioned image 2-d reconstruction that the k=40 two field picture that obtains is rebuild, it shown in Fig. 3 a the interior cylindrical cross-section image of rebuilding, cylinder profile in extracting, recording diameter is 30.2mm.Getting center of rotation is the center of circle, and diameter is 30 works, one standard rounds, and shown in Fig. 3 b, dotted line is a standard round, and solid line is the interior cylinder profile line that extracts, and contour line and standard round are very identical.Both maximum deviations are in the bottom, and maximum deviation and scale error see Table 1.
2) outline precision analysis
Measuring a diameter is 90mm polyamine fat column, adopt the method for above-mentioned image 2-d reconstruction that its image is rebuild, the result carries out profile to image and extracts shown in Fig. 4 a, measures its outline diameter size and is 89mm, getting center of rotation is the center of circle, diameter is 90 works, one standard rounds, and shown in Fig. 4 b, dotted line is a standard round, solid line is the outside cylinder contour line that extracts, and contour line and standard round are very identical.Both maximum deviations are in the lower-left side, and maximum deviation and scale error see Table 1.
Table 1 inside and outside contour precision analysis result
Measuring object | Diameter dimension | Actual measurement circle and standard round+0.8 maximum deviation distance ± 1.01 |
Standard value | Measured value | Relative error |
Interior profile | ??30mm | ?30.2mm | ????
Outward=0.7%
| ????±1.01 ????
In=
|
Outline | ??90mm | ??89mm | ????
In=1.1%
| ????
Outward=
|
The image total error | ????1.1% | |
Comprehensive 1), 2) image outline that can obtain by ultrasonic measurement and feature extraction and the maximum deviation of profile in kind apart from δ=± 1.01mm, size relative error ε=1.1%, as shown in table 1.Above-mentioned error result is rebuild the CAD 3D of deformed limb not to be influenced, and ultrasonic measurement produces δ and ε is acceptable.
Embodiment 1
As measuring object, method is as follows to select the fresh pig foreleg of peeling off from live body (the about 240mm of mean radius, highly about 160mm) for use:
A. image acquisition
Earlier the pig foreleg is immersed in the liquid storing barrel 3 of storage full water, the lower end places on the pallet blended rubber mud to fix, and demarcates center of rotation, measure the anglec of rotation and be set at 9 °, probe moves down distance for 5mm, on each horizontal direction, probe is made wheel measuring around one week of the leg of pork counterclockwise with per 9 ° interval, obtain k=360 °/9 °=40 two field pictures, pixel size is 768 * 576, on the vertical direction, probe moves down 5mm at every turn, survey 24 horizontal planes altogether, can record 24 * 40 two field pictures so altogether, referring to Fig. 5;
B. image 2-d reconstruction
The first step is determined the center of rotation of residual body earlier, determines its home position by the measurement standard column, if the center of circle overlaps with probe focusing center, then this center of circle promptly is a center of rotation;
Second step rotated to new coordinate system with 9 ° of the anglec of rotation of k two field picture when measuring separately around center of rotation, and according to the rotation of coordinate principle, its conversion expression formula is:
In the following formula, (x y) is the horizontal vertical coordinate point of original image coordinate system, (u v) is the horizontal vertical coordinate point of postrotational image coordinate system, again following formula is write as matrix form and is:
Here, R
aBe the rotation transformation matrix, (x y) is the coordinate system of single-frame images, and (u v) is the coordinate system of combination picture;
The 3rd step superposeed to 40 two field pictures that rotate to reposition,
If the single frames measurement image is A
k, combination picture is B
K+1,
A wherein
Ij (l)Expression m * n matrix A
l, m=768, n=576,1≤i≤m, 1≤j≤n, 1≤l≤k+1;
The 4th step rounded pixel value, was expressed as C
K+1={ B
K+1, wherein { x} represents to surpass the smallest positive integral of x;
The 5th step denoising, threshold value is made as 128,
C
K+1Expression to the image pixel after compound round with denoising after image pixel matrix, c
IjExpression is through compound, after the denoising image pixel capable at i, the gray value that j lists;
C. image is handled, extracted the edge feature of skeleton and skin
With function f (then the grey scale change gradient of this point is for x, y) any gray scale of any in the presentation video:
is a gradient operator, and T represents transposition, and Gx and Gy are illustrated in the intensity of variation of the gray value on x and the y direction, and gradient is a vector, and its amplitude is
The edge strength of representing certain dot image, satisfy following three conditions and both can find the edge of image point:
With this point be edge strength maximum in 3 * 3 neighborhoods at center less than certain threshold values,
2. this edge strength is greater than the edge strength along two neighbor pixels of this gradient direction,
3. the phase value of this point
Value is less than 45 °,
Wherein, (x y), discrete change into f (i j), is designated as f each some f of image array
I, j, gradient G
x, G
yDiscrete turning to:
G
x=[f
i-1,j-1+2f
i,j-1+f
i+1,j-1]-[f
i-1,j+1+2f
i,j+1+f
i-1,j+1]
G
y=[f
i+1,j-1+2f
i+1,j+f
i+1,j+1]-[f
i-1,j-1+2f
i-1,j+f
i-1,j+1]
The template of operator is
The above step of process is just finished the feature extraction to the skeleton and the skin of image.
Embodiment 2
Testee is with embodiment 1, and method is as follows:
A. image acquisition, earlier the pig foreleg is immersed in the liquid storing barrel 3 of storage full water, the lower end places on the pallet blended rubber mud to fix, demarcate center of rotation, measure the anglec of rotation and be set at 90 °, probe moves down distance and is 3mm, on each horizontal direction, probe is made wheel measuring around one week of the leg of pork counterclockwise with per 90 ° interval, obtains K=360 °/90 °=4 two field pictures, and pixel size is 768 * 576, on the vertical direction, probe moves down 3mm at every turn, surveys 45 horizontal planes altogether, can record 45 * 4 two field pictures so altogether.
Following steps are with embodiment 1.
Embodiment 3
Testee is with embodiment 1, and method is as follows:
A. image acquisition, earlier the pig foreleg is immersed in the liquid storing barrel 3 of storage full water, the lower end places on the pallet blended rubber mud to fix, demarcate center of rotation, measure the anglec of rotation and be set at 45 °, probe moves down distance and is 4mm, on each horizontal direction, probe is made wheel measuring around one week of the leg of pork counterclockwise with per 45 ° interval, obtains K=360 °/45 °=8 two field pictures, and pixel size is 768 * 576, on the vertical direction, probe moves down 4mm at every turn, surveys 35 horizontal planes altogether, can record 35 * 8 two field pictures so altogether.
Following steps are with embodiment 1.
Embodiment 4
Testee is with embodiment 1, and method is as follows:
A. image acquisition, earlier the pig foreleg is immersed in the liquid storing barrel 3 of storage full water, the lower end places on the pallet blended rubber mud to fix, demarcate center of rotation, measure the anglec of rotation and be set at 30 °, probe moves down distance and is 5mm, on each horizontal direction, probe is made wheel measuring around one week of the leg of pork counterclockwise with per 30 ° interval, obtains K=360 °/30 °=12 two field pictures, and pixel size is 768 * 576, on the vertical direction, probe moves down 5mm at every turn, surveys 25 horizontal planes altogether, can record 25 * 12 two field pictures so altogether.
Following steps are with embodiment 1.
Embodiment 5
Testee is with embodiment 1, and method is as follows:
A. image acquisition, earlier the pig foreleg is immersed in the liquid storing barrel 3 of storage full water, the lower end places on the pallet blended rubber mud to fix, demarcate center of rotation, measure the anglec of rotation and be set at 60 °, probe moves down distance and is 6mm, on each horizontal direction, probe is made wheel measuring around one week of the leg of pork counterclockwise with per 60 ° interval, obtains K=360 °/60 °=6 two field pictures, and pixel size is 768 * 576, on the vertical direction, probe moves down 6mm at every turn, surveys 8 horizontal planes altogether, can record 8 * 6 two field pictures so altogether.
Following steps are with embodiment 1.