CN102081737B - Multi-scale modeling method for pneumatic heat radiation images and application thereof - Google Patents
Multi-scale modeling method for pneumatic heat radiation images and application thereof Download PDFInfo
- Publication number
- CN102081737B CN102081737B CN201010614054A CN201010614054A CN102081737B CN 102081737 B CN102081737 B CN 102081737B CN 201010614054 A CN201010614054 A CN 201010614054A CN 201010614054 A CN201010614054 A CN 201010614054A CN 102081737 B CN102081737 B CN 102081737B
- Authority
- CN
- China
- Prior art keywords
- heat radiation
- image
- under
- pneumatic
- fingerprint
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Radiation Pyrometers (AREA)
- Image Input (AREA)
Abstract
The invention discloses a multi-scale modeling method for pneumatic heat radiation images, which comprises the following steps of: (1) registering each pneumatic heat radiation deterioration image, and solving the difference between each deterioration image and a reference image; (2) fitting each difference image in a full image region, namely under a first scale to obtain a fitting curved surface under the first scale; and (3) thinning the last fitting scale, and performing multi-scale curved surface approximation fitting on the difference image dk to obtain a curved surface polynomial fitted under each scale, namely forming a window heat radiation fingerprint library of pneumatic heat radiation deterioration image sequences under corresponding pneumatic heat environments. The invention also discloses application of the method in image correction. The contrast of the corrected target region is obviously promoted, and the promotion of the contrast is more obvious together with the thinning of the scales; and the method can be widely applied in image correction.
Description
Technical field
The invention belongs to the interdisciplinary science technical field that Pneumatic optical combines with Flame Image Process, be specifically related to multiple dimensioned modeling method of a kind of pneumatic heat radiation and the application in image rectification thereof.
Background technology
Pneumatic optical is that the subject of flow field to the influence of high-speed aircraft imaging detection streamed in research at a high speed.When the high-speed aircraft that has an optical imagery detection system flew in the endoatmosphere, the interaction between optical window and the incoming flow formed complicated flow field.Because the effect of air viscosity will be blocked with the contacted air-flow in optical window surface, make gas velocity reduce, and near window surface, form the boundary layer.Each layer that has very big velocity gradient in the boundary layer can produce strong friction, and the kinetic energy of air-flow irreversibly becomes heat energy, causes the rising of window wall surface temperature.High temperature gas flow will constantly conduct heat to the low temperature wall, cause very strong pneumatic heating.Optical window is in the serious pneumatic thermal environment by pneumatic heating, produces heat radiated noise, reduces the signal to noise ratio (S/N ratio) and the picture quality of Photodetection system.
Flying speed is big more, and air-flow is just serious more in the degree of aircraft surface heating.The irradiance superposition of the irradiance of the irradiance of air-flow and window and background outside window, imaging sensor will get into inelastic region or saturated, and that causes the scenery effective information loses or the reduction of signal to noise ratio (S/N ratio), signal to noise ratio the decline of detection performance or disabler.Therefore, need carry out pneumatic heat radiation and proofread and correct, to improve signal to noise ratio (S/N ratio).
Because pneumatic thermal-radiating degradation model is the unknown and random variation, degraded image also contains sensor noise, has increased the difficulty that image recovers or proofreaies and correct, and does not also have pertinent literature to report the bearing calibration of pneumatic heat radiation degraded image at present.
Summary of the invention
The present invention proposes the multiple dimensioned modeling method of a kind of pneumatic heat radiation image, this method utilizes least square approximation that the degradation characteristics of pneumatic heat radiation image is carried out modeling from a plurality of yardsticks, obtains the window heat radiation fingerprint under certain pneumatic thermal environment.The invention allows for a kind of window heat radiation fingerprint that utilizes above-mentioned modeling to obtain pneumatic heat radiation degeneration image sequence is carried out method of correcting, can proofread and correct recovery to pneumatic heat radiation image effectively, improve the signal to noise ratio (S/N ratio) and the picture quality of image.
Among the present invention, the optical window that is in the pneumatic thermal environment has the rule along with temperature and pressure change, is called window heat radiation fingerprint.
The concrete steps of the multiple dimensioned modeling method of a kind of pneumatic heat radiation image provided by the invention comprise:
(1) obtains the pneumatic heat radiation degeneration image sequence (f of (being under uniform temperature and the pressure conditions) under certain pneumatic thermal environment through imaging device
1, f
2, f
3..., f
N-2, f
N-1, f
n), and with every two field picture in the image sequence all with benchmark image f
0Divide into groups in pairs, each picture group picture constitutes the base conditioning object of proofreading and correct and recovering computing:
(f
0，f
1)，(f
0，f
2)，(f
0，f
3)，......，(f
0，f
n-1)，(f
0，f
n)
N is the frame number of degeneration image sequence, f
kFor being T in temperature
k, pressure is P
kCondition under the pneumatic heat radiation degraded image that obtains, k is the frame number of pneumatic heat radiation degraded image;
(2) to each picture group picture (f
0, f
k) carry out registration, obtain the image combination (f behind the registration
0, f '
k) and off-set value (Vx
k, Vy
k), Vx
kBe the side-play amount on the x direction of principal axis, Vy
kBe the side-play amount on the y direction of principal axis;
(3) to each the picture group picture (f behind the registration
0, f '
k), ask the difference d of its pneumatic heat radiation degraded image and benchmark image
k=f '
k-f
0
(4) utilize the least square approximation principle, to error image d
kCarry out multiple dimensioned curved surface and approach match, obtain the curved surface polynomial expression of match under each yardstick
Be window heat radiation fingerprint, scale=max, mid, min represents large scale (i.e. first yardstick), mesoscale (i.e. second yardstick) and small scale (i.e. the 3rd yardstick) respectively, and p, q are the highest polynomial power power, if the bicubic polynomial expression is then got p=q=3; Be specially:
Large scale least square approximation: establish pneumatic heat radiation degraded image size and be N
Max* M
Max, to the N in the full figure zone
Max* M
MaxIndividual difference point is sampled, and gets N '
Max* M '
MaxIndividual difference point (x
u, y
v) (u=0,1 ..., N '
Max-1; V=0,1 ..., M '
Max-1) carries out the polynomial surface match of large scale, obtain the curved surface polynomial expression of match under the large scale
It is the heat radiation fingerprint under the large scale;
Mesoscale least square approximation: analyze the heat radiation fingerprint under the large scale that obtains
Further Meso-Scale Analysis is carried out in zone to error is relatively large, and the size of establishing selected mesoscale zone is N
Mid* M
Mid, to the N in the zone
Mid* M
MidIndividual difference point is sampled, and gets N '
Mid* M '
MidIndividual difference point (x
u, y
v) (u=0,1 ..., N '
Mid-1; V=0,1 ..., M '
Mid-1) carries out the polynomial surface match of mesoscale, obtain the curved surface polynomial expression of match under the mesoscale
It is the heat radiation fingerprint under the mesoscale;
Small scale least square approximation: analyze the heat radiation fingerprint under the mesoscale that obtains
Further small scale analysis is carried out in the relatively large zone of error, and the size of establishing selected small scale zone is N
Min* M
Min, to the N in the zone
Min* M
MinIndividual difference point is sampled, and gets N '
Min* M '
MinIndividual difference point (x
u, y
v) (u=0,1 ..., N '
Min-1; V=0,1 ..., M '
Min-1) carries out the polynomial surface match of small scale, obtain the curved surface polynomial expression of match under the small scale
It is the heat radiation fingerprint under the small scale;
Heat radiation fingerprint under the small scale that analysis obtains
is if still exist the relatively large zone of error; Then can copy above-mentioned steps that the more analyses of small scale such as microscale are carried out in the big zone of selected error, obtain the heat radiation fingerprint under the corresponding scale;
Comprehensive said process can obtain the light shaft offset amount (Vx under certain pneumatic thermal environment
k, Vy
k) and a plurality of yardsticks such as large scale, mesoscale, small scale under the heat radiation fingerprint
(scale=max, mid min), can set up multiple dimensioned pneumatic heat radiation image rectification fingerprint base.
The pneumatic heat radiation image rectification fingerprint base that utilizes said method to obtain carries out the method for image rectification, is specially:
(1) the pneumatic heat radiation degeneration image sequence (g of input
1, g
2, g
3..., g
M-2, g
M-1, g
m), m is the frame number of degeneration image sequence, g
kBe temperature T
k, pressure P
kPneumatic heat radiation degraded image under the condition, k is the frame number of pneumatic heat radiation degraded image;
(2) to pneumatic heat radiation degraded image g arbitrarily
k, in pneumatic heat radiation image rectification fingerprint base, obtain temperature T
kWith pressure P
kThe heat radiation fingerprint of the corresponding scale of the selection under the condition
(scale=max, mid, min) and corresponding light shaft offset amount (Vx
k, Vy
k), pneumatic heat radiation degraded image is proofreaied and correct, obtain its correcting result g ' under a plurality of yardsticks
K, scale
Pneumatic heat radiation degraded image before and after proofreading and correct is carried out target area contrast analysis relatively, and the target area contrast after can finding to proofread and correct obviously promotes, and along with the refinement of yardstick, contrast promotes more obvious, has proved the validity of this method.
Description of drawings
Fig. 1 is the pneumatic thermoradiation efficiency synoptic diagram of high-speed aircraft optical window;
Fig. 2 is the process flow diagram of setting up of the multiple dimensioned pneumatic heat radiation fingerprint base of the present invention;
Fig. 3 is the multiple dimensioned pneumatic heat radiation method for correcting image process flow diagram of the present invention;
Fig. 4 is the pneumatic heat radiation image of a present invention multiscale analysis process flow diagram;
Fig. 5 is the process flow diagram of setting up of the multiple dimensioned pneumatic heat radiation fingerprint of the present invention;
Fig. 6 (a) is the benchmark image of pneumatic heat radiation degeneration image sequence;
Fig. 6 (b) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
1Two field picture;
Fig. 6 (c) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
2Two field picture;
Fig. 6 (d) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
3Two field picture;
Fig. 6 (e) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
4Two field picture;
Fig. 6 (f) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
5Two field picture;
Fig. 6 (g) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
6Two field picture;
Fig. 6 (h) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
7Two field picture;
Fig. 6 (i) is the f of the pneumatic heat radiation degeneration image sequence in the modeling method of the present invention
8Two field picture;
Fig. 7 is the target position of form center curve of deviation figure behind the pneumatic heat radiation degraded image registration among Fig. 6 (b)～6 (i);
Fig. 8 (a) is the 3-D display of the difference of benchmark image among the middle degraded image of the Fig. 6 (b) behind the registration and Fig. 6 (a);
Fig. 8 (b) is the 3-D display of the difference of benchmark image among the middle degraded image of the Fig. 6 (c) behind the registration and Fig. 6 (a);
Fig. 8 (c) is the 3-D display of the difference of benchmark image among the middle degraded image of the Fig. 6 (d) behind the registration and Fig. 6 (a);
Fig. 8 (d) is the 3-D display of the difference of benchmark image among the middle degraded image of the Fig. 6 (e) behind the registration and Fig. 6 (a);
Fig. 8 (e) is the 3-D display of the difference of benchmark image among the middle degraded image of the Fig. 6 (f) behind the registration and Fig. 6 (a);
Fig. 8 (f) is the 3-D display of the difference of benchmark image among the middle degraded image of the Fig. 6 (g) behind the registration and Fig. 6 (a);
Fig. 8 (g) is the 3-D display of the difference of benchmark image among the middle degraded image of the Fig. 6 (h) behind the registration and Fig. 6 (a);
Figure (8h) is the 3-D display of the difference of benchmark image among the middle degraded image of the figure (6i) behind the registration and Fig. 6 (a);
Fig. 9 (a) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (b);
Fig. 9 (b) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (c);
Fig. 9 (c) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (d);
Fig. 9 (d) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (e);
Fig. 9 (e) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (f);
Fig. 9 (f) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (g);
Fig. 9 (g) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (h);
Fig. 9 (h) is the 3-D display of the heat radiation fingerprint of degraded image under large scale among Fig. 6 (i);
Figure 10 (a) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (b);
Figure 10 (b) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (c);
Figure 10 (c) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (d);
Figure 10 (d) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (e);
Figure 10 (e) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (f);
Figure 10 (f) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (g);
Figure 10 (g) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (h);
Figure 10 (h) is the 3-D display of the heat radiation fingerprint of degraded image under mesoscale among Fig. 6 (i);
Figure 11 (a) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (b);
Figure 11 (b) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (c);
Figure 11 (c) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (d);
Figure 11 (d) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (e);
Figure 11 (e) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (f);
Figure 11 (f) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (g);
Figure 11 (g) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (h);
Figure 11 (h) is the 3-D display of the heat radiation fingerprint of degraded image under small scale among Fig. 6 (i);
Figure 12 is that the data overlapping region synoptic diagram that obtains is suitably extended on the isoplanatic region border outward;
Figure 13 suitably extends the data overlapping region synoptic diagram on the one dimension direction that obtains outward with the isoplanatic region border;
The actual pneumatic heat radiation degraded image of Figure 14 for needing to proofread and correct;
Figure 15 (a) is the heat radiation correcting image of pneumatic heat radiation degraded image under large scale among Figure 14;
Figure 15 (b) is the heat radiation correcting image of pneumatic heat radiation degraded image under mesoscale among Figure 14;
Figure 15 (c) is the heat radiation correcting image of pneumatic heat radiation degraded image under small scale among Figure 14;
Figure 16 (a) is the synoptic diagram of choosing of target area ((170,53), (237,124));
Figure 16 (b) carries out the result that zone (Figure 16 (a)) contrast is estimated to the correcting image of pneumatic heat radiation degraded image under a plurality of yardsticks among Figure 14; 5 scales of horizontal ordinate are represented benchmark image (Fig. 6 (a)), pneumatic heat radiation degraded image (Figure 14), degraded image correcting image (Figure 15 (a)), the correcting image (Figure 15 (b)) under the mesoscale and the correcting image (Figure 15 (c)) under the small scale under large scale from left to right successively, the region contrast that the ordinate presentation video is corresponding;
Figure 17 (a) is the synoptic diagram of choosing of target area ((103,154), (145,195));
Figure 17 (b) carries out the result that zone (Figure 17 (a)) contrast is estimated to the correcting image of pneumatic heat radiation degraded image under a plurality of yardsticks among Figure 14; 5 scales of horizontal ordinate are represented benchmark image (Fig. 6 (a)), pneumatic heat radiation degraded image (Figure 14), degraded image correcting image (Figure 15 (a)), the correcting image (Figure 15 (b)) under the mesoscale and the correcting image (Figure 15 (c)) under the small scale under large scale from left to right successively, the region contrast that the ordinate presentation video is corresponding.
Embodiment
Below in conjunction with accompanying drawing and specific embodiment the present invention is done further detailed explanation.
The multiple dimensioned modeling method of a kind of pneumatic heat radiation image comprises the steps:
(1) obtains the pneumatic heat radiation degeneration image sequence (f of (being under uniform temperature and the pressure conditions) under certain pneumatic thermal environment through imaging device
1, f
2, f
3..., f
N-2, f
N-1, f
n), and with itself and benchmark image f
0Divide into groups in pairs, each picture group picture constitutes the base conditioning object of proofreading and correct and recovering computing:
(f
0，f
1)，(f
0，f
2)，(f
0，f
3)，......，(f
0，f
n-1)，(f
0，f
n)
N is the frame number of degeneration image sequence, f
kFor being T in temperature
k, pressure is P
kCondition under the pneumatic heat radiation degraded image that obtains, k is the frame number of pneumatic heat radiation degraded image.
Respectively with the f of pneumatic heat radiation degeneration image sequence
1Frame degraded image and benchmark image, f
2Frame degraded image and benchmark image divide into groups in pairs, and the pneumatic heat radiation degraded image of back is done similar processing.
(2) to each picture group picture (f
0, f
k) carry out registration, obtain the image combination (f behind the registration
0, f '
k) and off-set value (Vx
k, Vy
k), Vx
kBe the side-play amount on the x direction of principal axis, Vy
kBe the side-play amount on the y direction of principal axis.
(3) to each the picture group picture (f behind the registration
0, f '
k), ask the difference d of its pneumatic heat radiation degraded image and benchmark image
k=f '
k-f
0
For the first picture group picture (f behind the registration
0, f '
1), obtain the f behind the registration
1Pneumatic heat radiation degraded image of frame and benchmark image f
0Difference d
1=f '
1-f
0, by that analogy, for the k picture group picture (f behind the registration
0, f '
k), the f behind the registration
kPneumatic heat radiation degraded image of frame and benchmark image f
0Difference be d
k=f '
k-f
0, obtain the error image sequence (d of pneumatic heat radiation degraded image and benchmark image thus
1, d
2, d
3..., d
N-2, d
N-1, d
n).
(4) utilize the least square approximation principle, to error image d
kCarry out multiple dimensioned curved surface and approach match, obtain the curved surface polynomial expression of match under each yardstick
Scale=max, mid, min represents large scale, mesoscale and small scale respectively, and p, q are the highest polynomial power power, if the bicubic polynomial expression is then got p=q=3;
According to step (4.1) to (4.4) to error image d
kCarry out surface fitting, its detailed process is:
(4.1) to the N * M in the image rectangular area point (x
u, y
v) (u=0,1 ..., N-1; V=0,1 ..., M-1; ) on functional value z
Uv, make the least square fitting polynomial expression do
Wherein, p, q are the highest polynomial power power;
(4.2) fixing y; X is constructed M least square fitting polynomial expression:
wherein;
(i=0; 1; ..., p) be mutually orthogonal polynomial expression, construct by following recursion formula:
Order
Have
β
i=η
i/ η
I-1, can get according to the principle of least square:
(4.3) the least square fitting polynomial expression of structure y:
Wherein, ψ
j(y) (j=0,1 ..., q) be mutually orthogonal polynomial expression, construct by following recursion formula:
ψ
0(y)＝1
ψ
1(y)＝y-α′
0
ψ
j(y)＝(y-α′
j)ψ
j-1(y)-β′
jψ
j-2(y)
Order
Have
β '
j=δ
j/ δ
J-1, can get according to the principle of least square:
(4.4) result who derives in integrating step (4.2), (4.3) can obtain the polynomial expression of surface fitting:
The polynomial form that converts standard to is:
(5) large scale least square approximation: establish the image size and be N
Max* M
Max, to the N in the full figure zone
Max* M
MaxIndividual difference point is sampled, and gets N '
Max* M '
MaxIndividual difference point (x
u, y
v) (u=0,1 ..., N '
Max-1; V=0,1 ..., M '
Max-1), carries out the polynomial surface match of large scale, obtain the curved surface polynomial expression of match under the large scale according to step (4.1) to (4.4)
Be degraded image f
kHeat radiation fingerprint under large scale.
(6) mesoscale least square approximation: analyze the heat radiation fingerprint under the large scale that obtains
Further Meso-Scale Analysis is carried out in zone to error is relatively large, and the size of establishing selected mesoscale zone is N
Mid* M
Mid, to the N in the zone
Mid* M
MidIndividual difference point is sampled, and gets N '
Mid* M '
MidIndividual difference point (x
u, y
v) (u=0,1 ..., N '
Mid-1; V=0,1 ..., M '
Mid-1) carries out the polynomial surface match of mesoscale, obtain the curved surface polynomial expression of match under the mesoscale
Be degraded image f
kHeat radiation fingerprint under mesoscale;
According to step (6.1) to (6.3) pneumatic heat radiation degraded image is carried out the analysis of mesoscale, its detailed process is:
(6.1) with the heat radiation fingerprint under the large scale that obtains in the step (5)
Be divided into M
2* M
2Sub-block (for Meso-Scale Analysis, M
2Generally getting 2～4 gets final product), each sub-piece is used
(s is the sequence number of sub-piece, s=1,2, L, M
2* M
2) expression, same, with the error image d that obtains in the step (3)
kCorrespondence is divided into M
2* M
2Sub-block, each sub-piece is used d
K, mid (s)Expression can be similar in each sub-piece and regard isoplanatic region as, and certain pixel value is suitably extended on the isoplanatic region border outward;
(6.2) N in the calculating full figure zone
Max* M
MaxIndividual point (x
u, y
v) (u=0,1 ..., N
Max-1; V=0,1 ..., M
MaxThe relative error of the large scale heat radiation fingerprint-1)
And N in each sub-piece zone
Mid (s)* M
Mid (s)Individual point (x
u, y
v) (u=0,1 ..., N
Mid (s)-1; V=0,1 ..., M
Mid (s)The relative error of the large scale heat radiation fingerprint-1)
|| be absolute value sign, if
The mesoscale heat radiation fingerprint of then corresponding sub-piece
If
Then to the sub-piece d of difference
K, mid (s)N in the zone
Mid (s)* M
Mid (s)Individual difference point is sampled, and gets N '
Mid (s)* M '
Mid (s)Individual difference point (x
u, y
v) (u=0,1 ..., N '
Mid (s)-1; V=0,1 ..., M '
Mid (s)-1), carries out the polynomial surface match of mesoscale, obtain the curved surface polynomial expression of match under the mesoscale according to step (4.1) to (4.4)
The promptly corresponding heat radiation fingerprint of sub-piece under mesoscale;
(6.3) with stitching algorithm with the heat radiation fingerprint of each sub-piece under mesoscale
(s=1,2, L, M
2* M
2) be stitched together, obtain degraded image f
kHeat radiation fingerprint under mesoscale
Its detailed process is: according to the distance structure weighting coefficient of each sub-piece overlay region pixel to the border, use the overlay region data to accomplish the splicing of gradual change, stitching image is visual to isolate sense to remove.
Shown in figure 12, [1] [2], zone [3] be not have certain one-level yardstick model under and the part of other region overlappings, need not do other processing, directly use initial value to get final product, and regional [4] are the laps in two zones, and regional [5] are the laps in 4 zones; In the process of splicing and stack, adopt weighted-average method, be that example is come this weighting coefficient of simplified illustration with overlapping on the one dimension direction.Adjacent two image blocks are represented with X, Y respectively image represented that with Z shown in figure 13, the size of X, Y is respectively W after both spliced
X, W
Y, L extends in the border outward in the block district of image boundary, and border, promptly block district is at the W of X
X-L row are at the L of Y row.X, Y have removed the l row that have blocky effect through after the image correction process, the overlay region width 2d=2 (L-l) of two image blocks that are left so, and promptly the border, bulk district of this moment is at the W of X
X-d row, at the d of Y row, the weight coefficient of each transitional element of both sides doubling of the image district is pressed 50%/d=1/ (2d) recursion.
(7) small scale least square approximation: analyze the heat radiation fingerprint under the mesoscale that obtains
Further small scale analysis is carried out in the relatively large zone of error, and the size of establishing selected small scale zone is N
Min* M
Min, to the N in the zone
Min* M
MinIndividual difference point is sampled, and gets N '
Min* M '
MinIndividual difference point (x
u, y
v) (u=0,1 ..., N '
Min-1; V=0,1 ..., M '
Min-1) carries out the polynomial surface match of small scale, obtain the curved surface polynomial expression of match under the small scale
Be degraded image f
kHeat radiation fingerprint under small scale;
According to step (7.1) to (7.3) pneumatic heat radiation degraded image is carried out the analysis of small scale, its detailed process is:
(7.1) with the heat radiation fingerprint under the mesoscale that obtains in the step (6)
Be divided into M
3* M
3Sub-block is (for small scale analysis, M
3Generally getting 4～8 gets final product), each sub-piece is used
(s is the sequence number of sub-piece, s=1,2, L, M
3* M
3) expression, same, with the error image d that obtains in the step (3)
kCorrespondence is divided into M
3* M
3Sub-block, each sub-piece is used d
K, min (s)Expression can be similar in each sub-piece and regard isoplanatic region as, and certain pixel value is suitably extended on the isoplanatic region border outward;
(7.2) N in the calculating full figure zone
Max* M
MaxIndividual point (x
u, y
v) (u=0,1 ..., N
Max-1; V=0,1 ..., M
MaxThe relative error of the mesoscale heat radiation fingerprint-1)
And N in each sub-piece zone
Min (s)* M
Min (s)Individual point (x
u, y
v) (u=0,1 ..., N
Min (s)-1; V=0,1 ..., M
Min (s)The relative error of the mesoscale heat radiation fingerprint-1)
|| be absolute value sign, if
The small scale heat radiation fingerprint of then corresponding sub-piece
If
Then to the sub-piece d of difference
K, min (s)N in the zone
Min (s)* M
Min (s)Individual difference point is sampled, and gets N '
Min (s)* M '
Min (s)Individual difference point (x
u, y
v) (u=0,1 ..., N '
Min (s)-1; V=0,1 ..., M '
Min (s)-1), carries out the polynomial surface match of small scale, obtain the curved surface polynomial expression of match under the small scale according to step (4.1) to (4.4)
The promptly corresponding heat radiation fingerprint of sub-piece under small scale;
(7.3) according to the stitching algorithm in the step (6.3) with the heat radiation fingerprint of each sub-piece under small scale
(s=1,2, L, M
3* M
3) be stitched together, obtain degraded image f
kHeat radiation fingerprint under small scale.
(8) analyze heat radiation fingerprint
under the small scale obtain if still there is the relatively large zone of error; Then can carry out the more analyses of small scale such as microscale to the big zone of selected error, obtain the heat radiation fingerprint under the corresponding scale according to above-mentioned steps.
(9) comprehensive step (2), (5), (6), (7), (8) can obtain the light shaft offset amount (Vx under certain pneumatic thermal environment
k, Vy
k) and a plurality of yardsticks such as large scale, mesoscale, small scale under the heat radiation fingerprint, set up multiple dimensioned pneumatic heat radiation image rectification fingerprint base.
The pneumatic heat radiation image rectification fingerprint base that utilizes said method to obtain carries out the method for image rectification, and concrete steps are:
(1) the pneumatic heat radiation degeneration image sequence (g of input
1, g
2, g
3..., g
M-2, g
M-1, g
m), m is the frame number of degeneration image sequence, g
kBe temperature T
k, pressure P
kPneumatic heat radiation degraded image under the condition, k is the frame number of pneumatic heat radiation degraded image;
(2) to pneumatic heat radiation degraded image g arbitrarily
k, in pneumatic heat radiation image rectification fingerprint base, obtain temperature T
kWith pressure P
kThe heat radiation fingerprint of a plurality of yardsticks under the condition
(scale=max, mid, min) and corresponding light shaft offset amount (Vx
k, Vy
k), pneumatic heat radiation degraded image is proofreaied and correct, obtain its correcting result g ' under a plurality of yardsticks
K, scale
According to step (2.1) to (2.2) pneumatic heat radiation degraded image is carried out multiple dimensioned correction, its detailed process is:
(2.1) utilize the light shaft offset amount (Vx that obtains
k, Vy
k), obtain the pneumatic heat radiation degraded image g ' behind the registration
k(x, y)=g
k(x-Vx
k, y-Vy
k);
(2.2) as required pneumatic heat radiation degraded image is carried out multiple dimensioned correction, if to correcting image less demanding on local detail, then can be to pneumatic heat radiation degraded image g
kCarry out the correction of large scale, with g '
kDeduct the large scale heat radiation fingerprint that from above-mentioned fingerprint base, obtains
Obtain pneumatic heat radiation degraded image g
kCorrecting image under large scale
If to correcting image having relatively high expectations on local detail, then can be to pneumatic heat radiation degraded image g
kCarry out the correction of mesoscale or small scale, with g '
kDeduct the mesoscale heat radiation fingerprint that from above-mentioned fingerprint base, obtains
Obtain pneumatic heat radiation degraded image g
kCorrecting image under mesoscale
Or with g '
kDeduct the small scale heat radiation fingerprint that from above-mentioned fingerprint base, obtains
Obtain pneumatic heat radiation degraded image g
kCorrecting image under small scale
Pneumatic heat radiation degraded image before and after proofreading and correct is carried out target area contrast analysis relatively, and the target area contrast after can finding to proofread and correct obviously promotes, and along with the refinement of yardstick, contrast promotes more obvious, has proved the validity of this method.
Claims (8)
1. multiple dimensioned modeling method of pneumatic heat radiation image; The curved surface that carries out under a plurality of yardsticks through the error image to each image in the pneumatic heat radiation degeneration image sequence approaches match; Thereby obtain the window heat radiation fingerprint base under the corresponding pneumatic thermal environment, these method concrete steps comprise:
(1) respectively each pneumatic heat radiation degraded image is carried out registration according to benchmark image, and ask the difference of each degraded image and benchmark image, obtain separately light shaft offset amount and error image;
(2) each error image is promptly carried out match under first yardstick in the full figure zone, obtain the fitting surface under this first yardstick, be the window heat radiation fingerprint under this first yardstick;
(3) the match yardstick to the last time carries out refinement; Be about to last fitting surface and be divided into a plurality of segmented areas; Calculate the error of fitting of each segmented areas; If wherein the error of fitting of arbitrary segmented areas is not less than the error threshold value under the preset current refinement yardstick, then this segmented areas is carried out match once more, obtain the fitting surface of this segmented areas; Be the window heat radiation fingerprint of this segmented areas under this current refinement yardstick; And then obtain the window heat radiation fingerprint of whole pneumatic heat radiation degraded image under this refinement yardstick, repeat said process, until the error of fitting of the segmented areas of institute's refinement all less than it till error threshold value under the corresponding refinement yardstick; Otherwise the window heat radiation fingerprint under the last match yardstick is promptly as the window heat radiation fingerprint of pneumatic heat radiation degraded image under this refinement yardstick;
Through above-mentioned steps, obtain the window heat radiation fingerprint under a plurality of yardsticks, promptly constitute the window heat radiation fingerprint base of pneumatic heat radiation degeneration image sequence under corresponding pneumatic thermal environment.
2. method according to claim 1 is characterized in that, said match detailed process is: the difference point of said error image on the corresponding zone of match yardstick sampled, and the difference point that obtains to sample carries out the polynomial surface match and approaches.
3. method according to claim 1 and 2; It is characterized in that; The error of fitting of said segmented areas refers to the relative error on this segmented areas, after promptly the difference of corresponding each point takes absolute value in the fitting surface each point under the last match yardstick on this segmented areas and the error image and with corresponding benchmark image on the ratio of each point sum.
4. method according to claim 1 and 2; It is characterized in that; The error threshold value of said current refinement yardstick refers to, carries out each segmented areas behind piecemeal error of fitting sum and ratio of said segmented areas number under last once match yardstick under the current refinement yardstick.
5. method according to claim 1 and 2; It is characterized in that; In the said step (3); Said error of fitting is the window heat radiation fingerprint that this zone obtains in the match of last once yardstick less than the window heat radiation fingerprint of the segmented areas of error threshold value, and said whole pneumatic heat radiation degraded image is spliced through the window heat radiation fingerprint of each segmented areas under this refinement yardstick at the window heat radiation fingerprint under this refinement yardstick each other.
6. method according to claim 1 and 2 is characterized in that, the said least square fitting that fits to.
7. application rights requires the multiple dimensioned modeling method of pneumatic heat radiation image described in one of 1-6 that pneumatic heat radiation degraded image is carried out method of correcting, it is characterized in that, comprises the steps:
(1) input pneumatic heat radiation degraded image to be corrected;
(2) utilize the described method of one of claim 1-6 to obtain pneumatic heat radiation image rectification fingerprint base; From this fingerprint base, obtain heat radiation fingerprint and the corresponding light shaft offset amount of this degraded image under pneumatic thermal environment; Again this degraded image is proofreaied and correct; Promptly at first utilize the light shaft offset amount that this degraded image is carried out registration, the image behind the registration deducts the heat radiation fingerprint under the corresponding scale again, can obtain the correcting result under this corresponding scale.
8. method according to claim 7 is characterized in that said corresponding scale is confirmed according to the accuracy requirement of image to be corrected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010614054A CN102081737B (en) | 2010-12-30 | 2010-12-30 | Multi-scale modeling method for pneumatic heat radiation images and application thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010614054A CN102081737B (en) | 2010-12-30 | 2010-12-30 | Multi-scale modeling method for pneumatic heat radiation images and application thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102081737A CN102081737A (en) | 2011-06-01 |
| CN102081737B true CN102081737B (en) | 2012-10-10 |
Family
ID=44087692
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010614054A Active CN102081737B (en) | 2010-12-30 | 2010-12-30 | Multi-scale modeling method for pneumatic heat radiation images and application thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102081737B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093425B (en) * | 2012-12-31 | 2016-01-20 | 华中科技大学 | A kind of method for building up of pneumatic thermal radiation fingerprint map database and application thereof |
| CN105160687B (en) * | 2015-10-21 | 2018-01-02 | 华中科技大学 | A kind of Pneumatic optical heat radiated noise bearing calibration and system |
| CN105654430B (en) * | 2015-12-24 | 2017-04-19 | 华中科技大学 | Contrast constrained pneumatic thermal radiation correction method |
| CN110222301B (en) * | 2019-06-03 | 2020-09-04 | 中国科学院遥感与数字地球研究所 | Surface solar short wave radiation calculation method under haze condition |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101027679B (en) * | 2004-09-09 | 2010-04-21 | 奥普提克斯晶硅有限公司 | System and method for representing a general two dimensional spatial transformation |
| CN100349183C (en) * | 2004-11-18 | 2007-11-14 | 国家***第一海洋研究所 | Two-dimensional image data separation with multiple scale information |
| TWI353561B (en) * | 2007-12-21 | 2011-12-01 | Ind Tech Res Inst | 3d image detecting, editing and rebuilding system |
-
2010
- 2010-12-30 CN CN201010614054A patent/CN102081737B/en active Active
Non-Patent Citations (1)
| Title |
|---|
| 王瑞胡.基于多尺度曲面拟合的细胞图像表面分割算法.《计算机工程与设计》.2009,(第17期),4041-4043. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102081737A (en) | 2011-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10079975B2 (en) | Image distortion correction of a camera with a rolling shutter | |
| CN102081737B (en) | Multi-scale modeling method for pneumatic heat radiation images and application thereof | |
| CN102663693B (en) | Least square method-based adaptive radiation correction method for linear array push-broom image | |
| CN102254313B (en) | Image defogging method based on restoration and fusion of images on foggy days | |
| US10348985B2 (en) | Turbulence-free camera system and related method of image enhancement | |
| US11869218B2 (en) | System and method for camera calibration | |
| WO2017107343A1 (en) | Contrast constrained aerothermal radiation correction method | |
| CN101937561A (en) | Image recovery strengthening method for space-based wide-view field imaging point source/spot target | |
| JP6563050B2 (en) | System and method for image motion compensation | |
| CN104616247A (en) | Method for aerial photography map splicing based on super-pixels and SIFT | |
| US11935249B2 (en) | System and method for egomotion estimation | |
| CN106534614A (en) | Rapid movement compensation method of moving target detection under mobile camera | |
| CN103093425B (en) | A kind of method for building up of pneumatic thermal radiation fingerprint map database and application thereof | |
| CN103793891A (en) | Low-complexity panorama image joint method | |
| CN108109118B (en) | Aerial image geometric correction method without control points | |
| CN103337080A (en) | Registration technology of infrared image and visible image based on Hausdorff distance in gradient direction | |
| CN102651134B (en) | Constant-speed blurred image construction method and device based on splicing of two frames of static images | |
| CN103985089A (en) | Image stripe correction method in which weight edge analysis and intra-frame iteration are combined | |
| Nejadasl et al. | Optical flow based vehicle tracking strengthened by statistical decisions | |
| Luo et al. | Vision based displacement sensor with heat haze filtering capability | |
| US20090154771A1 (en) | Automated Determination of Cross-Stream Wind Velocity from an Image Series | |
| CN112241967B (en) | Target tracking method, device, medium and equipment | |
| Liu et al. | Analysis and modeling of aerothermal radiation based on experimental data | |
| CN107655507B (en) | A kind of focal plane arc joining method | |
| US20230230279A1 (en) | Single camera calibration |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |