CN102663720B - Image splicing method based on minimum mean square error criterion - Google Patents
Image splicing method based on minimum mean square error criterion Download PDFInfo
- Publication number
- CN102663720B CN102663720B CN201210092941.3A CN201210092941A CN102663720B CN 102663720 B CN102663720 B CN 102663720B CN 201210092941 A CN201210092941 A CN 201210092941A CN 102663720 B CN102663720 B CN 102663720B
- Authority
- CN
- China
- Prior art keywords
- registration
- region
- pixel
- square error
- overlapping region
- 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
Abstract
The invention provides an image splicing method based on a minimum mean square error criterion, relating to a high precision image splicing method, belonging to the field of image processing technology. A purpose of the invention is applying the minimum mean square error criterion to realize the registration of best overlap regions with high accuracy. The method comprises the following steps of: carrying out rough registration; sliding a registration area; calculating epsilon [L*D], sliding an image splicing area around a rough registration splicing range, and calculating the mean square error value epsilon [L*D] of an overlapping area successively; carrying out epsilon [min] updating, updating a large value with a minimum mean square error, and recording a corresponding overlapping area; repeating the operation in a sliding range, wherein when L is not equal to m or D is not equal to n, epsilon [L*D] is larger than epsilon [opt]; when a mean square error of an examine overlapping area shows an obvious valley, expressing an area reaching a minimum pixel difference, and finally using the overlapping area corresponding to the minimum mean square error epsilon [min]= epsilon [opt] as an optimal overlapping area X[M-m:M-1,N-n:n-1] and Y[0:m-1,0:n-1] of registration; and carrying out area ratio Gauss transition. The method is suitable for the image splicing of two-dimensional planar graphs with a same pixel density degree, and is applied to splicing of continuously shot images of a same image acquisition device.
Description
Technical field
The present invention relates to a kind of high precision image joining method, belong to technical field of image processing.
Background technology
Image Mosaics, as an important branch of image processing techniques, is applied to the various fields that need to obtain high resolving power or wide visual angle image.For current acquisition technology, there will be due to the different faint difference of brightness of image causing of acquisition angles.Therefore in the splicing of two high-definition pictures, cannot find in the overlapping region, border of two pictures the desirable pixel region overlapping, therefore leave obvious splicing vestige or crack, affect visual effect.Image Mosaics comprises the registration of image and two processes of the fusion of image.The method of the image registration generally adopting is at present the method for extract minutiae.
Summary of the invention
The object of this invention is to provide a kind of image split-joint method based on minimum mean square error criterion, realize the best overlapping region of high registration accuracy to apply minimum mean square error criterion.The present invention only proposes for being spliced into condition between 2 d plane picture and the identical picture of pixel density degree, is applied to the splicing that completes to same image capture device continuously shot images.
The present invention solves the problems of the technologies described above the technical scheme of taking to be:
A kind of image split-joint method based on minimum mean square error criterion of the present invention: set image X and the Y that two resolution are M × N and splice, its best overlapping region is the region of a m × n; Setting these two picture formats is black and white BMP file, and pixel span is X
i, j, Y
i, j∈ 0 ..., 255}, (i=0,1 ... M-1, j=0,1 ... N-1); Take the image upper left corner as X
0,0, the lower right corner is X
m-1, N-1, best overlapping region can be expressed as in X and Y: X
m-m:M-1, N-n:N-1and Y
0:m-1,0:n-1;
Described method realizes according to following steps:
Step 2, slip registration region: near the Image Mosaics region of sliding thick Registration and connection scope;
Step 3, calculating ε
l × D:
When investigation overlapping region is (as shown in Figure 1) L × D size, the expression formula of square error is:
Near the Image Mosaics region of sliding thick Registration and connection scope, even X
m-L, N-Dand Y
l, Dmeet:
and
and successively calculate the square mean error amount ε of overlapping region
l × D;
Step 4, ε
minupgrade: upgrade large value with little value square error, and record corresponding overlapping region; In sliding scale, repeat this operation, in the time of L ≠ m or D ≠ n, have ε
l × D> ε
opt; In the time that obvious valley appears in the square error of investigation overlapping region, represent to arrive the region of minimum pixel difference, finally with minimum value square error ε
min=ε
optthe best overlapping region X that corresponding overlapping region is registration
m-m:M-1, N-n:N-1and Y
0:m-1,0:n-1;
, in the time that registration region is best overlapping region, square error minimum is:
X
m-m+i, N-n+jfor pixel X
m-m+i, N-n+jpixel value, y
i, jfor pixel Y
i, jpixel value;
Step 5, Area Ratio Gauss transition: determine behind optimal registration region, for the each pixel structure Gaussian distribution model corresponding with pixel value in registration region, calculate the upper left bottom right Area Ratio of each pixel in registration overlapping region, and the mode of approaching according to Area Ratio is determined the pixel value of pixel.
The invention has the beneficial effects as follows:
The present invention mainly emphasizes to apply minimum mean square error criterion and realizes the best overlapping region of high registration accuracy.Image Mosaics comprises the registration of image and two processes of the fusion of image.Different from the method registration of the extract minutiae generally adopting at present, the present invention completes the location in Image Mosaics region take least mean-square error (MMSE) criterion on the area of investigation overlapping region as registration decision rule, avoid the generation in crack; The inventive method application minimum mean square error criterion has been realized the best overlapping region of high registration accuracy; And the method that adopts transition to merge to the pixel of splicing regions.And then from visually eliminating the splicing vestige causing due to luminance difference XOR difference in exposure, greatly improve the joining quality of image.
Accompanying drawing explanation
Fig. 1 is the splicing schematic diagram of the inventive method, and Fig. 2 is the procedure chart of realizing Area Ratio Gauss transition, and Fig. 3 is the inventive method FB(flow block).
Embodiment
Embodiment one: as shown in Figures 1 to 3, a kind of image split-joint method based on minimum mean square error criterion described in present embodiment: set image X and the Y that two resolution are M × N and splice, its best overlapping region is the region of a m × n; Setting these two picture formats is black and white BMP file, and pixel span is X
i, j, Y
i, j∈ 0 ..., 255}, (i=0,1 ... M-1, j=0,1 ... N-1); Take the image upper left corner as X
0,0, the lower right corner is X
m-1, N-1, best overlapping region can be expressed as in X and Y: X
m-m:M-1, N-n:N-1and Y
0:m-1,0:n-1;
Described method realizes according to following steps:
Thick registration can be realized by the mode of feature point extraction; The mode of feature point extraction is prior art category;
Step 2, slip registration region: near the Image Mosaics region of sliding thick Registration and connection scope;
Step 3, calculating ε
l × D:
When investigation overlapping region is (as shown in Figure 1) L × D size, the expression formula of square error is:
Near the Image Mosaics region of sliding thick Registration and connection scope, even X
m-L, N-Dand Y
l, Dmeet:
and
and successively calculate the square mean error amount ε of overlapping region
l × D;
Step 4, ε
minupgrade: upgrade large value with little value square error, and record corresponding overlapping region; In sliding scale, repeat this operation, in the time of L ≠ m or D ≠ n, have ε
l × D> ε
opt; In the time that obvious valley appears in the square error of investigation overlapping region, represent to arrive the region of minimum pixel difference, finally with minimum value square error ε
min=ε
optthe best overlapping region X that corresponding overlapping region is registration
m-m:M-1, N-n:N-1and Y
0:m-1,0:n-1;
, in the time that registration region is best overlapping region, square error minimum is:
X
m-m+i, N-n+jfor pixel X
m-m+i, N-n+jpixel value, y
i, jfor pixel Y
i, jpixel value;
Step 5, Area Ratio Gauss transition: determine behind optimal registration region, for the each pixel structure Gaussian distribution model corresponding with pixel value in registration region, calculate the upper left bottom right Area Ratio of each pixel in registration overlapping region, and the mode of approaching according to Area Ratio is determined the pixel value of pixel.
The registration process of step 1 to four image, step 5 is the fusion process of image.By investigating the square error situation of change in sliding area, search best overlapping region, realize high-precision Image Mosaics.
Embodiment two: as shown in Figure 3, in present embodiment, the detailed process of the Area Ratio Gauss transition described in step 5 is:
Step 5 (one), definition overlapping region Z
m × non pixel be Z
l, d, in image X and Y, the pixel value of corresponding pixel points is respectively x
m-m+l, N-n+dand y
l, d; Through pixel Z
l, dthe separatrix that is m/n as slope, separatrix top left region area is S
x, separatrix lower right area area is S
y; Now crossing separatrix Area Ratio is S
x/ S
y;
Step 5 (two), the transition of employing Gauss model make the luminance transition in Image Mosaics region more level and smooth, find respective pixel value z in Gaussian distribution figure
l, d, make z
l, dleft side area Z
xwith right side area Z
ymeet:
Z
X/Z
Y→S
X/S
Y (4)
In formula, " → " represents to approach;
Step 5 (three), can determine overlapping region Z after completing above-mentioned two steps
m × nat Z
l, dlocational pixel value; The variance of Gaussian distribution can be determined according to being spliced the luminance difference of picture, and the average of Gaussian distribution is μ
z=(x
m-m+l, N-n+d+ y
l, d)/2;
Step 5 (four), repeat above-mentioned steps, can determine thus the pixel value of all pixels in registration overlapping region.Other step is identical with embodiment one.
Principle of work:
Square error in the communications field for weighing the estimated performance of receiving terminal to transmitting terminal signal.If transmitting terminal signal is t, receiving terminal to the estimated value of transmitting terminal signal is
the estimated value square error ε of receiving end is expressed as:
Wherein E{} represents the expectation of element or vector.The square error of the signal of receiving terminal estimated value and transmitting terminal is less, the information sequence that demodulates transmitting terminal of the easier zero defect of receiving end.Be that square error is the important indicator that affects the receiving terminal bit error rate.
Judgment condition take minimum mean square error criterion as registration, is considered as transmitting terminal signal and receiving terminal signal by two width picture region of registration region.The overlapping region of mobile two pictures, investigate the square mean error amount of two pixels corresponding on the area of overlapping region, in the time there is minimum value in square mean error amount, represent the difference minimum of two width pictures on registration overlapping region now, therefore judgement is the registration overlapping region of two pictures.
Take two resolution shown in Fig. 1 as M × N image X and Y be spliced into example, its best overlapping region is the region of a m × n.If these two picture formats are black and white BMP file, pixel span is X
i, j, Y
i, j∈ 0 ..., 255}, (i=0,1 ... M-1, j=0,1 ... N-1).Take the image upper left corner as X
0,0, the lower right corner is X
m-1, N-1, best overlapping region can be expressed as in X and Y: X
m-m:M-1, N-n:N-1and Y
0:m-1,0:n-1., in the time that registration region is best overlapping region, square error minimum is:
X
m-m+i, N-n+jfor pixel X
m-m+i, N-n+jpixel value, y
i, jfor pixel Y
i, jpixel value.When investigating overlapping region for L as shown in Figure 1 × D is when big or small, the expression formula of square error is:
, in the time of L ≠ m or D ≠ n, there is ε
l × D> ε
opt, therefore, in the time that obvious valley appears in the square error of investigation overlapping region, represent to arrive the region of minimum pixel difference, therefore registration region is best overlapping region.
Behind the best overlapping region of registration, the pixel of overlapping region, due to luminance difference, must adopt the suitable excessive method of pixel.Traditional Gauss's class transition algorithm can reasonablely address this problem.Because common splicing regions is two dimension splicing, therefore adopt Gauss's transition method of dividing based on Area Ratio.The Gauss who divides based on Area Ratio is the position in registration region according to pixel excessively, divide its upper left and lower right area by the size of registration region rectangle, and in Gaussian distribution curve, finding suitable pixel, the left and right area of changing the time is than the Area Ratio that should approach this pixel upper left and bottom right pixel point in registration region.With the overlapping region Z shown in Fig. 2
m × non pixel Z
l, dfor example, now in X and Y, the pixel value of corresponding pixel points is respectively x
m-m+l, N-n+dand y
l, d.Cross and be Z
l, dslope is the separatrix of m/n, and separatrix top left region area is S
x, separatrix lower right area area is S
y.Now crossing separatrix Area Ratio is S
x/ S
y.
For making the brightness in Image Mosaics region excessively more level and smooth, adopt Gauss model excessive.Can in brachymemma Gaussian distribution figure, find as shown in Figure 2 respective pixel value z
l, d, make z
l, dleft side area Z
xwith right side area Z
ymeet:
Z
X/Z
Y→S
X/S
Y (4)
And then determine at Z
l, dlocational pixel value.The variance of brachymemma Gaussian distribution can be determined according to being spliced the luminance difference of picture, and the average of Gaussian distribution is μ
z=(x
m-m+l, N-n+d+ y
l, d)/2.Can determine thus the pixel value of all pixels in registration overlapping region.
Claims (1)
1. the image split-joint method based on minimum mean square error criterion, sets image X and the Y that two resolution are M × N and splices, and its best overlapping region is the region of a m × n; Setting these two picture formats is black and white BMP file, and pixel span is X
i,j, Y
i,j∈ 0 ..., 255}, (i=0,1 ... M-1, j=0,1 ... N-1); Take the image upper left corner as X
0,0, the lower right corner is X
m-1, N-1, best overlapping region can be expressed as in X and Y: X
m-m:M-1, N-n:N-1and Y
0:m-1,0:n-1; Described method realizes according to following steps: step 1, thick registration: the splicing regions of image X and Y is completed to thick registration, tentatively determine that Image Mosaics region sliding scale is:
Wherein,
for the minimum value of default m,
for the maximal value of default m,
for the minimum value of default n,
for the maximal value of default n;
Step 2, slip registration region: near the Image Mosaics region of sliding thick Registration and connection scope;
Step 3, calculating ε
l × D:
When investigation overlapping region is (as shown in Figure 1) L × D size, the expression formula of square error is:
Near the Image Mosaics region of sliding thick Registration and connection scope, even, X
m-L, N-Dand Y
l, Dmeet:
and
and successively calculate the square mean error amount ε of overlapping region
l × D;
Step 4, ε
minupgrade: upgrade large value with little value square error, and record corresponding overlapping region; In sliding scale, repeat this operation, in the time of L ≠ m or D ≠ n, have ε
l × D> ε
opt; In the time that obvious valley appears in the square error of investigation overlapping region, represent to arrive the region of minimum pixel difference, finally with minimum value square error ε
min=ε
optthe best overlapping region X that corresponding overlapping region is registration
m-m:M-1, N-n:N-1and Y
0:m-1,0:n-1;
, in the time that registration region is best overlapping region, square error minimum is:
X
m-m+i, N-n+jfor pixel X
m-m+i, N-n+jpixel value, y
i,jfor pixel Y
i,jpixel value;
Step 5, Area Ratio Gauss transition: determine behind optimal registration region, for the each pixel structure Gaussian distribution model corresponding with pixel value in registration region, calculate the upper left bottom right Area Ratio of each pixel in registration overlapping region, and the mode of approaching according to Area Ratio is determined the pixel value of pixel;
It is characterized in that: the detailed process of described Area Ratio Gauss transition is:
(1), definition overlapping region Z
m × non pixel be Z
l,d, in image X and Y, the pixel value of corresponding pixel points is respectively x
m-m+l, N-n+dand y
l,d; Through pixel Z
l,dthe separatrix that is m/n as slope, separatrix top left region area is S
x, separatrix lower right area area is S
y; Now crossing separatrix Area Ratio is S
x/ S
y;
(2), adopt Gauss model transition to make the luminance transition in Image Mosaics region more level and smooth, in Gaussian distribution figure, find respective pixel value z
l,d, make z
l,dleft side area Z
xwith right side area Z
ymeet:
Z
X/Z
Y→S
X/S
Y (4)
In formula, " → " represents to approach;
(3), can determine overlapping region Z after completing above-mentioned two steps
m × nat Z
l,dlocational pixel value; The variance of Gaussian distribution can be determined according to being spliced the luminance difference of picture, and the average of Gaussian distribution is μ
z=(x
m-m+l, N-n+d+ y
l,d)/2;
(4), repeat above-mentioned steps, can determine thus the pixel value of all pixels in registration overlapping region.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210092941.3A CN102663720B (en) | 2012-03-31 | 2012-03-31 | Image splicing method based on minimum mean square error criterion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210092941.3A CN102663720B (en) | 2012-03-31 | 2012-03-31 | Image splicing method based on minimum mean square error criterion |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102663720A CN102663720A (en) | 2012-09-12 |
CN102663720B true CN102663720B (en) | 2014-06-04 |
Family
ID=46773199
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210092941.3A Active CN102663720B (en) | 2012-03-31 | 2012-03-31 | Image splicing method based on minimum mean square error criterion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102663720B (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104184950B (en) * | 2014-09-10 | 2017-12-08 | 北京奇艺世纪科技有限公司 | A kind of video image joining method and device |
CN105554449B (en) * | 2015-12-11 | 2018-04-27 | 浙江宇视科技有限公司 | A kind of method and device for being used to quickly splice camera review |
CN105844683B (en) * | 2016-03-23 | 2019-02-05 | 深圳市富途网络科技有限公司 | The implementation method of pixel difference frame-by-frame cartoons based on Canvas and WebWorker |
CN105976319A (en) * | 2016-05-06 | 2016-09-28 | 安徽伟合电子科技有限公司 | Boundary reproduction method applied to image splicing |
CN108510439B (en) * | 2017-02-28 | 2019-08-16 | 贝壳找房(北京)科技有限公司 | Joining method, device and the terminal of point cloud data |
US11847766B2 (en) | 2018-08-22 | 2023-12-19 | Genemind Biosciences Company Limited | Method and device for detecting bright spots on image, and computer program product |
US20210217186A1 (en) * | 2018-08-22 | 2021-07-15 | Genemind Biosciences Company Limited | Method and device for image registration, and computer program product |
EP3843033A4 (en) | 2018-08-22 | 2021-11-24 | GeneMind Biosciences Company Limited | Method for constructing sequencing template based on image, and base recognition method and device |
CN111598177A (en) * | 2020-05-19 | 2020-08-28 | 中国科学院空天信息创新研究院 | Self-adaptive maximum sliding window matching method facing low-overlapping image matching |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6018349A (en) * | 1997-08-01 | 2000-01-25 | Microsoft Corporation | Patch-based alignment method and apparatus for construction of image mosaics |
CN101052100A (en) * | 2007-03-29 | 2007-10-10 | 上海交通大学 | Multiple exposure image intensifying method |
CN101221661A (en) * | 2008-01-29 | 2008-07-16 | 深圳市迅雷网络技术有限公司 | Image registration method and device |
-
2012
- 2012-03-31 CN CN201210092941.3A patent/CN102663720B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6018349A (en) * | 1997-08-01 | 2000-01-25 | Microsoft Corporation | Patch-based alignment method and apparatus for construction of image mosaics |
CN101052100A (en) * | 2007-03-29 | 2007-10-10 | 上海交通大学 | Multiple exposure image intensifying method |
CN101221661A (en) * | 2008-01-29 | 2008-07-16 | 深圳市迅雷网络技术有限公司 | Image registration method and device |
Non-Patent Citations (4)
Title |
---|
《图像拼接技术研究与实现》;谭康;《南京理工大学硕士论文》;20070131;第18-21页,第41-42页,图3.1,图3.2(a),图3.2(b),图3.4,图3.5,图4.1 * |
《基于SIFT特征遥感影像自动配准与拼接》;程焱等;《遥感技术与应用》;20081231;第23卷(第6期);第721-728页 * |
程焱等.《基于SIFT特征遥感影像自动配准与拼接》.《遥感技术与应用》.2008,第23卷(第6期),第721-728页. |
谭康.《图像拼接技术研究与实现》.《南京理工大学硕士论文》.2007,第1-60页. |
Also Published As
Publication number | Publication date |
---|---|
CN102663720A (en) | 2012-09-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102663720B (en) | Image splicing method based on minimum mean square error criterion | |
Vaudrey et al. | Differences between stereo and motion behaviour on synthetic and real-world stereo sequences | |
CN111612763A (en) | Mobile phone screen defect detection method, device and system, computer equipment and medium | |
CN103325095B (en) | A kind of openness image repair method of sample block of the bonding position factor | |
CN104867135A (en) | High-precision stereo matching method based on guiding image guidance | |
CN104850850A (en) | Binocular stereoscopic vision image feature extraction method combining shape and color | |
CN108596975A (en) | A kind of Stereo Matching Algorithm for weak texture region | |
CN110335199A (en) | A kind of image processing method, device, electronic equipment and storage medium | |
CN103293168B (en) | Fruit surface defect detection method based on visual saliency | |
CN114820579A (en) | Semantic segmentation based image composite defect detection method and system | |
CN104851089A (en) | Static scene foreground segmentation method and device based on three-dimensional light field | |
CN116206133A (en) | RGB-D significance target detection method | |
CN114120272A (en) | Multi-supervision intelligent lane line semantic segmentation method fusing edge detection | |
CN111523439B (en) | Method, system, device and medium for target detection based on deep learning | |
CN113159043A (en) | Feature point matching method and system based on semantic information | |
CN109858349A (en) | A kind of traffic sign recognition method and its device based on improvement YOLO model | |
CN103167247B (en) | A kind of video sequence color image joining method | |
CN116543227A (en) | Remote sensing image scene classification method based on graph convolution network | |
CN117197624A (en) | Infrared-visible light image fusion method based on attention mechanism | |
Si et al. | A comprehensive benchmark analysis for sand dust image reconstruction | |
Quan et al. | CrackViT: a unified CNN-transformer model for pixel-level crack extraction | |
CN111881914B (en) | License plate character segmentation method and system based on self-learning threshold | |
Shit et al. | An encoder‐decoder based CNN architecture using end to end dehaze and detection network for proper image visualization and detection | |
CN109801273B (en) | Light field image quality evaluation method based on polar plane linear similarity | |
CN116071721A (en) | Transformer-based high-precision map real-time prediction method and system |
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 |