CN105913407B - A method of poly focal power image co-registration is optimized based on differential chart - Google Patents
A method of poly focal power image co-registration is optimized based on differential chart Download PDFInfo
- Publication number
- CN105913407B CN105913407B CN201610210196.6A CN201610210196A CN105913407B CN 105913407 B CN105913407 B CN 105913407B CN 201610210196 A CN201610210196 A CN 201610210196A CN 105913407 B CN105913407 B CN 105913407B
- Authority
- CN
- China
- Prior art keywords
- frequency information
- image
- map2
- map1
- fusion
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Abstract
The present invention relates to a kind of methods optimized to multi-focus image fusion based on differential chart, be primarily based on it is multiple dimensioned it is multi-direction multiple focussing image is tentatively merged, the information of the remaining information and focal zone then measured by differential chart is updated initial blending image.It is multiple dimensioned to be merged with multi-direction two methods, obtained image-region accurate positioning, precision is high, suitable for the vision system of the mankind, being updated operation by the residual risk original fusion image of differential chart keeps blending image more accurate, image co-registration quality higher has wide practical use in military and civil field.
Description
Technical field
The invention belongs to image processing fields and information to merge field, more particularly to a kind of to be based on differential chart to multi-focus
The method for spending image co-registration optimization has wide practical use in military and civil field.
Background technology
Poly focal power image fusion technology is by the image co-registration of different focusing a to image, in conjunction with integrated information
Complementarity forms when property more than one, the image that multiple view information merges, obtains more comprehensive scene information.
Currently, multi-focus image fusion is broadly divided into two class of spatial domain and transform domain.Multiple focussing image based on spatial domain
Integration technology is divided into blending algorithm (such as Weighted Average Algorithm) and segmented areas blending algorithm for pixel, segmented areas fusion
What algorithm considered is contacting between the gray value of pixel and the point and adjacent pixel, the image definition that method obtains and
It will be better than pixel fusion method effect in terms of contrast.Compared with the integration technology based on spatial domain, based on transform domain
Integration technology is stronger to the zone location accuracy of image, precision higher.Image fusion technology of the multi-scale transform in transform domain
The irreplaceable key player of middle performer, wherein method the most famous is：Laplacian-pyramid method and discrete wavelet
It converts (DWT).But in both the above method, if for the location misalignment of source images, or go out in image acquisition process
Existing moving region, the syncretizing effect of image can significantly decline.So in order to solve this problem, and there is scholar to propose translation
Constant M8003 line dual-tree complex wavelet transform.In recent years, and it is proposed that some new method such as wavelet transforms
(DCT), non-down sampling contourlet transform (NSST) quaternion wavelet transformation (QWT).But transformation series is merged in fusion process
Number can replace original pixel value, multiple dimensioned variation that may become the main disadvantage in fusion method.
Another influences the fusion rule that fusion product qualitative factor is sub-band.In recent years, many scholars are directed to
A large amount of research has been done in this work.And achieve certain achievement in research.Most research is all based on the dilute of Multiple Spanning Tree (MST) region
It dredges and indicates total frame (SR).The method is to follow absolute maximum principle for high-frequency sub-band to merge, and low frequency sub-band is adopted
The method based on SR is taken to be merged.This approach avoid the offsets of smooth detail, but it remains unchanged, there is some defects, examples
Such as the inherent shortcoming of independent image feature.Then, in order to solve the problems, such as this, spatial frequency is had also been proposed with PCNN in non-lower sampling
The method being combined in contourlet transform domain, the method overcome the limit of labyrinth and excessive iterations in transformation
System, but still can not solve the problems, such as excessive artificial setup parameter.Therefore, and it is proposed that it is based on neighborhood characteristics NSCT
It converts, low frequency sub-band information follows the Weighted Fusion method based on neighboring region energy in this method, and high-frequency sub-band follows neighborhood spy
Sign is merged, but the method is concentrated and considers low frequency region energy information, but not for the extraction of effective coverage information
There is further processing.
Invention content
The present invention proposes a kind of method optimized to multi-focus image fusion based on differential chart, based on multiple dimensioned multi-direction
Multiple focussing image is tentatively merged, the information of the remaining information and focal zone then measured by differential chart is to initial
Blending image is updated, and the quality of obtained multi-focus image fusion is more preferable, precision higher.The technical solution adopted by the present invention
It comprises the following specific steps that：
Step 1：The pretreatment of image；
To source images IAAnd IBGaussian filtering is carried out, the average value of source images is calculated：
IAIndicate source images A, IBIndicate source images B, IAVGIndicate the mean value of source images A and B；
Step 2：Decomposition multi-direction to Image Multiscale obtains original fusion image；
Use neighborhood distance filter to source images I firstAAnd IBMulti-resolution decomposition is carried out, the number of plies is four layers, each layer of packet
Containing high-frequency information and low-frequency information；
Then the multi-direction decomposition methods of NSCT are used to carry out multi-direction decomposition to high-frequency information, wherein first layer decomposition direction is
16, the second layer is 8, and third layer is 4, last layer is 1, obtains IAHigh-frequency information IA,hWith low-frequency information IA,l, IB
High-frequency information IB,hWith low-frequency information IB,l；
Spatial frequency values of the pixel based on neighborhood in high-frequency information are finally calculated, the decision of high-frequency information fusion is constructed
Matrix tentatively merges high-frequency information by decision matrix, and low-frequency information, which is taken, averages, the high frequency letter after preliminary fusion
Low-frequency information after ceasing and averaging passes through the inverse transformation of multi-direction multi-scale transform, forms original fusion image IIN；
Step 3：Structure error image simultaneously calculates energy value；
Indicate source images IAWith mean value image IAVGIn the difference of pixel (i, j),
Indicate original fusion image IINWith mean value image IAVGIn the difference of pixel (i, j)；
Calculate energy value：
Indicate IIN(i, j) and IAVGThe difference of (i, j)Corresponding energy value,Indicate IAVG(i, j) and IAThe difference of (i, j)Corresponding energy value, M × N indicate pre-
If Size of Neighborhood, (i+m, j+n) indicates any point in neighborhood M × N centered on (i, j)；
Using two-sided filter to energy matrixWithProtect at the denoising of side
Reason, respectively obtains matrix char1 and char2；
Step 4：Construct initial bianry image；
(1) initial binary map MAP1, MAP2 are constructed by char1 and char2：
MAP1 (i, j) indicates that pixel (i, j) corresponding value in binary map MAP1, MAP2 (i, j) indicate pixel
(i, j) corresponding value in binary map MAP2, char1 (i, j) indicate pixel (i, j) corresponding value in char1,
Char2 (i, j) indicates that pixel (i, j) corresponding value in char2, δ are the threshold value of setting；
(2) bianry image is modified：
X × Y indicates that preset Size of Neighborhood, (i+a, j+b) indicate any one in neighborhood X × Y centered on (i, j)
Point；
MAP1 and MAP2 is modified respectively by above formula, obtains revised binary map MAP1' and MAP2', for
In source images focus and the equal unobvious area of non-focusing (i, j) | MAP1'(i, j)=MAP2'(i, j) handled, take MAP1'
(i, j)=MAP2'(i, j)=0.5；
Step 5：By modified binary map MAP1', MAP2' is to source images IAAnd IBIt is merged.
(1) fusion based on spatial domain：
More modified binary map MAP1', MAP2', if meeting MAP1'(i, j)=0, MAP2'(i, j)=1, then it selects
Source images IB(i, j) is filled, if meeting MAP1'(i, j) and=1, MAP2'(i, j)=0, then select source images IA(i, j) into
Row filling, other the case where then select original fusion image IIN(i, j) is filled, IF(i, j) is melting for last optimization generation
Close result；
(2) fusion based on transform domain：
More modified binary map MAP1', MAP2', if meeting MAP1'(i, j)=1, MAP2'(i, j)=0, then it selects
Source images IAHigh-frequency information IA,h(i, j) and low-frequency information IA,l(i, j) is the high-frequency information and low-frequency information of blending image, if
Meet MAP1'(i, j)=0, MAP2'(i, j)=1, then select source images IBHigh-frequency information IB,h(i, j) generation and low-frequency information
IB,l(i, j) is the high-frequency information and low-frequency information of blending image, if meeting MAP1'(i, j) and=MAP2'(i, j)=0.5, then it selects
Select original fusion image IINHigh-frequency information IIN, h(i, j) and low-frequency information IIN,l(i, j) be blending image high-frequency information and
Then low-frequency information carries out inverse transformation to the high and low frequency information of obtained blending image, obtains final fusion results IF。
Beneficial effects of the present invention
This method is updated blending image using differential chart, makes remaining multiple focussing image information more accurate
Blending image is updated, make that blending image is more accurate, quality higher.It is multiple dimensioned to be merged with multi-direction two methods,
Obtained image-region accurate positioning, precision is high, suitable for the vision system of the mankind, and multidirectional method compensates for multi-scale method
Some drawbacks, it is multi-direction to compensate for the case where some edge effective informations are lost in multi-scale image, pass through multi-direction fusion
To its edge, these important informations have obtained effective preservation to technology with lines.
Description of the drawings：
Fig. 1 is the flow chart of the method for the present invention；
Fig. 2 is the flow chart to form original fusion image；
Fig. 3 is the image co-registration flow chart based on spatial domain in the present invention；
Fig. 4 is the image co-registration flow chart based on transform domain in the present invention；
Fig. 5 is two groups of poly focal power source images to be fused, and (a) and (b) is that flower schemes, and (c) and (d) is lab figures；
Fig. 6 is fusion results of the flower figures under the method for the present invention and other four kinds of methods, wherein (a) and (b) is respectively
For the syncretizing effect figure proposed by the present invention based on spatial domain and transform domain, (c)-(f) is respectively to be adopted under four kinds of control methods are non-
Sample profile wave changes (NSCT), NSCT contrast enhancement process (NSCT-Co), neighborhood apart from (ND) and neighborhood apart from multi-direction side
The syncretizing effect figure of method (MMND)；
Fig. 7 is the differential chart of corresponding fusion figure and source images (b) in Fig. 6；
Fig. 8 is fusion results of the lab figures under the method for the present invention and other four kinds of methods, wherein (a) and (b) is respectively this
The syncretizing effect figure based on spatial domain and transform domain proposed is invented, (c)-(f) is respectively four kinds of control methods non-lower sampling wheels
Wide wave variation (NSCT), NSCT contrast enhancement process (NSCT-Co), neighborhood distance (ND) and neighborhood are apart from multi-direction method
(MMND) syncretizing effect figure；
Fig. 9 is the differential chart of corresponding blending image and source images (d) in Fig. 8.
Specific implementation mode
In the following with reference to the drawings and specific embodiments to the present invention is further illustrated.
Embodiment 1：As shown in Figure 1, the present invention is based on it is multiple dimensioned it is multi-direction multiple focussing image is tentatively merged, then pass through
Differential chart carries out spatial domain and transformation area update to initial blending image, obtains final blending image.As in Figure 2-4, have
Body step includes:
Step 1：The pretreatment of image：
To source images IAAnd IBGaussian filtering is carried out, the average value of source images is calculated：
IAIndicate source images A, IBIndicate source images B, IAVGIndicate the mean value of source images A and B；
Step 2：Decomposition multi-direction to Image Multiscale obtains original fusion image；
Use neighborhood distance filter to source images I firstAAnd IBMulti-resolution decomposition is carried out, the number of plies is four layers, each layer of packet
Containing high-frequency information and low-frequency information；
Then the multi-direction decomposition methods of NSCT are used to carry out multi-direction decomposition to high-frequency information, wherein first layer decomposition direction is
16, the second layer is 8, and third layer is 4, last layer is 1, obtains IAHigh-frequency information IA,hWith low-frequency information IA,l, IB
High-frequency information IB,hWith low-frequency information IB,l；
Spatial frequency values of the pixel based on neighborhood in high-frequency information are finally calculated, the decision of high-frequency information fusion is constructed
Matrix tentatively merges high-frequency information by decision matrix, and low-frequency information, which is taken, averages, the high frequency letter after preliminary fusion
Low-frequency information after ceasing and averaging passes through the inverse transformation of multi-direction multi-scale transform, forms original fusion image IIN；
Step 3：Structure error image simultaneously calculates energy value；
Indicate source images IAWith mean value image IAVGIn the difference of pixel (i, j),
Indicate original fusion image IINWith mean value image IAVGIn the difference of pixel (i, j)；
Calculate energy value：
Indicate IIN(i, j) and IAVGThe difference of (i, j)Corresponding energy value,Indicate IAVG(i, j) and IAThe difference of (i, j)Corresponding energy value, M × N indicate pre-
If Size of Neighborhood, (i+m, j+n) indicates any point in neighborhood M × N centered on (i, j)；M × N=in the present embodiment
11×11
Using two-sided filter to energy matrixWithIt carries out further
Processing, respectively obtain matrix char1 and char2.The effect of two-sided filter herein is the proximity and pixel to energy value
It is worth the compromise of similarity, while considers spatial information (si) and grey similarity, achievees the purpose that protect side denoising.Wherein bilateral filter
The parameter of wave device is set as, and window size is window=11 × 11, and Gauss variance is sigma=21.
Step 4：Construct initial bianry image；
(1) initial binary map MAP1, MAP2 are constructed by char1 and char2：
MAP1 (i, j) indicates that pixel (i, j) corresponding value in binary map MAP1, MAP2 (i, j) indicate pixel
(i, j) corresponding value in binary map MAP2, char1 (i, j) indicate pixel (i, j) corresponding value in char1,
Char2 (i, j) indicates that pixel (i, j) corresponding value in char2, δ be the threshold value set, δ in the present embodiment=
0.0025。
(2) and then again bianry image is modified：
X × Y indicates that preset Size of Neighborhood, (i+a, j+b) indicate any one in neighborhood X × Y centered on (i, j)
Point.
MAP1 and MAP2 is modified respectively by above formula, obtains revised binary map MAP1' and MAP2', herein
The purpose that place carries out pixel consistency checking is bad influence to be had to fusion process in the acnode of some mistake choosings, leakage choosing, therefore
By pixel consistency pixel is selected to ignore these mistakes.For region (i, j) | MAP1'(i, j)=MAP2'(i, j), i.e.,
Focusing and the equal unobvious region of non-focusing, handle these regions in two Zhang Yuan's images：MAP1'(i, j)=MAP2'(i,
J)=0.5.
Step 5：By modified binary map MAP1', MAP2' is to source images IAAnd IBIt is merged.
(1) fusion based on spatial domain：
More modified binary map MAP1', MAP2', if meeting MAP1'(i, j)=0, MAP2'(i, j)=1, then it selects
Source images IB(i, j) is filled, if meeting MAP1'(i, j) and=1, MAP2'(i, j)=0, then select source images IA(i, j) into
Row filling, other the case where then select original fusion image IIN(i, j) is filled.IF(i, j) is melting for last optimization generation
Close result；
(2) fusion based on transform domain：
More modified binary map MAP1', MAP2', if meeting MAP1'(i, j)=1, MAP2'(i, j)=0, then it selects
Source images IAHigh-frequency information IA,h(i, j) and low-frequency information IA,l(i, j) is the high-frequency information and low-frequency information of blending image, if
Meet MAP1'(i, j)=0, MAP2'(i, j)=1, then select source images IBHigh-frequency information IB,h(i, j) generation and low-frequency information
IB,l(i, j) is the high-frequency information and low-frequency information of blending image, if meeting MAP1'(i, j) and=MAP2'(i, j)=0.5, then it selects
Select original fusion image IINHigh-frequency information IIN, h(i, j) and low-frequency information IIN,l(i, j) be blending image high-frequency information and
Then low-frequency information carries out inverse transformation to the high and low frequency information of obtained blending image, obtains final fusion results IF。
Experimental result
For superiority of the comparative descriptions this method compared with conventional method, four kinds of control methods are selected here：It is adopted under non-
Sample profile wave changes (NSCT), NSCT contrast enhancement process (NSCT-Co), neighborhood apart from (ND) and neighborhood apart from multi-direction side
Method (MMND).
The one group of source images chosen in the present embodiment as shown in Figure 5 (a) and (b), are schemed for static flower.Fusion knot
Fruit is as shown in fig. 6, (a) and (b) respectively represents the optimization method (MMND-SD) in the present invention based on spatial domain and be based on transform domain
Optimization method (MMND-TD) fusion results.Fig. 6 is visually it does not appear that the effect of multi-focus region fusion, is
More more obvious with other methods, it is poor that blending image and source images are made, and selection differential chart visual effect is more apparent here
Source images (b) make difference.It is observed that (a) and (b) is that method syncretizing effect of the invention is good in Fig. 7 in red frame region,
Substantially the information of remaining source images is not observed in red frame.In remaining differential chart, source can be obviously observed all remaining
Image information (e) and in (f) obviously also includes the texture information of wall, image bottom right can be obviously observed in (c) and (d)
The information of angle blade is without without fusion completely.
Other than evaluating fusion results from subjective vision, objective evaluation index is additionally used here to carry out objective comment
Valence.Using interactive information MI, nonlinear correlation comentropy NCIE, based on multiple dimensioned measure Q_ in objective evaluation index
M, Chen Shi measures Q_CB.Index value is bigger, illustrates that syncretizing effect is better, as shown in table 1, four indexs of the method for the present invention
Numerical value is bigger than other four kinds of methods, illustrates to take the image syncretizing effect of the method for the present invention good.
Table 1：Flower schemes the objective evaluation of different fusion methods
Embodiment 2；The source images that the present embodiment is chosen are (c) and (d) in Fig. 5, to there is a group picture of weak vibrations,
Have certain representativeness, remaining step identical as implementing 1 in multi-focus image fusion.
To make differential chart as shown in Figure 9 as shown in figure 8, choosing source images (d) for fusion results, it can be seen that the method for the present invention
Blending image does not have apparent remaining information, and in other four figures, it has apparent regional choice misalignment on the head of people and leads
The residue of the fuse information of cause.Objective evaluation index is as shown in table 2, the numerical value of four indexs of the method for the present invention than other four
Kind of method it is big, illustrate to take the image syncretizing effect of the method for the present invention good.
Table 2：Lab schemes the objective evaluation of different fusion methods
Specific embodiments of the present invention are explained in detail above in conjunction with attached drawing, but the present invention is not limited to above-mentioned realities
Example is applied, it within the knowledge of a person skilled in the art, can also be without departing from the purpose of the present invention
Various changes can be made.
Claims (2)
1. a kind of method optimized to poly focal power image co-registration based on differential chart, it is characterised in that：It comprises the following specific steps that：
Step 1：The pretreatment of image；
To source images IAAnd IBGaussian filtering is carried out, the average value of source images is calculated：
IAIndicate source images A, IBIndicate source images B, IAVGIndicate the mean value of source images A and B；
Step 2：Decomposition multi-direction to Image Multiscale obtains original fusion image；
Use neighborhood distance filter to source images I firstAAnd IBMulti-resolution decomposition is carried out, the number of plies is four layers, and each layer includes height
Frequency information and low-frequency information；
Then the multi-direction decomposition methods of NSCT are used to carry out multi-direction decomposition to high-frequency information, it is 16 that wherein first layer, which decomposes direction,
A, the second layer is 8, and third layer is 4, last layer is 1, obtains IAHigh-frequency information IA,hWith low-frequency information IA,l, IB's
High-frequency information IB,hWith low-frequency information IB,l；
Spatial frequency values of the pixel based on neighborhood in high-frequency information are finally calculated, the decision square of high-frequency information fusion is constructed
Battle array, tentatively merges high-frequency information by decision matrix, and low-frequency information, which is taken, averages, the high-frequency information after preliminary fusion
Pass through the inverse transformation of multi-direction multi-scale transform with the low-frequency information after averaging, forms original fusion image IIN；
Step 3：Structure error image simultaneously calculates energy value；
Indicate source images IAWith mean value image IAVGIn the difference of pixel (i, j),Table
Show original fusion image IINWith mean value image IAVGIn the difference of pixel (i, j)；
Calculate energy value：
Indicate IIN(i, j) and IAVGThe difference of (i, j)Corresponding energy value,Indicate IAVG(i, j) and IAThe difference of (i, j)Corresponding energy value, M × N indicate pre-
If Size of Neighborhood, (i+m, j+n) indicates any point in neighborhood M × N centered on (i, j)；
Using two-sided filter to energy matrixWithIt carries out protecting side denoising,
Respectively obtain matrix char1 and char2；
Step 4：Construct initial bianry image；
(1) initial binary map MAP1, MAP2 are constructed by char1 and char2：
MAP1 (i, j) indicates that pixel (i, j) corresponding value in binary map MAP1, MAP2 (i, j) indicate pixel (i, j)
The corresponding value in binary map MAP2, char1 (i, j) indicate pixel (i, j) corresponding value, char2 in char1
(i, j) indicates that pixel (i, j) corresponding value in char2, δ are the threshold value of setting；
(2) bianry image is modified：
X × Y indicates that preset Size of Neighborhood, (i+a, j+b) indicate any point in neighborhood X × Y centered on (i, j)；
MAP1 and MAP2 is modified respectively by above formula, revised binary map MAP1' and MAP2' is obtained, for source figure
As in focus and the equal unobvious area of non-focusing (i, j) | MAP1'(i, j)=MAP2'(i, j) handled, take MAP1'(i, j)
=MAP2'(i, j)=0.5；
Step 5：By modified binary map MAP1', MAP2' is to source images IAAnd IBIt is merged.
2. the method according to claim 1 optimized to poly focal power image co-registration based on differential chart, it is characterised in that：Step
Fusion rule in rapid 5 is as follows：
(1) fusion based on spatial domain：
More modified binary map MAP1', MAP2', if meeting MAP1'(i, j)=0, MAP2'(i, j)=1, then select source figure
As IB(i, j) is filled, if meeting MAP1'(i, j) and=1, MAP2'(i, j)=0, then select source images IA(i, j) is filled out
Fill, other the case where then select original fusion image IIN(i, j) is filled, IF(i, j) is the fusion knot that last optimization generates
Fruit；
(2) fusion based on transform domain：
More modified binary map MAP1' and MAP2', if meeting MAP1'(i, j)=1, MAP2'(i, j)=0, then select source figure
As IAHigh-frequency information IA,h(i, j) and low-frequency information IA,l(i, j) is the high-frequency information and low-frequency information of blending image, if meeting
MAP1'(i, j)=0, MAP2'(i, j)=1, then select source images IBHigh-frequency information IB,h(i, j) and low-frequency information IB,l(i,
J) it is the high-frequency information and low-frequency information of blending image, if meeting MAP1'(i, j)=MAP2'(i, j)=0.5, then selection is initial
Blending image IINHigh-frequency information IIN, h(i, j) and low-frequency information IIN,l(i, j) is the high-frequency information and low frequency letter of blending image
Then breath carries out inverse transformation to the high and low frequency information of obtained blending image, obtains final fusion results IF。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610210196.6A CN105913407B (en) | 2016-04-06 | 2016-04-06 | A method of poly focal power image co-registration is optimized based on differential chart |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610210196.6A CN105913407B (en) | 2016-04-06 | 2016-04-06 | A method of poly focal power image co-registration is optimized based on differential chart |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105913407A CN105913407A (en) | 2016-08-31 |
| CN105913407B true CN105913407B (en) | 2018-09-28 |
Family
ID=56745700
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610210196.6A Active CN105913407B (en) | 2016-04-06 | 2016-04-06 | A method of poly focal power image co-registration is optimized based on differential chart |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105913407B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107403416B (en) * | 2017-07-26 | 2020-07-28 | 温州大学 | NSCT-based medical ultrasonic image denoising method with improved filtering and threshold function |
| CN108171676B (en) * | 2017-12-01 | 2019-10-11 | 西安电子科技大学 | Multi-focus image fusing method based on curvature filtering |
| CN108399645B (en) * | 2018-02-13 | 2022-01-25 | 中国传媒大学 | Image coding method and device based on contourlet transformation |
| CN109658371B (en) * | 2018-12-05 | 2020-12-15 | 北京林业大学 | Fusion method and system of infrared image and visible light image and related equipment |
| CN109754386A (en) * | 2019-01-15 | 2019-05-14 | 哈尔滨工程大学 | A kind of progressive Forward-looking Sonar image interfusion method |
| CN113947554B (en) * | 2020-07-17 | 2023-07-14 | 四川大学 | Multi-focus image fusion method based on NSST and significant information extraction |
| CN113379660B (en) * | 2021-06-15 | 2022-09-30 | 深圳市赛蓝科技有限公司 | Multi-dimensional rule multi-focus image fusion method and system |
| CN115205181B (en) * | 2022-09-15 | 2022-11-18 | 季华实验室 | Multi-focus image fusion method and device, electronic equipment and storage medium |
| CN116916166B (en) * | 2023-09-12 | 2023-11-17 | 湖南湘银河传感科技有限公司 | Telemetry terminal based on AI image analysis |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854265A (en) * | 2012-12-03 | 2014-06-11 | 西安元朔科技有限公司 | Novel multi-focus image fusion technology |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9092875B2 (en) * | 2011-04-12 | 2015-07-28 | Panasonic Intellectual Property Management Co., Ltd. | Motion estimation apparatus, depth estimation apparatus, and motion estimation method |
-
2016
- 2016-04-06 CN CN201610210196.6A patent/CN105913407B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854265A (en) * | 2012-12-03 | 2014-06-11 | 西安元朔科技有限公司 | Novel multi-focus image fusion technology |
Non-Patent Citations (3)
| Title |
|---|
| Adaptive multi-focus image fusion using a wavelet-based statistical sharpness measure;Jing Tian et al.;《Signal Processing》;20120930;第92卷(第9期);第2137-2146页 * |
| Multifocus Image Fusion Based on NSCT and Focused Area Detection;Yong Yang et al.;《IEEE Sensors Journal》;20150531;第15卷(第5期);第2824-2838页 * |
| 结合小波变换和自适应分块的多聚焦图像快速融合;刘羽 等;《中国图象图形学报》;20131130;第18卷(第11期);第1435-1444页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105913407A (en) | 2016-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105913407B (en) | A method of poly focal power image co-registration is optimized based on differential chart | |
| CN109242888B (en) | Infrared and visible light image fusion method combining image significance and non-subsampled contourlet transformation | |
| CN109685732A (en) | A kind of depth image high-precision restorative procedure captured based on boundary | |
| CN106339998A (en) | Multi-focus image fusion method based on contrast pyramid transformation | |
| CN105445277A (en) | Visual and intelligent detection method for surface quality of FPC (Flexible Printed Circuit) | |
| CN107369148A (en) | Based on the multi-focus image fusing method for improving SML and Steerable filter | |
| CN106898048B (en) | A kind of undistorted integration imaging 3 D displaying method being suitable for complex scene | |
| CN106846289A (en) | A kind of infrared light intensity and polarization image fusion method based on conspicuousness migration with details classification | |
| CN107909560A (en) | A kind of multi-focus image fusing method and system based on SiR | |
| CN108596975A (en) | A kind of Stereo Matching Algorithm for weak texture region | |
| Bai | Morphological infrared image enhancement based on multi-scale sequential toggle operator using opening and closing as primitives | |
| CN104966274B (en) | A kind of On Local Fuzzy restored method using image detection and extracted region | |
| CN106447640B (en) | Multi-focus image fusing method and device based on dictionary learning, rotation guiding filtering | |
| CN108564597A (en) | A kind of video foreground target extraction method of fusion gauss hybrid models and H-S optical flow methods | |
| CN110363719A (en) | A kind of cell layered image processing method and system | |
| CN109509163A (en) | A kind of multi-focus image fusing method and system based on FGF | |
| CN109118440B (en) | Single image defogging method based on transmissivity fusion and adaptive atmospheric light estimation | |
| CN106022337B (en) | A kind of planar target detection method based on continuous boundary feature | |
| CN109559273A (en) | A kind of quick joining method towards vehicle base map picture | |
| CN103942756B (en) | A kind of method of depth map post processing and filtering | |
| Chen et al. | A color-guided, region-adaptive and depth-selective unified framework for Kinect depth recovery | |
| CN109961016A (en) | The accurate dividing method of more gestures towards Intelligent household scene | |
| Yu et al. | Image and video dehazing using view-based cluster segmentation | |
| CN106530277B (en) | A kind of image interfusion method based on small echo directional correlation coefficient | |
| CN113887624A (en) | Improved feature stereo matching method based on binocular vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210224 Address after: 650000 room 1701, 17th floor, block a, science and Technology Information Innovation Incubation Center, Chenggong District, Kunming City, Yunnan Province Patentee after: YUNNAN UNITED VISUAL TECHNOLOGY Co.,Ltd. Address before: 650093 No. 253, Xuefu Road, Wuhua District, Yunnan, Kunming Patentee before: Kunming University of Science and Technology |
|
| TR01 | Transfer of patent right |