CN109102465A - A kind of calculation method of the content erotic image auto zoom of conspicuousness depth of field feature - Google Patents
A kind of calculation method of the content erotic image auto zoom of conspicuousness depth of field feature Download PDFInfo
- Publication number
- CN109102465A CN109102465A CN201810962052.5A CN201810962052A CN109102465A CN 109102465 A CN109102465 A CN 109102465A CN 201810962052 A CN201810962052 A CN 201810962052A CN 109102465 A CN109102465 A CN 109102465A
- Authority
- CN
- China
- Prior art keywords
- depth
- image
- calculates
- conspicuousness
- field
- 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.)
- Pending
Links
- 238000004364 calculation method Methods 0.000 title claims abstract description 19
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 63
- 230000008859 change Effects 0.000 claims abstract description 7
- 238000009795 derivation Methods 0.000 claims abstract description 4
- 230000002708 enhancing effect Effects 0.000 claims abstract description 4
- 230000000007 visual effect Effects 0.000 claims description 8
- 230000015556 catabolic process Effects 0.000 claims description 3
- 210000005079 cognition system Anatomy 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000000034 method Methods 0.000 description 31
- 238000011160 research Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000009966 trimming Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/403—Edge-driven scaling; Edge-based scaling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of calculation methods of the content erotic image auto zoom of conspicuousness depth of field feature, several patch are arbitrarily chosen from 1000 width high-definition images first including A1, A2 (using the Gaussian Blur of σ=2) by (being trained to obtain fuzzy dictionary Dblur, A3 subsequently for each of target image G patch be decomposed into several ground atoms cum rights superposition and, for A4 by calculating the number discovery of the ground atom that decomposites of target patch, the detailed information of clear image is relatively more, needs DblurIn more atoms approached, blurred picture can be analyzed to less atom superposition, and A5 is by the atom number of picture breakdownfblurIt is described as the fuzzy depth of field and proposes depth of field estimating algorithm; obtain A0 depth of field estimating algorithm; A6 calculates the Protection formula of background by carrying out derivation enhancing in the direction x and the direction y for depth of field F; A7 keeps reducing the change to non-significant regional structure information while marking area, can alleviate the distortion phenomenon in the non-significant region of bring.
Description
Technical field
The present invention relates to image scaling field, the content erotic image auto zoom of specially a kind of conspicuousness depth of field feature
Calculation method.
Background technique
It is universal with equipment such as mobile phone, digital cameras, people all the time not in photo record life a bit
One drop.However, different displays equipment (mobile phone, Pad, TV, webpage etc.) will lead to photo content to adapt to screen in length and breadth
Than generating serious distortion (stretch, compress or distort).This research is directed to this problem, proposes based on conspicuousness depth of field feature
Content erotic image auto zoom algorithm, so that image usually needs to be showed in the equipment of different sizes and resolution ratio
It is real.After the completion of shooting, resolution ratio and aspect ratio cannot change automatically universal electric image with the size of display equipment.?
When display, switching will lead to image by serious stretching or compression in different sizes, to influence the normal vision body of image
It tests and is lost with image information, saliency feature can guarantee that image important information, important area content are not lost；Based on side
The depth of field algorithm for estimating of edge fuzzy behaviour can be improved the integrality of image structure information.
Traditional images Zoom method includes linear-scale scaling (Scale), cuts out scaling (Crop), pixel scaling
(Pixel) and optimal scaling (Optimal)[1].Wherein Scale method using equal proportion scaling stretch principle, easily cause image because
It generates main target greatly for ratio variation and is compressed or widen；Crop method ensures image by cutting the peripheral part of image
The content of center portion, if the important information of cropped picture is distributed in picture side, this method can generate serious information and lose
Lose phenomenon；Pixel algorithm and Optimal algorithm, which are used, completes scaling based on single pixel importance, since pixel can not embody
The defects of image global feature, this method inevitably generate noise, sawtooth wave, in addition to this, it is soft that popular commercial repairs figure
Part Photoshop (PS) has increased content erotic trimming operation: Content-aware Scale newly also for this problem
(CAS), but image scaled is seriously damaged, since mobile phone, widescreen display etc. occur, so that meeting during image scaling
There is aspect ratio and changes larger situation.For large scale image scaling, due to the presence of significant characteristics often will cause it is aobvious
It writes and occurs distortion (important information overprotection phenomenon) around object.
Existing zoom technology will appear loss important information, object edge distortion, image structure information in image scaling
The problems such as imperfect, this research combine relevant psychological theory, improve traditional area conspicuousness calculation method and introduce non-aobvious
The concept that regional structure information is kept is write, image (slight) fuzzy behaviour is estimated using sparse learning algorithm to obtain fuzzy word
Allusion quotation proposes picture structure description and assists conspicuousness to calculate with this can have while keeping significant information, edge complete
Effect reduces the loss or damage of structural information, and the image after making scaling more meets the visual experience of people.Using originally researching and proposing
Algorithm establish content erotic image auto zoom prototype system, can quickly, easily by target image according to real world devices
Size be automatically performed scaling, according to the significant characteristics of domain type conspicuousness calculation method although available image, still
Due to lacking the constraint for image boundary, the image after leading to scaling can compare at edge (especially important area)
Serious deformation.This research proposes that the conspicuousness of edge sensitive calculates for the problem that this phenomenon, solves edge deformation.
Summary of the invention
The purpose of the present invention is to provide a kind of calculating sides of the content erotic image auto zoom of conspicuousness depth of field feature
Method can play multi-function unit using being equipped in cabinet, to solve the problems mentioned in the above background technology.
To achieve the above object, the invention provides the following technical scheme: a kind of content erotic figure of conspicuousness depth of field feature
As the calculation method of auto zoom, including the calculating of the S0 Boundary-Type conspicuousness depth of field, A0 depth of field estimating algorithm and the non-significant region A8
Deformation calculation.
In the above-mentioned methods: the S0 Boundary-Type conspicuousness depth of field calculates, which is characterized in that step and calculating are as follows:
S1: ridge line can be generated in image object edge with watershed algorithm first and be pre-processed；
S2: image is divided into numerous zonule G；
It calculates: G (x, y)=max (grad { f (x, y) }, g θ),
S3: the visual cognition system of the Psychological Angle mankind is by recognizing after successive ignition；
S4:G is merged by similarity calculation with adjacent region according to different scale and obtained: color discrimination is got over
Greatly, conspicuousness is higher,
It calculates:
Obtain Gi(i=3,5,9) layer, color discrimination is bigger, and conspicuousness is higher.
S5: position discrimination conspicuousness, which is obtained, wants high close to middle section pixel ratio neighboring pixel conspicuousness,
It calculates:
S0: the Boundary-Type conspicuousness depth of field calculates；
It calculates: Si=norm (Ci)·norm(Hi),.
In the above-mentioned methods: A0 depth of field estimating algorithm, which is characterized in that step and calculating are as follows:
A1: arbitrarily choosing several patch from 1000 width high-definition images first herein,
A2: fuzzy dictionary D is obtained by being trained (using the Gaussian Blur of σ=2)blur,
It calculates: calculating:||xi||0≤ k,
A3: subsequently for each of target image G patch be decomposed into several ground atoms cum rights superposition and.
It calculates:||Gi-Dblurxi||2≤ k,
A4: the number by calculating the ground atom that target patch is decomposited finds that the detailed information of clear image compares
It is more, need DblurIn more atoms approached, blurred picture can be analyzed to less atom superposition；
A5: the atom number fblur of picture breakdown is described as the fuzzy depth of field and proposes depth of field estimating algorithm；
It calculates: fblur=| | xi||0,
A0: depth of field estimating algorithm；
It calculates:
Target image I is inputted, dictionary Dblur. is obscured
Export .fblur.
Step1. I is decomposed by what target image I had an overlapping0, I1, I2..., In.
Step2.for I=0 to n.
Step3. fuzzy decomposition is carried out by formula 5, obtains x0, x1..., xn.
Step4. f is found out by formula 60, f1..., fn.
Step5.end。
In the above-mentioned methods: the deformation calculation in the non-significant region A8, which is characterized in that step and calculating are as follows:
A0: depth of field estimating algorithm；
A6: the Protection formula of background is calculated by carrying out derivation enhancing in the direction x and the direction y for depth of field F；
It calculates:
A7: it keeps reducing the change to non-significant regional structure information while marking area, non-significant area can be alleviated
The distortion phenomenon in domain；
A8: non-significant regional deformation calculates；
It calculates:
Input: target image G obscures dictionary Dblur, image scaling ratio r.
Output: the image G ' after scaling
Step1.W (G) → Ge, n=| Ge||0.
Step2. by Dblur, image depth numerical value F. is obtained using fuzzy depth of field algorithm
Step3. pass through formula (8), F → F '
Step4.for i=1to3.
Step4.1.for j=1ton.
Step4.1.1. color discrimination Cij. is calculated by formula (2)
Step4.1.2. pass through formula (3) calculating position discrimination Hij.
Step4.1.3. conspicuousness is carried out by formula (4) calculate Sij.
Step4.1.4.
Step4.2.end.
Step4.3. by S 'ijGenerate Si.
Step5.end.
Step6. conspicuousness fusion is carried out to { Si }, generates S (S, r) → Standard Seam Carving
Algorithm completes image scaling G '.
Compared with prior art, the beneficial effects of the present invention are:
1) the content erotic image auto zoom algorithm of the conspicuousness depth of field feature solves between image and original image due to ruler
Very little change and caused by perception difference, the image after scaling will not generate deformation at edge (especially important area),
Image after making scaling can accurately express the information that original image can be transmitted, and important information can be made to keep and not lose, to image
Depth information calculated, during scaling be added depth information constraint so that scaling after image in structure with
There is still similitudes for original image, improve the energy line penetration probability for marking area by the improvement to algorithm, can be effective
Avoid non-significant region distort the phenomenon that；
2) the significant characteristics calculation method of edge sensitive is proposed.Traditional conspicuousness calculating belongs to regional calculating, passes through
Feature extraction and calculation goes out the pixel region for attracting people to pay attention to.Method proposed in this paper joined side while conspicuousness calculates
Edge constraint.This method has many potential applications in computer vision, such as: image segmentation, object identification etc..
3) Image Zooming Algorithm of the depth estimation based on sparse study is proposed, for describing the structural information of image.Root
According to sparse learning characteristic, by indicate sliding window learning operator number as measurement standard, for indicating depth information.It introduces
Image scaling after depth information meets the mankind to the perception characteristics of image, reduces a possibility that marginal information distorts.
4) it is firstly introduced the holding of the structural information in non-significant region.Conventional algorithm is mainly in marking area information
It keeps, this method is proposed for the energy line global optimum for remaining able to ensure to search of non-significant area information.
Detailed description of the invention
Attached drawing 1 is the step flow chart of the method for the present invention；
Attached drawing 2 is the method flow diagram that the conspicuousness depth of field in edge of the present invention calculates；
Attached drawing 3 is first group of questionnaire survey data statistics chart percentage schematic diagram；
Attached drawing 4 is first group of questionnaire survey data statistics chart coordinate percentage schematic diagram；
Attached drawing 5 is second group of questionnaire survey data statistics chart percentage schematic diagram；
Attached drawing 6 is second group of questionnaire survey data statistics chart percentage schematic diagram；
Attached drawing 7 is first group and second group of questionnaire comprehensive survey tables of data；
Attached drawing 8 is first group and second group of questionnaire comprehensive survey partial data table.
Specific embodiment
Technical solution in the embodiment of the present invention carries out close Chu, is fully described by, it is clear that described embodiment is only
It is a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Please refer to Fig. 1-8, the present invention provides a kind of technical solution: a kind of content erotic image of conspicuousness depth of field feature from
The calculation method of dynamic scaling calculates, the shape of A0 depth of field estimating algorithm and the non-significant region A8 including the S0 Boundary-Type conspicuousness depth of field
Become and calculate, the S0 Boundary-Type conspicuousness depth of field calculates, and step and calculating are as follows:
S1: ridge line can be generated in image object edge with watershed algorithm first and be pre-processed；
S2: image is divided into numerous zonule G；
It calculates: G (x, y)=max (grad { f (x, y) }, g θ),
S3: the visual cognition system of the Psychological Angle mankind is by recognizing after successive ignition；
S4:G is merged by similarity calculation with adjacent region according to different scale and obtained: color discrimination is got over
Greatly, conspicuousness is higher,
It calculates:
Obtain Gi(i=3,5,9) layer, color discrimination is bigger, and conspicuousness is higher.
S5: position discrimination conspicuousness, which is obtained, wants high close to middle section pixel ratio neighboring pixel conspicuousness,
It calculates:
S0: the Boundary-Type conspicuousness depth of field calculates；
It calculates: Si=norm (Ci)·norm(Hi),
A0 depth of field estimating algorithm, step and calculating are as follows:
A1: arbitrarily choosing several patch from 1000 width high-definition images first herein,
A2: fuzzy dictionary D is obtained by being trained (using the Gaussian Blur of σ=2)blur,
It calculates: calculating:||xi||0≤ k,
A3: subsequently for each of target image G patch be decomposed into several ground atoms cum rights superposition and.
It calculates:||Gi-Dblurxi||2≤ k,
A4: the number by calculating the ground atom that target patch is decomposited finds that the detailed information of clear image compares
It is more, need DblurIn more atoms approached, blurred picture can be analyzed to less atom superposition；
A5: the atom number fblur of picture breakdown is described as the fuzzy depth of field and proposes depth of field estimating algorithm；
It calculates: fblur=| | xi||0,
A0: depth of field estimating algorithm；
It calculates:
Target image I is inputted, dictionary Dblur. is obscured
Export .fblur.
Step1. I is decomposed by what target image I had an overlapping0, I1, I2..., In.
Step2.for I=0 to n.
Step3. fuzzy decomposition is carried out by formula 5, obtains x0, x1..., xn.
Step4. f is found out by formula 60, f1..., fn.
Step5.end.
The deformation calculation in the non-significant region A8, step and calculating are as follows:
A0: depth of field estimating algorithm；
A6: the Protection formula of background is calculated by carrying out derivation enhancing in the direction x and the direction y for depth of field F；
It calculates:
A7: it keeps reducing the change to non-significant regional structure information while marking area, non-significant area can be alleviated
The distortion phenomenon in domain；
A8: non-significant regional deformation calculates；
It calculates:
Input: target image G obscures dictionary Dblur, image scaling ratio r.
Output: the image G ' after scaling
Step1.W (G) → Ge, n=| Ge||0.
Step2. by Dblur, image depth numerical value F. is obtained using fuzzy depth of field algorithm
Step3. pass through formula (8), F → F '
Step4.for i=1 to 3.
Step4.1.for j=1ton.
Step4.1.1. color discrimination Cij. is calculated by formula (2)
Step4.1.2. pass through formula (3) calculating position discrimination Hij.
Step4.1.3. conspicuousness is carried out by formula (4) calculate Sij.
Step4.1.4.
Step4.2.end.
Step4.3. by S 'ijGenerate Si.
Step5.end.
Step6. conspicuousness fusion is carried out to { Si }, generates S (S, r) → Standard Seam Carving
Algorithm completes image scaling G '.
Experiment and analysis
It include first SC with classical way in terms of algorithm performance[1]、SNS[18]、WSM[7], SDS (SA+DE+SC)[11]Method
Compare, verifies this paper algorithm and important information, shape border are saved completely.Secondly, herein by the average energy value is compared in number
It is compared according to aspect and above method.Then the research method for using questionnaire, choose different crowd and various ways into
It has gone user's questionnaire survey, has verified the real reliability of context of methods,
It is compared with classic algorithm
In order to embody the versatility of experiment, found at random from Network Picture Database (Google and Baidu) herein test image into
Row comparative experiments.Comparison algorithm is mainly chosen classic algorithm SC and algorithm SNS, WSM, SDS for newly releasing in recent years and is calculated herein
Method is tested.Wherein, SNS algorithm completes figure using the deformation method based on Mesh by gradient information and conspicuousness information
The scaling of picture.WSM algorithm combines the energy balane that marginal information completes image on the basis of conspicuousness information.SDS algorithm
Depth information combination energy balane by obtaining image carries out the scaling of image,
Experiment display show that circle mark is the region compared with original image, SC and SNS algorithm is all different degrees of
The size of personage in figure is reduced, although WSM algorithm ensure that personage is constant, but the fence on the left of photo is had occurred bright
Aobvious deformation, SDS algorithm distort the shape of the doggie on the right side of image, context of methods place all with original image
Keep almost the same, and there is no too many differences for visual effect, after SC, SNS, WSM and SDS algorithm reduce 30% to original image,
Different degrees of deformation all has occurred in the shape in the image lower right corner, and the content erotic image auto zoom of conspicuousness depth of field feature is calculated
Method is keeping the complete aspect effect in boundary more preferable, and many algorithms can just obtain more satisfied effect when small scale scales,
But it will appear different degrees of distortion when large percentage (generally more than 50%).This paper algorithm is while keeping important information
For image structure information, the structural information in especially non-significant region is also kept, therefore is had in large scale scaling
There is preferable effect,
Energy function analysis
It removes except visual comparison, calculates that several main method the average energy value are higher to show algorithm effect more herein
It is good.SC, SNS, WSM, SDS, Ours+f are compared herein ' (SA) and Ours+f ' (Dblur), data are shown in Table 1.
1 experimental result of table
As it can be seen from table 1 being essentially all each in terms of the average energy value of algorithm proposed in this paper after scaling
First 2 of test sample.Although the numerical value that standard SC algorithm obtains sometimes is higher, SC will lead to significant area in visual effect
The deformation in domain.Therefore, method proposed in this paper all has good performance for image scaling in terms of visual effect and data.
User's evaluation analysis
It is commented in a manner of questionnaire in randomly selecting user of 30 ages between 15-27 on network herein
Valence.Randomly select 5 groups of pictures, be divided into two groups: group 1 by this paper algorithm with cut out, compress, the content in Photoshop software is known
The result not scaled is compared, and group 2 is then compared with the method for current paper.Compare content be to same image into
Row same ratio reduces, and asks user by observation selection, which, which is opened, is best suitable for the in-mind anticipation of user.The purpose of this experiment is to adopt
With the mode of human-computer interaction and the research method of questionnaire, by artificially evaluating method proposed in this paper and conventional method and
Fresh approach is compared.
User's evaluation investigation
We are by the way of network-based questionnaire survey.The reason of taking which is because by present network hand
Section can rapidly investigate questionnaire.Compared with the research method of traditional questionnaire survey, network-based investigation method has
Feedback speed is fast, and data are accurate, is easy to the features such as statisticalling analyze.
First group of questionnaire:
Questionnaire feedback is as follows, and user's evaluation shows that data are shown in that annex, this paper algorithm have biggish excellent compared to traditional algorithm
Gesture, this paper algorithm has 63.56% probability to be selected in four kinds of algorithms, sees Figure of description 3, Figure of description 4, explanation
Book attached drawing 7 and Figure of description 8.
Second group of questionnaire:
User's evaluation is as follows, and it is excellent that data show that this paper algorithm still has compared to newest Seam Carving innovatory algorithm
Gesture illustrates that this paper algorithm is selected number shown in attached drawing 5,8 data of Figure of description 6, Figure of description 7 and Figure of description
Account for the 55.33% of total amount.Result above suffices to show that the advantage that this paper algorithm is possessed in image domains.
It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments, Er Qie
In the case where without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter
From the point of view of which point, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the present invention is by appended power
Benefit requires rather than above description limits, it is intended that all by what is fallen within the meaning and scope of the equivalent elements of the claims
Variation is included within the present invention, and any reference signs in the claims should not be construed as limiting the involved claims.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Anyone skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should all cover within the scope of guarantor of the invention.
Claims (4)
1. a kind of calculation method of the content erotic image auto zoom of conspicuousness depth of field feature, including S0 Boundary-Type conspicuousness scape
It calculates deeply, the deformation calculation of A0 depth of field estimating algorithm and the non-significant region A8.
2. the S0 Boundary-Type conspicuousness depth of field according to claim 1 calculates, which is characterized in that step and calculating are as follows:
S1: ridge line can be generated in image object edge with watershed algorithm first and be pre-processed；
S2: image is divided into numerous zonule G；
It calculates: G (x, y)=max (grad { f (x, y) }, g θ),
S3: the visual cognition system of the Psychological Angle mankind is by recognizing after successive ignition；
S4:G is merged by similarity calculation with adjacent region according to different scale and obtained: color discrimination is bigger, shows
Work property is higher,
It calculates:
Obtain Gi(i=3,5,9) layer, color discrimination is bigger, and conspicuousness is higher.
S5: position discrimination conspicuousness, which is obtained, wants high close to middle section pixel ratio neighboring pixel conspicuousness,
It calculates:
S0: the Boundary-Type conspicuousness depth of field calculates；
It calculates: Si=norm (Ci)·norm(Hi),
3. A0 depth of field estimating algorithm according to claim 1, which is characterized in that step and calculating are as follows:
A1: arbitrarily choosing several patch from 1000 width high-definition images first herein,
A2: fuzzy dictionary D is obtained by being trained (using the Gaussian Blur of σ=2)blur,
It calculates: calculating:||xi||0≤ k,
A3: subsequently for each of target image G patch be decomposed into several ground atoms cum rights superposition and.
It calculates: ||Gi-Dblurxi||2≤ k,
A4: the number by calculating the ground atom that target patch is decomposited finds that the detailed information of clear image is relatively more,
Need DblurIn more atoms approached, blurred picture can be analyzed to less atom superposition；
A5: the atom number fblur of picture breakdown is described as the fuzzy depth of field and proposes depth of field estimating algorithm；
It calculates: fblur=| | xi||0,
A0: depth of field estimating algorithm；
It calculates:
Target image I is inputted, dictionary Dblur. is obscured
Export .fblur.
Stepl. I is decomposed by what target image I had an overlapping0, I1, I2..., In.
Step2.for I=0 to n.
Step3. fuzzy decomposition is carried out by formula 5, obtains x0, x1..., xn.
Step4. f is found out by formula 60, f1..., fn.
Step5.end.
4. according to claim 1 with the deformation calculation in the non-significant region A8 described in 3, which is characterized in that step and calculate it is as follows:
A0: depth of field estimating algorithm；
A6: the Protection formula of background is calculated by carrying out derivation enhancing in the direction x and the direction y for depth of field F；
It calculates:
A7: it keeps reducing the change to non-significant regional structure information while marking area, non-significant region can be alleviated
Distortion phenomenon；
A8: non-significant regional deformation calculates；
It calculates:
Input: target image G obscures dictionary Dblur, image scaling ratio r.
Output: the image G ' after scaling
Step1.W (G) → Ge, n=| Ge||0.
Step2. by Dblur, image depth numerical value F. is obtained using fuzzy depth of field algorithm
Step3. pass through formula (8), F → F '
Step4.for i=1 to 3.
Step4.1.for j=1 to n.
Step4.1.1. color discrimination Cij. is calculated by formula (2)
Step4.1.2. pass through formula (3) calculating position discrimination Hij.
Step4.1.3. conspicuousness is carried out by formula (4) calculate Sij.
Step4.1.4.
Step4.2.end.
Step4.3. by S 'ijGenerate Si.
Step5.end.
Step6. conspicuousness fusion is carried out to { Si }, it is complete generates S (S, r) → Standard Seam Carving algorithm
At image scaling G '.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810962052.5A CN109102465A (en) | 2018-08-22 | 2018-08-22 | A kind of calculation method of the content erotic image auto zoom of conspicuousness depth of field feature |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810962052.5A CN109102465A (en) | 2018-08-22 | 2018-08-22 | A kind of calculation method of the content erotic image auto zoom of conspicuousness depth of field feature |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109102465A true CN109102465A (en) | 2018-12-28 |
Family
ID=64850722
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810962052.5A Pending CN109102465A (en) | 2018-08-22 | 2018-08-22 | A kind of calculation method of the content erotic image auto zoom of conspicuousness depth of field feature |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109102465A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111242880A (en) * | 2019-12-30 | 2020-06-05 | 广州市明美光电技术有限公司 | Multi-depth-of-field image superposition method, equipment and medium for microscope |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104463873A (en) * | 2014-12-10 | 2015-03-25 | 西安电子科技大学 | Image target repositioning method based on local uniform scaling |
| US20160104054A1 (en) * | 2014-10-08 | 2016-04-14 | Adobe Systems Incorporated | Saliency Map Computation |
| CN107992874A (en) * | 2017-12-20 | 2018-05-04 | 武汉大学 | Image well-marked target method for extracting region and system based on iteration rarefaction representation |
-
2018
- 2018-08-22 CN CN201810962052.5A patent/CN109102465A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160104054A1 (en) * | 2014-10-08 | 2016-04-14 | Adobe Systems Incorporated | Saliency Map Computation |
| CN104463873A (en) * | 2014-12-10 | 2015-03-25 | 西安电子科技大学 | Image target repositioning method based on local uniform scaling |
| CN107992874A (en) * | 2017-12-20 | 2018-05-04 | 武汉大学 | Image well-marked target method for extracting region and system based on iteration rarefaction representation |
Non-Patent Citations (1)
| Title |
|---|
| 岳朋飞 等: "结合边缘模糊的景深显著性在图像缩放中的研究", 《计算机辅助设计与图形学学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111242880A (en) * | 2019-12-30 | 2020-06-05 | 广州市明美光电技术有限公司 | Multi-depth-of-field image superposition method, equipment and medium for microscope |
| CN111242880B (en) * | 2019-12-30 | 2023-05-02 | 广州市明美光电技术有限公司 | Multi-depth-of-field image superposition method, equipment and medium for microscope |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| DE112016006081T5 (en) | Automatic real-time vehicle camera calibration | |
| CN109146892A (en) | A kind of image cropping method and device based on aesthetics | |
| JP6075110B2 (en) | Image processing apparatus, image processing method, and image processing program | |
| CN109145803B (en) | Gesture recognition method and device, electronic equipment and computer readable storage medium | |
| CN110175504A (en) | A kind of target detection and alignment schemes based on multitask concatenated convolutional network | |
| CN111160291B (en) | Human eye detection method based on depth information and CNN | |
| CN112381183B (en) | Target detection method and device, electronic equipment and storage medium | |
| CN105488758A (en) | Image scaling method based on content awareness | |
| CN111259878A (en) | Method and equipment for detecting text | |
| JP2019117577A (en) | Program, learning processing method, learning model, data structure, learning device and object recognition device | |
| CN103347196A (en) | Method for evaluating stereo image vision comfort level based on machine learning | |
| US20150138193A1 (en) | Method and device for panorama-based inter-viewpoint walkthrough, and machine readable medium | |
| CN111209811B (en) | Method and system for detecting eyeball attention position in real time | |
| DE102018003475A1 (en) | Form-based graphic search | |
| US11380035B2 (en) | Method and apparatus for generating map | |
| US20130201358A1 (en) | Efficient Line Detection Method | |
| US20220277580A1 (en) | Hand posture estimation method and apparatus, and computer storage medium | |
| JP7264929B2 (en) | Backgroundless image generation method and apparatus, electronic device, storage medium, and computer program | |
| CN107545544A (en) | The method and apparatus for removing the turbid object in image | |
| CN103914829B (en) | Method for detecting edge of noisy image | |
| CN107464258A (en) | A kind of method that three-dimensional planar is extracted from image | |
| CN114332794B (en) | Target detection method, system, device and medium for train line image | |
| CN109102465A (en) | A kind of calculation method of the content erotic image auto zoom of conspicuousness depth of field feature | |
| CN105376553A (en) | Three-dimensional video repositioning method | |
| CN103646397A (en) | Real-time synthetic aperture perspective imaging method based on multi-source data fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181228 |
|
| WD01 | Invention patent application deemed withdrawn after publication |