CN101655913A - Computer generated image passive detection method based on fractal dimension - Google Patents
Computer generated image passive detection method based on fractal dimension Download PDFInfo
- Publication number
- CN101655913A CN101655913A CN200910195780A CN200910195780A CN101655913A CN 101655913 A CN101655913 A CN 101655913A CN 200910195780 A CN200910195780 A CN 200910195780A CN 200910195780 A CN200910195780 A CN 200910195780A CN 101655913 A CN101655913 A CN 101655913A
- Authority
- CN
- China
- Prior art keywords
- image
- sub
- piece
- fractal dimension
- grid
- 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
Images
Abstract
The invention relates to a computer generated image passive detection method based on fractal dimension in the technical field of digital images, which comprises the following steps: firstly, respectively performing sub-block screening to images in a natural image and computer-generated image library in a training stage; secondly, respectively computing fractal dimension of each image sub-block, and further obtaining a set of eigenvectors; thirdly, training the series of eigenvectors by a support vector machine to obtain parameters of an optimized classifier; performing the similar processingto images to be detected to obtain a fractal eigenvector, and performing classification authentication by using the optimized classifier obtained in the training state to obtain a detection result. The invention increases an expression window on the basis of the traditional difference box fractal dimension method and a data multiplexing and variable box height mechanism based on dynamic planning idea, which enable the method to have greater improvement on computation complexity and detection rate.
Description
Technical field
What the present invention relates to is the detection method in a kind of Digital image technology field, specifically is a kind of computer generated image passive detection method of fractal dimension.
Background technology
In recent years, the develop rapidly of computer picture generation technique, the image that computing machine is generated more and more approaches real scene, to such an extent as to human eye almost can't make a distinction itself and natural image, the authenticity and the uniqueness of image are subjected to huge challenge.Computer generated image is the general data source that the false proof passive evaluation work of digital image information is faced, and the discriminating of natural image and computer generated image is distinguished have vital status in various fields such as crime survey, news report, intelligence analyses.The research of existing passive evaluation at computer generated image is less, and achievement is also few.With regard to the research work of present disclosed report, main direction of studying is based on that Integral Dimension space and the fractional dimension space launch.
In the Integral Dimension space, S.Lyu and H.Farid article " How realistic isphotorealistic? [J] " (" how true to nature photo-realistic images has? ") (IEEE Trans.SignalProcessing, 53 (2): 845-850,2005) (IEEE signal Processing periodical) proposes at first testing image to be carried out wavelet transformation, on wavelet field, extract average, variance, skewness, four statistics of kurtosis then, with these four statistics as statistical nature, then to these proper vectors with the machine learning method judgement of classifying.This method is only statistically analyzed image, do not point out the essential difference of computer generated image and natural image, so resolution is unsatisfactory.
In the fractional dimension space, Tian-Tsong Ng is at article " Physics-motivated features fordistinguishing photographic images and computer graphics[C] " (" distinguishing natural image and computer generated image based on physical characteristics ") (ACM Multimedia, Singapore, 2005) (ACM multimedia conferencing) utilizes to calculate and generates existing three intrinsic othernesses---object model difference between image and the natural image itself, light transmission difference and image acquisition procedures difference are extracted this otherness feature on the fractal dimension basis.But because the related mathematical model of this method in theory also is not very ripe, it is will be further in addition perfect to still need, and therefore, the final detection identification result of this method still is not very desirable yet.
Summary of the invention
The object of the invention is at the deficiencies in the prior art, and a kind of computer generated image passive detection method of fractal dimension is provided.The present invention has improved the computer generated image detection method based on difference box fractal dimension.Extract image difference box fractal dimension feature, by the method detection computations machine generation image of machine learning, it all reaches result preferably on detection accuracy and operating efficiency.
The present invention is achieved by the following technical solutions:
The present invention at first carries out sub-piece screening in the training stage respectively to the every width of cloth image in natural image and the computer generated image storehouse, then each image subblock is calculated its fractal dimension respectively, and then obtain an eigenvectors, at last should the series feature train and obtain the optimum classifier parameter with SVM (Support Vector Machines, support vector machine); At test phase, at first testing image is carried out as above similar processing and obtain the fractal characteristic vector, the optimum classifier that utilizes the training stage to obtain is then classified and is differentiated and then obtain testing result.
The described training stage, comprise the steps:
1. for reducing complexity, at first with every width of cloth image interception 512 * 512 sub-pieces in natural image and the computer generated image storehouse as pending image;
2. the every width of cloth image division that obtains in inciting somebody to action 1. respectively is several image subblocks of 64 * 64, to each image subblock, calculates its variances sigma
i, more relatively with the variance ω of itself and entire image.If sub-piece variances sigma
iLess than ω, it is slower for the variation of entire image to illustrate that then this sub-piece changes, and picture material is milder, thus with its deletion, otherwise keep this sub-piece, obtain L sub-piece through this step,
When L＞32, then from L sub-piece, give up (L-32) height piece of variance less than ω;
When L＜32, then from L sub-piece, choose variance 2., until obtaining 32 sub-pieces of variance greater than ω less than (32-L) height piece of ω and repeating step.
3. to above-mentioned 32 sub-pieces that obtain, adopt difference box fractal dimension method to calculate its fractal dimension respectively, obtain one group of 32 dimensional feature vector.
The calculation procedure of described difference box fractal dimension is:
1) the sub-piece of M * M size is divided for the grid pointwise with s*s, chosen s=2 herein, 3, obtain the pixel grey scale maximal value I in each grid respectively
MaxWith minimum value I
Min
2) calculate corresponding to pixel extreme value in the grid of 2 and 3 multiple size based on dynamic programming method.2,4,8 to be example, the maximin of the grid interior pixel gray scale that previous step is calculated each 2 * 2 grid in rapid is preserved as separating of ground floor minor structure, and it is right to obtain 32 * 32 groups of maximin altogether at this moment; Ask the maximal value and the minimum value of 4 * 4 grid interior pixel gray scales, as can be known in the ground floor minor structure therewith the sub-centering of maximin of corresponding 4 group of 2 * 2 grid of 4 * 4 grid position search and get final product, the maximal value of trying to achieve, minimum value are preserved as separating of second layer minor structure; And the like, ask 2
i* 2
iThe maximin of the grid interior pixel gray scale during division.This step finally obtains with s=2, and 3,4,6,8,12 is the maximal value and the minimum value of the pairing pixel grey scale of grid of yardstick.
3) judge whether selected s can be divided exactly by M, as can not, then carry out the expansion of image subblock window by duplicating to enlarge, obtain the image block of (s*M) * (s*M).Make r=s/M, image is imagined as curve in the three dimensions, then the xy plane is divided into the grid of many s * s, on each grid, is the box of a row s*s*s ', and wherein s ' is variable box height, is specially s '=maxI
k-minI
k, I
k(k=1,2, L, n) size of n grey scale pixel value in expression s * s unit area.If (i, j) the pairing box number of grid is
n
r(i，j)＝(maxI
k-minI
k)/r (1)
Covering the required box number of entire image piece is
And the box number of M under the yardstick s * sub-piece of M original image is N
r=N
r'/s
2, cause the variation of r by the variation of s, use least square fitting lgN
r-lg (1/r), the slope of obtaining are box dimension D, i.e. image fractal dimension.
4. repeating step 1. to step 3., obtain 32 dimension fractal dimension proper vectors of every width of cloth image in natural image and the computer generated image storehouse, respectively computer generated image proper vector and natural image proper vector are sent into SVM and train, the sorter model that obtains after the training is used for follow-up decision.
Described test phase comprises the steps:
1. input image to be declared intercepts 512 * 512 image blocks and is used for subsequent treatment, and calculates its variance ω;
2. the every width of cloth image division that obtains in inciting somebody to action 1. respectively is several image subblocks of 64 * 64, to each image subblock, calculates its variances sigma
i, more relatively with the variance ω of itself and entire image.If sub-piece variances sigma
iLess than ω, it is slower for the variation of entire image to illustrate that then this sub-piece changes, and picture material is milder, thus with its deletion, otherwise keep this sub-piece, obtain L sub-piece through this step,
When L＞32, then from L sub-piece, give up (L-32) height piece of variance less than ω;
When L＜32, then from L sub-piece, choose variance 2. less than (32-L) height piece of ω and repeating step, finally obtain 32 the sub-pieces of variance greater than ω.
3. 32 sub-pieces to 2. being obtained by step calculate its fractal dimension by the calculation procedure of described fractal dimension respectively, finally obtain one group of 32 dimensional feature vector;
4. the sorter model that will obtain after will being trained by the 32 dimensional feature vectors input that 3. step obtains detects processing, obtains final identification result.
The present invention adopts " box dimension " thought to extract image subblock fractal dimension feature, than traditional fractal dimension feature extracting method very big improvement has been arranged on complexity; The present invention has increased image subblock screening operation steps, greatly reduce the treatment capacity of subsequent operation step on the one hand, difference on self-similarity embodies better with natural image and computer generated image on the other hand, makes verification and measurement ratio be greatly improved.
Description of drawings
Fig. 1 is a feature extraction process flow diagram of the present invention.
Fig. 2 is image subblock screening process figure of the present invention.
Fig. 3 is improved difference box fractal dimension computing method process flow diagram among the present invention.
Fig. 4 is test demonstration example schematic;
Wherein: a, 512*512 picture, picture behind b, the piecemeal, c, screening back picture.
Embodiment
Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment has provided detailed embodiment and process being to implement under the prerequisite with the technical solution of the present invention, but protection scope of the present invention is not limited to following embodiment.
As Fig. 1, Fig. 2 and shown in Figure 3, present embodiment may further comprise the steps:
1. training stage
1. the training plan valut is set up---choose 400 width of cloth natural images, and 400 width of cloth computer generated images, picture format is jpeg, and picture size does not wait from 400*600 to 1024*768.A picture part is from the picture library (http://www.ee.columbia.edu/ln/dvmm/downloads/PIM_PRCG_dataset/) of Columbia University, and another part is from favorites.Image content has been contained various aspects such as personage, natural land, building, still life.Wherein, the software that computer generated image used has 3Dmax, Maya, Renderman, Shake, ZBrush or the like;
2. image subblock screens module---and arbitrary width of cloth figure in the picture library is intercepted 512 * 512 sizes as pending image (as Fig. 4 .a), calculate its population variance ω; To its piecemeal that carries out 64 * 64, obtain Fig. 4 .b then, each image subblock calculates its variances sigma respectively
i, relatively with the variance ω of itself and entire image.If sub-piece variances sigma
iLess than ω, it is slower for the variation of entire image to illustrate that then this sub-piece changes, and picture material is milder, thus with its deletion, otherwise keep this sub-piece, obtain L sub-piece through this step.If little (L-32) height piece of variance is then given up in L＞32 from L sub-piece; If L＜32 are then chosen little (32-L) height piece of variance and carried out repetition from L sub-piece, finally obtain 32 bigger sub-pieces of variance, shown in Fig. 4 .c;
3. the sub-piece fractal dimension of computed image---employing difference box fractal dimension method is calculated the fractal dimension of above-mentioned 32 sub-pieces respectively, obtains one group of 32 dimensional feature vector.Concrete measure is that 64 * 64 big boy's pieces of M=64 are divided for the grid pointwise with s*s, chooses s=2 herein, 3, obtain one group of each grid pixel maximal value I respectively
MaxAnd I
MinCalculate corresponding to pixel extreme value in the grid of 2 and 3 multiple size based on dynamic programming method, get s=2,3,4,6,8,12; Judge whether selected s can be divided exactly by M, as can not, then carry out the expansion of image subblock window by duplicating to enlarge, obtain the image block of (s*M) * (s*M); Make r=s/M, image is imagined as curve in the three dimensions, then the xy plane is divided into the grid of many s * s, on each grid, is the box of a row s*s*s ', and wherein s ' is variable box height, is specially s=maxI
k-minI
k, I
k(k=1,2, L, n) size of n grey scale pixel value in expression s * s unit area.If (i, j) the box number of grid indication is n
r(i, j)=(maxI
k-minI
k)/r, the required box number of covering entire image piece is
And the box number of M under the yardstick s * sub-piece of M original image is N
r=N
r'/s
2, cause the variation of r by the variation of s, with least square fitting 1g N
r-1g (1/r), the slope of obtaining are box dimension D, i.e. image fractal dimension.In this example, specifically each sub-piece fractal dimension is as shown in table 1, respectively corresponding above-mentioned each sub-piece position, each position in its table;
Table 1 fractal dimension
4. SVM support vector machine---a kind of machine learning method is quoted the Libsvm software package herein and is handled (http://www.csie.ntu.edu.tw/ ~ cjlin/libsvm).With in the image library pairing 32 dimensional feature vectors of totally 800 width of cloth images be organized into the required structure of Libsvm (idiographic flow seen http://www.csie.ntu.edu.tw/ ~ cjlin/papers/guide/guide.pdf), wherein, if numeral 1 is represented natural image, 0 represents computer generated image.
2. test phase
1. test pattern input---choose any natural image or computer generated image, form is jpeg, and picture size is one of any in from 400*600 to 1024*768;
2. image subblock screening module---as pending image, then to its piecemeal that carries out 64 * 64, concrete steps are 2. identical with the training stage step with test pattern intercepting 512 * 512 sizes.
3. the sub-piece fractal dimension of computed image---adopt above-mentioned difference box fractal dimension method to calculate the fractal dimension of above-mentioned 32 sub-pieces respectively, obtain one group of 32 dimensional feature vector;
4. SVM support vector machine identification module---quote the Libsvm software package herein and handle that (concrete software package can be downloaded http://www.csie.ntu.edu.tw/ ~ cjlin/libsvm) from here.Being proper vector with 32 of this single image, (idiographic flow is referring to http://www.csie.ntu.edu.tw/ ~ cjlin/papers/guide/guide.pdf) as the original input of SVM Discr., finally obtain authenticating value 1 or 0,1 differentiation is a natural image, and 0 is computer generated image.
Claims (4)
1, a kind of computer generated image passive detection method of fractal dimension, it is characterized in that, at first the every width of cloth image in natural image and the computer generated image storehouse is carried out sub-piece screening respectively in the training stage, then each image subblock is calculated its fractal dimension respectively, and then obtain an eigenvectors, at last should the series feature train to obtain the optimum classifier parameter with support vector machine; At test phase, at first testing image is carried out as above similar processing and obtain the fractal characteristic vector, the optimum classifier that utilizes the training stage to obtain is then classified and is differentiated and then obtain testing result.
2, the computer generated image passive detection method of fractal dimension according to claim 1 is characterized in that, the described training stage, comprises the steps:
1. for reducing complexity, at first with every width of cloth image interception 512 * 512 sub-pieces in natural image and the computer generated image storehouse as pending image, and calculate its variance ω;
2. the every width of cloth image division that obtains in inciting somebody to action 1. respectively is several image subblocks of 64 * 64, to each image subblock, calculates its variances sigma
iAgain relatively, if sub-piece variances sigma with the variance ω of itself and entire image
iLess than ω, it is slower for the variation of entire image to illustrate that then this sub-piece changes, and picture material is milder, thus with its deletion, otherwise keep this sub-piece, obtain L sub-piece through this step,
When L＞32, then from L sub-piece, give up (L-32) height piece of variance less than ω;
When L＜32, then from L sub-piece, choose variance 2. less than (32-L) height piece of ω and repeating step, finally obtain 32 the sub-pieces of variance greater than ω;
3. to above-mentioned 32 sub-pieces that obtain, adopt difference box fractal dimension method to calculate its fractal dimension respectively, obtain one group of 32 dimensional feature vector;
4. repeating step 1. to step 3., obtain 32 dimension fractal dimension proper vectors of every width of cloth image in natural image and the computer generated image storehouse, respectively computer generated image proper vector and natural image proper vector are sent into SVM and train, the sorter model that obtains after the training is used for follow-up decision.
3, the computer generated image passive detection method of fractal dimension according to claim 2 is characterized in that, the calculation procedure of described difference box fractal dimension is:
1) the sub-piece of M * M size is divided for the grid pointwise with s*s, chosen s=2 herein, 3, obtain the pixel maximal value I in every group of grid respectively
MaxWith minimum value I
Min
2) calculate corresponding to pixel extreme value in the grid of 2 and 3 multiple size based on dynamic programming method.2,4,8 to be example, the maximin of the grid interior pixel gray scale that previous step is calculated each 2 * 2 grid in rapid is preserved as separating of ground floor minor structure, and it is right to obtain 32 * 32 groups of maximin altogether at this moment; Ask the maximal value and the minimum value of 4 * 4 grid interior pixel gray scales, the sub-centering of maximin of corresponding 4 group of 2 * 2 grid of 4 * 4 grid position is searched and is got final product therewith in the ground floor minor structure, and the maximal value of trying to achieve, minimum value are preserved as separating of second layer minor structure; And the like, ask 2
i* 2
iThe maximin of the grid interior pixel gray scale during division.This step finally obtains with s=2, and 3,4,6,8,12 is the maximal value and the minimum value of the pairing pixel grey scale of grid of yardstick;
3) judge whether selected s can be divided exactly by M, as can not, then carry out the expansion of image subblock window by duplicating to enlarge, obtain the image block of (s*M) * (s*M),
Make r=s/M, image is imagined as curve in the three dimensions, then the xy plane is divided into the grid of many s * s, on each grid, is the box of a row s*s*s ', and wherein s ' is variable box height, s '=maxI
k-minI
k, I
k(k=1,2, L, n) size of n grey scale pixel value in expression s * s unit area;
If the (i, j) the pairing box number of grid is:
n
r(i，j)＝(maxI
k-minI
k)/r
Covering the required box number of entire image piece is
And the box number of M under the yardstick s * sub-piece of M original image is N
r=N
r'/s
2, cause the variation of r by the variation of s, use least square fitting lgN
r-lg (1/r), the slope of obtaining are box dimension D, i.e. image fractal dimension.
4, the computer generated image passive detection method of fractal dimension according to claim 2 is characterized in that, described test phase comprises the steps:
1. input image to be declared intercepts 512 * 512 image blocks and is used for subsequent treatment, and calculates its variance ω;
2. the image block that obtains in inciting somebody to action 1. is divided into several image subblocks of 64 * 64, to each image subblock, calculates its variances sigma
i, more relatively with the variance ω of itself and entire image, if sub-piece variances sigma
iLess than ω, it is slower for the variation of entire image to illustrate that then this sub-piece changes, and picture material is milder, thus with its deletion, otherwise keep this sub-piece, obtain L sub-piece through this step,
When L＞32, then from L sub-piece, give up (L-32) height piece of variance less than ω;
When L＜32, then from L sub-piece, choose variance 2. less than (32-L) height piece of ω and repeating step, finally obtain 32 the sub-pieces of variance greater than ω;
3. 32 sub-pieces to 2. being obtained by step calculate its fractal dimension by the calculation procedure of described fractal dimension respectively, finally obtain one group of 32 dimensional feature vector;
4. the sorter model that will be obtained by the 32 dimensional feature vectors input training stage that 3. step obtains detects processing, obtains final identification result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910195780A CN101655913A (en) | 2009-09-17 | 2009-09-17 | Computer generated image passive detection method based on fractal dimension |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910195780A CN101655913A (en) | 2009-09-17 | 2009-09-17 | Computer generated image passive detection method based on fractal dimension |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101655913A true CN101655913A (en) | 2010-02-24 |
Family
ID=41710198
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200910195780A Pending CN101655913A (en) | 2009-09-17 | 2009-09-17 | Computer generated image passive detection method based on fractal dimension |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101655913A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102621154A (en) * | 2012-04-10 | 2012-08-01 | 河海大学常州校区 | Method and device for automatically detecting cloth defects on line based on improved differential box multi-fractal algorithm |
| CN103150573A (en) * | 2012-12-24 | 2013-06-12 | 西交利物浦大学 | Nerve dendritic spine image classification method based on multiresolution fractal features |
| CN103775071A (en) * | 2014-01-15 | 2014-05-07 | 四川大学 | Method for measuring cranny evolvement of mining coal-rock mass |
| CN103808646A (en) * | 2014-02-07 | 2014-05-21 | 沈阳理工大学 | Image analyzer and method for measuring line fractal dimension of granular material |
| CN104296667A (en) * | 2014-11-07 | 2015-01-21 | 重庆邮电大学 | Micro electromechanical system in-plane displacement measuring method based on optimized box dimension image matching |
| CN107256571A (en) * | 2017-05-15 | 2017-10-17 | 北京理工大学 | A kind of Fractal Dimension Estimation based on deep learning Yu adaptive differential box |
| CN107274005A (en) * | 2017-05-19 | 2017-10-20 | 安徽四创电子股份有限公司 | The Urban Storm Flood disaster alarm method and system of fractal algorithm based on big data |
| CN108804848A (en) * | 2018-06-22 | 2018-11-13 | 西南石油大学 | A kind of computational methods of log box counting dimension |
| CN109444015A (en) * | 2018-10-31 | 2019-03-08 | 成都理工大学 | More kens, it is multiple dimensioned under shale reservoir microcellular system identification method |
| CN109658429A (en) * | 2018-12-21 | 2019-04-19 | 电子科技大学 | A kind of infrared image cirrus detection method based on boundary fractal dimension |
| CN110348459A (en) * | 2019-06-28 | 2019-10-18 | 西安理工大学 | Based on multiple dimensioned quick covering blanket method sonar image fractal characteristic extracting method |
| CN110555198A (en) * | 2018-05-31 | 2019-12-10 | 北京百度网讯科技有限公司 | method, apparatus, device and computer-readable storage medium for generating article |
| CN114443880A (en) * | 2022-01-24 | 2022-05-06 | 南昌市安厦施工图设计审查有限公司 | Picture examination method and picture examination system for large sample picture of fabricated building |
-
2009
- 2009-09-17 CN CN200910195780A patent/CN101655913A/en active Pending
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102621154A (en) * | 2012-04-10 | 2012-08-01 | 河海大学常州校区 | Method and device for automatically detecting cloth defects on line based on improved differential box multi-fractal algorithm |
| CN103150573A (en) * | 2012-12-24 | 2013-06-12 | 西交利物浦大学 | Nerve dendritic spine image classification method based on multiresolution fractal features |
| CN103150573B (en) * | 2012-12-24 | 2016-06-01 | 西交利物浦大学 | Based on the nerve dendritic spine image classification method of many resolving power fractal characteristic |
| CN103775071A (en) * | 2014-01-15 | 2014-05-07 | 四川大学 | Method for measuring cranny evolvement of mining coal-rock mass |
| CN103775071B (en) * | 2014-01-15 | 2016-08-24 | 四川大学 | Adopt the measuring method that coal and rock crack develops |
| CN103808646A (en) * | 2014-02-07 | 2014-05-21 | 沈阳理工大学 | Image analyzer and method for measuring line fractal dimension of granular material |
| CN103808646B (en) * | 2014-02-07 | 2016-04-06 | 沈阳理工大学 | Measure image analyzer and the method thereof of granular material stockline fractal dimension |
| CN104296667A (en) * | 2014-11-07 | 2015-01-21 | 重庆邮电大学 | Micro electromechanical system in-plane displacement measuring method based on optimized box dimension image matching |
| CN107256571A (en) * | 2017-05-15 | 2017-10-17 | 北京理工大学 | A kind of Fractal Dimension Estimation based on deep learning Yu adaptive differential box |
| CN107274005A (en) * | 2017-05-19 | 2017-10-20 | 安徽四创电子股份有限公司 | The Urban Storm Flood disaster alarm method and system of fractal algorithm based on big data |
| CN110555198A (en) * | 2018-05-31 | 2019-12-10 | 北京百度网讯科技有限公司 | method, apparatus, device and computer-readable storage medium for generating article |
| CN110555198B (en) * | 2018-05-31 | 2023-05-23 | 北京百度网讯科技有限公司 | Method, apparatus, device and computer readable storage medium for generating articles |
| CN108804848A (en) * | 2018-06-22 | 2018-11-13 | 西南石油大学 | A kind of computational methods of log box counting dimension |
| CN108804848B (en) * | 2018-06-22 | 2021-08-10 | 西南石油大学 | Method for calculating box dimension of logging curve |
| CN109444015A (en) * | 2018-10-31 | 2019-03-08 | 成都理工大学 | More kens, it is multiple dimensioned under shale reservoir microcellular system identification method |
| CN109658429A (en) * | 2018-12-21 | 2019-04-19 | 电子科技大学 | A kind of infrared image cirrus detection method based on boundary fractal dimension |
| CN110348459A (en) * | 2019-06-28 | 2019-10-18 | 西安理工大学 | Based on multiple dimensioned quick covering blanket method sonar image fractal characteristic extracting method |
| CN110348459B (en) * | 2019-06-28 | 2021-11-16 | 西安理工大学 | Sonar image fractal feature extraction method based on multi-scale rapid carpet covering method |
| CN114443880A (en) * | 2022-01-24 | 2022-05-06 | 南昌市安厦施工图设计审查有限公司 | Picture examination method and picture examination system for large sample picture of fabricated building |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101655913A (en) | Computer generated image passive detection method based on fractal dimension | |
| Gilani et al. | Learning from millions of 3D scans for large-scale 3D face recognition | |
| Majhi et al. | Novel features for off-line signature verification | |
| CN103136516B (en) | The face identification method that visible ray and Near Infrared Information merge and system | |
| CN107451510A (en) | Biopsy method and In vivo detection system | |
| CN106650830A (en) | Deep model and shallow model decision fusion-based pulmonary nodule CT image automatic classification method | |
| CN105096300B (en) | Method for checking object and equipment | |
| CN106296720A (en) | Human body based on binocular camera is towards recognition methods and system | |
| CN103390164A (en) | Object detection method based on depth image and implementing device thereof | |
| CN103729842A (en) | Fabric defect detection method based on local statistical characteristics and overall significance analysis | |
| CN101251896B (en) | Object detecting system and method based on multiple classifiers | |
| CN103020649A (en) | Forest type identification method based on texture information | |
| CN103729649A (en) | Image rotating angle detection method and device | |
| CN107240086B (en) | A kind of fabric defects detection method based on integral nomography | |
| CN108564111A (en) | A kind of image classification method based on neighborhood rough set feature selecting | |
| Kruthi et al. | Offline signature verification using support vector machine | |
| Winarno et al. | Multi-view faces detection using Viola-Jones method | |
| CN109190456A (en) | Pedestrian detection method is overlooked based on the multiple features fusion of converging channels feature and gray level co-occurrence matrixes | |
| CN103093243A (en) | High resolution panchromatic remote sensing image cloud discriminating method | |
| CN108446606A (en) | A kind of face critical point detection method based on acceleration binary features extraction | |
| CN114724218A (en) | Video detection method, device, equipment and medium | |
| CN109741351A (en) | A kind of classification responsive type edge detection method based on deep learning | |
| KR101584091B1 (en) | Multi-view object detection method using shared local features | |
| CN106446870B (en) | A kind of human body contour outline feature extracting method and device | |
| CN115797970B (en) | Dense pedestrian target detection method and system based on YOLOv5 model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20100224 |