CN108268896A - The nude picture detection method being combined based on HSV with SURF features - Google Patents
The nude picture detection method being combined based on HSV with SURF features Download PDFInfo
- Publication number
- CN108268896A CN108268896A CN201810049475.8A CN201810049475A CN108268896A CN 108268896 A CN108268896 A CN 108268896A CN 201810049475 A CN201810049475 A CN 201810049475A CN 108268896 A CN108268896 A CN 108268896A
- Authority
- CN
- China
- Prior art keywords
- surf
- rgb image
- image
- hsv
- normal
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
Abstract
The present invention provides a kind of nude picture detection method being combined based on HSV with SURF features, the skin area including obtaining sensitive RGB image and normal RGB image；The SURF visual vocabularies that the sensitive RGB image and the normal RGB image are obtained using SURF algorithm are expressed；The hsv color feature of the sensitive RGB image and the normal RGB image is obtained using hsv color model；Using SURF visual vocabularies expression and the hsv color feature as input parameter, BP neural network is trained；Export testing image recognition result.The present invention carries out nude picture detection using hsv color feature and the expression of SURF visual vocabularies, has the characteristics that processing speed is fast, accuracy rate is high.
Description
Technical field
The invention belongs to nude picture detection technical fields, are mutually tied based on HSV and SURF features in particular to one kind
The nude picture detection method of conjunction.
Background technology
With the fast development of Internet technology in recent years, network forum and portal website also rapidly develop, several
The every aspect of life is covered, thus the propagation of Internet picture information is also more and more extensive and easy, wherein harmful figure
The propagation of picture has a negative impact to teen-age physically and mentally healthy and social conduct.Due to forum post hair figure number crowd
More, forum administrator is allowed, which to audit all forum's pictures successively, will obviously consume a large amount of time and efforts, therefore a kind of effective
Nude picture detection method based on machine learning and machine vision seems outstanding for the workload for mitigating forum administrative staff
It is important.
Conventional method is based primarily upon the area accounting of skin area to differentiate sensitive image, but the method is easily to swimsuit etc.
Picture is judged by accident.Deep-neural-network is also commonly used for Image Classification Studies at present, but since this kind of mode of learning needs to carry out
It is a large amount of to calculate, it generally requires and arithmetic speed is accelerated using computer graphics processor (GPU).So currently the majority is quick
Sense image recognizing step is based primarily upon characteristics of image, generally includes feature extraction, training pattern and image and identifies 3 steps, and
For the prior art in feature extraction, mainly using SIFT corner features, processing speed is slow, and often results in picture feature and lack
It loses.
By being analyzed above it is found that existing nude picture detection method has the following disadvantages：
1st, the nude picture detection method of the prior art uses SIFT corner features, and processing speed is slow；
2nd, the nude picture detection method of the prior art be easy to cause picture feature missing, and accuracy rate is low.
Invention content
The present invention provides a kind of nude picture detection methods being combined based on HSV with SURF features, can effectively solve
The problem of certainly the nude picture detection method processing speed of the prior art is slow, additionally it is possible to solve the nude picture detection of the prior art
The problem of method accuracy rate is low.
In order to solve problem above, the present invention provides a kind of sensitive image knowledges being combined based on HSV with SURF features
Other method, technical solution are as follows：
A kind of nude picture detection method being combined based on HSV with SURF features, is included the following steps：
Step 1：Obtain the skin area of sensitive RGB image and normal RGB image；
Step 2：The SURF visual vocabularies of the sensitive RGB image and the normal RGB image are obtained using SURF algorithm
Expression；
Step 3：The hsv color that the sensitive RGB image and the normal RGB image are obtained using hsv color model is special
Sign；
Step 4：Using SURF visual vocabularies expression and the hsv color feature as input parameter, BP god is trained
Through network；
Step 5：Export testing image recognition result.
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：In step
In one, the human face region in the sensitive RGB image and the normal RGB image is detected using Haar-like features,
The skin area of the YCrCb color spaces detection judgement sensitive RGB image and the normal RGB image.
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：
The condition of the skin area of the YCrCb color spaces detection judgement sensitive RGB image and the normal RGB image is 77≤Cb
≤ 127 or 133≤Cr≤173.
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：In step
In two, the method that obtains SURF visual vocabularies expression include the use of SURF algorithm the sensitive RGB image and it is described just
The skin area extraction generation sensitive image SURF feature vectors and normal picture SURF feature vectors of normal RGB image；To described
The intersection of sensitive image SURF feature vectors and the normal picture SURF feature vectors carries out k-means clusters, and it is k to gather
Class, a cluster centre of common U={ u1, u2 ..uk }, k visual vocabulary calculate comparison word frequency Dw；Count the sensitive image
The occurrence number of the visual vocabulary of SURF feature vectors and the normal picture SURF feature vectors constructs SURF feature histograms
SurfBOWFeature carries out surfBOWFeature*Dw normalization.
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：It is described right
Calculation formula than word frequency Dw is：
FWPorn (i)=pCounts (i)/numPFeatures
FWNomal (i)=nCounts (i)/numNFeatures
Dw (i)=FWPorn* (FWPorn/FWNomal) 2
PCounts (i) is the number that visual vocabulary i occurs in the sensitive RGB image, and nCounts (i) is visual word
The number that remittance i occurs in the normal RGB image, numPFeatures are the sensitive RGB image and the normal RGB figures
The sum of sensitive features as in, numNFeatures are total for normal characteristics in the sensitive RGB image and the normal RGB image
Number.
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：In step
In three, 72 D feature vectors are constructed as the sensitive RGB image and described according to color quantization formula and HSVG=9H+3S+V
The hsv color feature of normal RGB image, 72 D feature vectors are：
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：In step
In four, the input parameter of the BP neural network is [SURF*0.8, HSV*0.2].
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：In step
Judged in five according to step 1, if the human face region of the testing image is not more than the 1/2 of the skin area, sentenced
The fixed testing image is normal picture.
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：In step
Judged in five according to step 1, if the gross area that the skin area of the testing image accounts for the testing image is not more than
When 10%, then judge the testing image for normal picture.
The nude picture detection method being combined as described above based on HSV with SURF features, further preferably：In step
When testing image described in five is unsatisfactory for the normal picture decision condition of step 1, is exported and schemed according to the BP neural network
As recognition result.
Analysis is it is found that compared with prior art, the advantages of the present invention are：
1st, it is provided by the invention based on HSV when the nude picture detection method that SURF features are combined is in feature extraction,
Using SURF corner features, have rotation translation indeformable and have preferable fault-tolerance to brightness change and noise, in image spy
There is good effect in sign extraction and identification field, and processing speed faster, has merged the angle point information of image and global color letter
Breath is as characteristics of image, and accuracy rate is high, so that the present invention has the characteristics that processing speed is fast, accuracy rate is high.
2nd, it is provided by the invention that training BP nerves are used with the nude picture detection method that SURF features are combined based on HSV
The method of network expresses the visual vocabulary of sensitive features, convenient and efficient when judging, so that the present invention has processing
Fireballing feature.
3rd, it is provided by the invention to be made based on HSV with the nude picture detection method that SURF features are combined using HSV systems
For image classification feature, closer perception of the people to color, so that the present invention has the characteristics that accuracy rate is high.
Description of the drawings
Fig. 1 is the overview flow chart of the present invention.
Fig. 2 is the body skin extracted region flow chart of the present invention.
Fig. 3 is the body skin zone marker design sketch of the present invention.
The SURF characteristic visuals lexicon model that Fig. 4 is the present invention expresses flow chart.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only the part of the embodiment of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, those of ordinary skill in the art obtained without making creative work it is all its
His embodiment, shall fall within the protection scope of the present invention.
As shown in Figure 1, provided by the invention included such as based on HSV with the nude picture detection method that SURF features are combined
Lower step：
Step 1：Obtain the skin area of sensitive RGB image and normal RGB image.
1.1 read in the sensitive RGB image of 320*280 and normal RGB image, utilize Haar-like features detection sensitivity RGB
Human face region in image and normal RGB image creates the identical Mask variables of the wide height of image, and human face region puts 1, remaining region
It sets to 0；
1.2 in YCrCb color spaces carry out detection of skin regions to sensitive RGB image and normal RGB image, and (wherein Y is
Luminance component, Cb are chroma blue component, and Cr is red chrominance component), according to 77≤Cb≤127 or 133≤Cr≤173 pair skin
Skin region is judged that Mask variables put 1 in area of skin color, other regions are set to 0, and design sketch is as shown in Figure 3.
Step 2：As shown in figure 4, the SURF that sensitive RGB image and the normal RGB image are obtained using SURF algorithm is regarded
Feel lexical representation.
2.1 detect characteristic point using SURF algorithm in sensitive RGB image and normal RGB image, and 1 is put in Mask variables
Region generates 128 dimension sensitive image SURF feature vectors and 128 dimension normal picture SURF feature vectors respectively, traverses all sensitivities
RGB image and normal RGB image tie up sensitive image SURF feature vectors and 128 dimension normal picture SURF features by the 128 of generation
Vector is stored in respectively in two two-dimensional arrays；
All 128 dimension sensitive image SURF feature vectors and 128 dimension normal picture SURF feature vectors are carried out group by 2.2
It closes, carries out k-means clusters, gather for 128 classes, a cluster centre of common U={ u1, u2 ..u128 }, 128 visual vocabularies, root
Comparison word frequency Dw is calculated according to following equation：
FWPorn (i)=pCounts (i)/numPFeatures
FWNomal (i)=nCounts (i)/numNFeatures
Dw1 (i)=FWPorn* (FWPorn/FWNomal) 2
PCounts (i) is the number that visual vocabulary i occurs in sensitive RGB image, and nCounts (i) is visual vocabulary i
The number occurred in normal RGB image, numPFeatures are sensitive features in sensitive RGB image and normal RGB image
Sum, numNFeatures are normal characteristics sum in sensitive RGB image and normal RGB image；
2.3 calculate each sensitive image SURF feature vectors and normal picture SURF feature vectors to each visual vocabulary
Euclidean distance, Euclidean distance minimum are each sensitive image SURF feature vectors and each normal picture SURF feature vectors institute
The visual vocabulary of category counts sensitive image SURF feature vectors and normal picture in every width sensitivity RGB image and normal RGB image
The occurrence number of the visual vocabulary of SURF feature vectors, the SURF feature histogram surfBOWFeature of 128 dimension of construction, carries out
SurfBOWFeature*Dw is normalized, and obtains the expression of SURF visual vocabularies.Normalizing formula is：
Step 3：The hsv color feature of sensitive RGB image and normal RGB image is obtained using hsv color model.
According to color quantization formula and HSVG=9H+3S+V 72 D feature vectors of construction as sensitive RGB image and normally
The hsv color feature of RGB image.
Step 4：Using the expression of SURF visual vocabularies and hsv color feature as input parameter [SURF*0.8, HSV*
0.2], training BP neural network.
In the present invention using BP neural network come training pattern, hidden layer contains 20 neurons, and output layer uses
Sigmoid exports the neural network for 2.
Step 5：Export testing image recognition result.
5.1 are judged according to step 1, if the human face region of testing image is not more than the 1/2 of skin area, judgement
Testing image is normal picture；
5.2 are judged according to step 1, if the gross area that the skin area of testing image accounts for testing image is not more than
When 10%, judgement testing image is normal picture；
It is defeated according to the BP neural network of step 4 when 5.3 testing images are unsatisfactory for the normal picture decision condition of step 1
Go out image recognition result.
As shown in Figure 1, the present invention excludes people by YCrCb features of skin colors and the method for detecting human face based on Haar features
Face region and only extraction body region extract SURF corner features, using SURF as interest region (ROI) in interest region
Corner feature has rotation translation indeformable and has preferable fault-tolerance to brightness change and noise, in image characteristics extraction and
There is good effect in identification field, and processing speed faster, using K-means clusters constructs BOW bag of words moulds to SURF corner features
Type and the comparison word frequency for counting normal RGB image and sensitive RGB image, with reference to global hsv color feature, closer to people to color
Perception, so as to constitute the positive and negative training sample set of sensitive image feature.The angle point information and the overall situation of image are merged
Color information as characteristics of image, accuracy rate is high.Using BP neural network training sample, structural classification model judges when side
Just quick, processing speed is fast.By extracting the color of image, angle point textural characteristics train classification models, with 1520 normograms
Picture and 1600 sensitive images are tested, and normal picture detection accuracy is 86.2%, and sensitive image detection accuracy is
86.6%.
Analysis is it is found that compared with prior art, the advantages of the present invention are：
1st, it is provided by the invention based on HSV when the nude picture detection method that SURF features are combined is in feature extraction,
Using SURF corner features, have rotation translation indeformable and have preferable fault-tolerance to brightness change and noise, in image spy
There is good effect in sign extraction and identification field, and processing speed faster, has merged the angle point information of image and global color letter
Breath is as characteristics of image, and accuracy rate is high, so that the present invention has the characteristics that processing speed is fast, accuracy rate is high.
2nd, it is provided by the invention that training BP nerves are used with the nude picture detection method that SURF features are combined based on HSV
The method of network expresses the visual vocabulary of sensitive features, convenient and efficient when judging, so that the present invention has processing
Fireballing feature.
3rd, it is provided by the invention to be made based on HSV with the nude picture detection method that SURF features are combined using HSV systems
For image classification feature, closer perception of the people to color, so that the present invention has the characteristics that accuracy rate is high.
As known by the technical knowledge, the present invention can pass through the embodiment party of other essence without departing from its spirit or essential feature
Case is realized.Therefore, embodiment disclosed above, all things considered are all merely illustrative, not the only.Institute
Have within the scope of the present invention or be included in the invention in the change being equal in the scope of the present invention.
Claims (10)
- A kind of 1. nude picture detection method being combined based on HSV with SURF features, which is characterized in that include the following steps：Step 1：Obtain the skin area of sensitive RGB image and normal RGB image；Step 2：The SURF visual vocabulary tables of the sensitive RGB image and the normal RGB image are obtained using SURF algorithm It reaches；Step 3：The hsv color feature of the sensitive RGB image and the normal RGB image is obtained using hsv color model；Step 4：Using SURF visual vocabularies expression and the hsv color feature as input parameter, BP nerve nets are trained Network；Step 5：Export testing image recognition result.
- 2. the nude picture detection method according to claim 1 being combined based on HSV with SURF features, feature are existed In：In step 1, the face in the Haar-like features detection sensitive RGB image and the normal RGB image is used Region, the skin area of the detection judgement sensitive RGB image and the normal RGB image in YCrCb color spaces.
- 3. the nude picture detection method according to claim 2 being combined based on HSV with SURF features, feature are existed In：In YCrCb color spaces, the condition of the skin area of the detection judgement sensitive RGB image and the normal RGB image is 77≤Cb≤127 or 133≤Cr≤173.
- 4. the nude picture detection method according to claim 1 or 2 being combined based on HSV with SURF features, feature It is：In step 2, the method for obtaining the SURF visual vocabularies expression includes：Using SURF algorithm in the skin area of the sensitive RGB image and normal RGB image extraction generation sensitive image SURF feature vectors and normal picture SURF feature vectors；K-means is carried out to the intersection of the sensitive image SURF feature vectors and the normal picture SURF feature vectors to gather Class is gathered for k class, a cluster centre of common U={ u1, u2 ..uk }, k visual vocabulary, calculating comparison word frequency Dw；Count the visual vocabulary of the sensitive image SURF feature vectors and the normal picture SURF feature vectors goes out occurrence Number constructs SURF feature histogram surfBOWFeature, carries out surfBOWFeature*Dw normalization.
- 5. the nude picture detection method according to claim 4 being combined based on HSV with SURF features, feature are existed In：The calculation formula of the comparison word frequency Dw is：FWPorn (i)=pCounts (i)/numPFeaturesFWNomal (i)=nCounts (i)/numNFeaturesDw (i)=FWPorn* (FWPorn/FWNomal) 2PCounts (i) is the number that visual vocabulary i occurs in the sensitive RGB image, and nCounts (i) is visual vocabulary i The number occurred in the normal RGB image, numPFeatures are the sensitive RGB image and the normal RGB image The sum of middle sensitive features, numNFeatures are total for normal characteristics in the sensitive RGB image and the normal RGB image Number.
- 6. the nude picture detection method according to claim 1 or 2 being combined based on HSV with SURF features, feature It is：In step 3,72 D feature vectors are constructed as the sensitivity RGB according to color quantization formula and HSVG=9H+3S+V The hsv color feature of image and the normal RGB image.
- 7. the nude picture detection method according to claim 1 or 2 being combined based on HSV with SURF features, feature It is：In step 4, the input parameter of the BP neural network is [SURF*0.8, HSV*0.2].
- 8. the nude picture detection method according to claim 2 being combined based on HSV with SURF features, feature are existed In：Judged in step 5 according to step 1, if the human face region of the testing image is no more than the skin area When 1/2, then judge the testing image for normal picture.
- 9. the nude picture detection method according to claim 2 being combined based on HSV with SURF features, feature are existed In：Judged in step 5 according to step 1, if the skin area of the testing image accounts for total face of the testing image When product is no more than 10%, then judge the testing image for normal picture.
- 10. the nude picture detection method being combined based on HSV with SURF features according to claim 8 or claim 9, feature It is：When the testing image described in step 5 is unsatisfactory for the decision condition of the normal picture of step 1, according to BP god Image recognition result is exported through network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810049475.8A CN108268896A (en) | 2018-01-18 | 2018-01-18 | The nude picture detection method being combined based on HSV with SURF features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810049475.8A CN108268896A (en) | 2018-01-18 | 2018-01-18 | The nude picture detection method being combined based on HSV with SURF features |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108268896A true CN108268896A (en) | 2018-07-10 |
Family
ID=62776131
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810049475.8A Pending CN108268896A (en) | 2018-01-18 | 2018-01-18 | The nude picture detection method being combined based on HSV with SURF features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108268896A (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001059612A2 (en) * | 2000-02-14 | 2001-08-16 | Adscience Limited | Improvements relating to data filtering |
| CN101763502A (en) * | 2008-12-24 | 2010-06-30 | 中国科学院自动化研究所 | High-efficiency method and system for sensitive image detection |
| CN101996314A (en) * | 2009-08-26 | 2011-03-30 | 厦门市美亚柏科信息股份有限公司 | Content-based human body upper part sensitive image identification method and device |
| CN102938054A (en) * | 2012-09-06 | 2013-02-20 | 北京工业大学 | Method for recognizing compressed-domain sensitive images based on visual attention models |
| CN103093467A (en) * | 2013-01-21 | 2013-05-08 | 杭州电子科技大学 | Shot boundary detection method based on double detection model |
| CN103679132A (en) * | 2013-07-15 | 2014-03-26 | 北京工业大学 | A sensitive image identification method and a system |
| US20140254874A1 (en) * | 2011-08-31 | 2014-09-11 | Metaio Gmbh | Method of detecting and describing features from an intensity image |
| US9003445B1 (en) * | 2012-05-10 | 2015-04-07 | Google Inc. | Context sensitive thumbnail generation |
| US20150294188A1 (en) * | 2014-02-19 | 2015-10-15 | Nant Holdings Ip, Llc | Invariant-Based Dimensional Reduction Of Object Recognition Features, Systems And Methods |
| CN105389593A (en) * | 2015-11-16 | 2016-03-09 | 上海交通大学 | Image object recognition method based on SURF |
| CN105678730A (en) * | 2014-11-17 | 2016-06-15 | 西安三茗科技有限责任公司 | Camera movement self-detecting method on the basis of image identification |
| US20170032224A1 (en) * | 2015-07-31 | 2017-02-02 | Xiaomi Inc. | Method, device and computer-readable medium for sensitive picture recognition |
| CN106664376A (en) * | 2014-06-10 | 2017-05-10 | 2Mee 有限公司 | Augmented reality apparatus and method |
-
2018
- 2018-01-18 CN CN201810049475.8A patent/CN108268896A/en active Pending
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001059612A2 (en) * | 2000-02-14 | 2001-08-16 | Adscience Limited | Improvements relating to data filtering |
| CN101763502A (en) * | 2008-12-24 | 2010-06-30 | 中国科学院自动化研究所 | High-efficiency method and system for sensitive image detection |
| CN101996314A (en) * | 2009-08-26 | 2011-03-30 | 厦门市美亚柏科信息股份有限公司 | Content-based human body upper part sensitive image identification method and device |
| US20140254874A1 (en) * | 2011-08-31 | 2014-09-11 | Metaio Gmbh | Method of detecting and describing features from an intensity image |
| US9003445B1 (en) * | 2012-05-10 | 2015-04-07 | Google Inc. | Context sensitive thumbnail generation |
| CN102938054A (en) * | 2012-09-06 | 2013-02-20 | 北京工业大学 | Method for recognizing compressed-domain sensitive images based on visual attention models |
| CN103093467A (en) * | 2013-01-21 | 2013-05-08 | 杭州电子科技大学 | Shot boundary detection method based on double detection model |
| CN103679132A (en) * | 2013-07-15 | 2014-03-26 | 北京工业大学 | A sensitive image identification method and a system |
| US20150294188A1 (en) * | 2014-02-19 | 2015-10-15 | Nant Holdings Ip, Llc | Invariant-Based Dimensional Reduction Of Object Recognition Features, Systems And Methods |
| CN106664376A (en) * | 2014-06-10 | 2017-05-10 | 2Mee 有限公司 | Augmented reality apparatus and method |
| CN105678730A (en) * | 2014-11-17 | 2016-06-15 | 西安三茗科技有限责任公司 | Camera movement self-detecting method on the basis of image identification |
| US20170032224A1 (en) * | 2015-07-31 | 2017-02-02 | Xiaomi Inc. | Method, device and computer-readable medium for sensitive picture recognition |
| CN105389593A (en) * | 2015-11-16 | 2016-03-09 | 上海交通大学 | Image object recognition method based on SURF |
Non-Patent Citations (4)
| Title |
|---|
| YIZHI LIU等: "Constructing SURF Visual-Words for Pornographic Images Detection", 《2009 12TH INTERNATIONAL CONFERENCE ON COMPUTER AND INFORMATION TECHNOLOGY (ICCIT 2009)》 * |
| YIZHI LIU等: "Constructing SURF Visual-Words for Pornographic Images Detection", 《2009 12TH INTERNATIONAL CONFERENCE ON COMPUTER AND INFORMATION TECHNOLOGY (ICCIT 2009)》, 23 December 2009 (2009-12-23), pages 404 - 407, XP031624645 * |
| 彭浩林等: "基于内容的敏感图像判别模型的设计", 《图形图像》, pages 48 - 49 * |
| 王小华等: "基于YCgCr颜色空间的不良图像肤色检测", 《计算机应用与软件》, vol. 28, no. 4, pages 151 - 152 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107103298B (en) | Pull-up counting system and method based on image processing | |
| Kivinen et al. | Visual boundary prediction: A deep neural prediction network and quality dissection | |
| CN103942577B (en) | Based on the personal identification method for establishing sample database and composite character certainly in video monitoring | |
| Osareh | Automated identification of diabetic retinal exudates and the optic disc | |
| CN107423690A (en) | A kind of face identification method and device | |
| Gizatdinova et al. | Feature-based detection of facial landmarks from neutral and expressive facial images | |
| CN107403142B (en) | A kind of detection method of micro- expression | |
| CN108280454A (en) | The nude picture detection method being combined with LBP features based on HSV | |
| CN110334781A (en) | A kind of zero sample learning algorithm based on Res-Gan | |
| CN107066972B (en) | Natural scene Method for text detection based on multichannel extremal region | |
| CN110874587B (en) | Face characteristic parameter extraction system | |
| CN112396573A (en) | Facial skin analysis method and system based on image recognition | |
| CN108629336A (en) | Face value calculating method based on human face characteristic point identification | |
| CN112200154A (en) | Face recognition method and device for mask, electronic equipment and storage medium | |
| CN103093180A (en) | Method and system for detecting pornography images | |
| Arif et al. | Human pose estimation and object interaction for sports behaviour | |
| CN108133219A (en) | The nude picture detection method being combined based on HSV, SURF with LBP features | |
| Fernando et al. | Novel approach to use HU moments with image processing techniques for real time sign language communication | |
| Gocławski et al. | Neural network segmentation of images from stained cucurbits leaves with colour symptoms of biotic and abiotic stresses | |
| Paul et al. | PCA based geometric modeling for automatic face detection | |
| CN110348320A (en) | A kind of face method for anti-counterfeit based on the fusion of more Damage degrees | |
| CN113221655A (en) | Face spoofing detection method based on feature space constraint | |
| CN108268896A (en) | The nude picture detection method being combined based on HSV with SURF features | |
| Chen et al. | Contour detection by simulating the curvature cell in the visual cortex and its application to object classification | |
| CN107066928A (en) | A kind of pedestrian detection method and system based on grader |
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 |