CN107358147A - Face recognition features' extraction algorithm based on local circulation graph structure - Google Patents
Face recognition features' extraction algorithm based on local circulation graph structure Download PDFInfo
- Publication number
- CN107358147A CN107358147A CN201710362000.XA CN201710362000A CN107358147A CN 107358147 A CN107358147 A CN 107358147A CN 201710362000 A CN201710362000 A CN 201710362000A CN 107358147 A CN107358147 A CN 107358147A
- Authority
- CN
- China
- Prior art keywords
- graph structure
- neighborhood
- face
- face recognition
- extraction algorithm
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of face recognition features' extraction algorithm based on local circulation graph structure, its technical characteristics is：In a width face gray level image, the neighborhood of 3 × 3 sizes is chosen, and corresponding pixel value is set；Circulation builds graph structure in certain sequence in selected neighborhood；According to the order of the graph structure size of compared pixels value successively, and use binary representation in the direction of arrows；Calculate the characteristic value of center pixel in neighborhood.The present invention is reasonable in design, can describe face characteristic exactly, make the characteristic value of resulting object pixel more representative, improve discrimination；The characteristic value that the present invention obtains the characteristic information of facial image is described more comprehensively, therefore effective recognition of face can be carried out, at the same when consumption in terms of relative to the symmetrical Local map structure algorithm based on multi-direction weight optimization also there is advantage.
Description
Technical field
The invention belongs to technical field of image processing, especially a kind of face recognition features based on local circulation graph structure
Extraction algorithm (LCGS).
Background technology
One complete face identification system includes three big module of target detection, feature extraction and recognition of face.Wherein, energy
The no effective face characteristic information of extraction is a very important step.The algorithm of extraction face characteristic has many kinds, generally may be used
Global characteristics are based on to be divided into and based on two kinds of local feature.Research shows, based on the extracting method of local feature in discrimination
Aspect outline is higher than the extracting method based on global characteristics.2011, inspired by dominant set in graph structure, Abusham etc.
People is applied it in face characteristic extraction, it is proposed that Local map construction operator (LGS).The operator be in 3 × 4 neighborhood,
5 pixels around center pixel are extracted, form a graph structure, then by two pixels in center pixel and its left side by inverse
Clockwise compares size successively, and center pixel and its right side adjacent pixel are compared to each other into size, finally that center pixel is right
Three pixels of side can so obtain 8 binary values according to size is compared successively clockwise, using it as in imago
The characteristic value of element.This method efficiently utilizes the characteristic information of object pixel surrounding pixel, can improve discrimination, but by
In it only just with the Pixel Information at left and right sides of object pixel, the information in other directions is not analyzed, and
An effective weight is not assigned to obtained binary coding, in terms of discrimination or Shortcomings.
In order to improve this deficiency, 2015, Dong Song et al. proposed the symmetrical Local map based on multi-direction weight optimization
Structure algorithm (MOW-SLGS), in 5 × 5 neighborhoods, 0 °, 45 °, 90 ° and 135 ° four direction around object pixel is calculated special
Value indicative, and its optimal weights is assigned, eigenvalue of maximum is finally taken as object pixel characteristic value.This method makes discrimination have
One step improves.
But LGS and MOW-SLGS takes 3 × 4 and 5 × 5 neighborhoods respectively when carrying out neighborhood division, but it is special calculating
During value indicative, the information of neighborhood inner periphery pixel is not all used, causes loss of learning, so as to which final face can be caused to know
Other effect is unsatisfactory.
The content of the invention
It is an object of the invention to overcome the deficiencies in the prior art, proposes that one kind is reasonable in design, discrimination is high and identification speed
The fast face recognition features' extraction algorithm based on local circulation graph structure of degree.
The present invention solves its technical problem and takes following technical scheme to realize：
A kind of face recognition features' extraction algorithm based on local circulation graph structure, comprises the following steps：
Step 1：In a width face gray level image, the neighborhood of 3 × 3 sizes is chosen, from left to right and from up in neighborhood
It is lower to be respectively：X1、X2、X3、X8、X0、X4、X5、X6、X7, and corresponding pixel value is set；
Step 2：Circulation builds graph structure in certain sequence in selected neighborhood；
Step 3：According to the order of the graph structure size of compared pixels value successively, and use binary representation in the direction of arrows；
Step 4：The characteristic value of center pixel in neighborhood is calculated according to obtained binary numeral.
Further, step 2 graph structure be in the following order structure and it is indicated with arrows：
X0→X8→X1→X0→X2→X3→X0→X4→X5→X0→X6→X7→X0。
Further, the step 3 is with the mode of binary representation：When pixel value becomes big between two pixels of arrow instruction
When, the binary system is set to 1, and when pixel value change is small, then binary system is set to 0.
Further, the specific method of the step 4 is：By binary numeral according to X0→X8→X1→X0→X2→X3→X0
→X4→X5→X0→X6→X7→X0Direction arrangement, then turn to decimal number, obtain final pixel value.
The advantages and positive effects of the present invention are：
The present invention is reasonable in design, and neighborhood is set to traditional 3 × 3 and by object pixel surrounding loop structural map knot by it
Structure so that the information of neighborhood inner periphery pixel can be utilized substantially effectively, can describe face characteristic exactly, so that
The characteristic value of resulting object pixel is more representative, allows final discrimination to be improved.Due to this obtained characteristic value
The characteristic information of facial image is described more comprehensively, therefore effective recognition of face can be carried out, at the same when consumption in terms of
Also there is advantage relative to the symmetrical Local map structure algorithm based on multi-direction weight optimization.
Brief description of the drawings
Fig. 1 is the schematic diagram that the present invention chooses neighborhood；
Fig. 2 is the schematic diagram of present invention structure graph structure；
Fig. 3 is the instance graph that the present invention chooses neighborhood；
Fig. 4 is the instance graph of present invention structure graph structure；
Fig. 5 is the present invention and the discrimination comparison diagram of LBP, LGS, MOW-SLGS on YALE face databases；
Fig. 6 is time-consuming comparison figure of the present invention with LBP, LGS, MOW-SLGS processing with piece image.
Embodiment
The embodiment of the present invention is further described below in conjunction with accompanying drawing：
A kind of face recognition features' extraction algorithm based on local circulation graph structure, comprises the following steps：
Step 1：In a width face gray level image, the neighborhood of 3 × 3 sizes is chosen, as shown in Figure 1.
In the present embodiment, selected neighborhood (nine pixels from left to right and from top to bottom：X1、X2、X3、X8、X0、X4、
X5、X6、X7) pixel value be respectively 8,18,26,13,22,28,21,31,20, as shown in Figure 3.
Step 2：The circulation structure graph structure in selected neighborhood, as shown in Figure 2.
In the present embodiment, the graph structure of structure is circulated as shown in figure 4, the order of the graph structure is X0→X8→X1→X0
→X2→X3→X0→X4→X5→X0→X6→X7→X0。
Step 3：According to the order of the graph structure size of compared pixels value successively, and use binary representation in the direction of arrows.
In this step, according to listed pixel value in instance graph, its size is compared successively, according to pixel value in the direction of the arrow
Change puts 1, diminished greatly, and the principle set to 0, it is respectively 0,0,1,0,1,0,1,0,1,1,0,1 to obtain binary numeral.
Step 4：Calculate the characteristic value of center pixel in neighborhood.
The circular of this step is：A string of binary values that step 3 is obtained arrange as follows：
X0→X8→X1→X0→X2→X3→X0→X4→X5→X0→X6→X7→X0
Then decimal number is converted into, obtains final pixel value.
By the calculating of this step, final characteristic value, i.e., 001010101101 can be obtained.It is translated into the decimal system
Value, is 0 × 2048+0 × 1024+1 × 512+0 × 256+1 × 128+0 × 64+1 × 32+0 × 16+1 × 8+1 × 4+0 × 2+1
× 1=685, then the characteristic value of final goal pixel is 685.
Further checking can be done to the present invention by experimental result.As shown in figure 5, on YALE databases, work as training
When sample data takes 1,2,3,4,5, the discrimination of (LCGS algorithms) of the invention is above LBP, LGS, MOW-SLGS algorithm, works as instruction
Practice sample number when measuring 1, the discriminations of LCGS algorithms can reach 83.13%, and LBP, LGS and MOW-SLGS algorithm
Discrimination is respectively 75.33%, 80.00%, 74.27%.Afterwards, gradually increase with the quantity of training sample, to face
Discrimination is also constantly increasing, and when the quantity of training sample takes 5, the discrimination of LCGS algorithms then reaches 99.44%, the same period
The discrimination of LBP, LGS and MOW-SLGS algorithm is respectively 96.78%, 96.67%, 97.11%.
As shown in fig. 6, it is shown that handled using algorithms of different with piece image spent time.It is as can be seen that of the invention
Take as 0.4493s, the time-consuming of LBP, LGS, MOW-SLGS algorithm is 0.2870s, 0.2886s, 1.0093s respectively.This hair
Although bright taking is higher than LBP and LGS, well below MOW-SLGS algorithms.
The present invention overcomes the problem of their feature information extractions are insufficient for LGS and MOW-SLGS algorithms,
Make the feature of extraction that more fully face be described.Therefore both algorithms are superior to the discrimination of facial image,
And when consumption also well below MOW-SLGS algorithms.
It is emphasized that embodiment of the present invention is illustrative, rather than it is limited, therefore present invention bag
Include and be not limited to embodiment described in embodiment, it is every by those skilled in the art's technique according to the invention scheme
The other embodiment drawn, also belongs to the scope of protection of the invention.
Claims (4)
1. a kind of face recognition features' extraction algorithm based on local circulation graph structure, it is characterised in that comprise the following steps：
Step 1：In a width face gray level image, choose the neighborhood of 3 × 3 sizes, in neighborhood from left to right and from top to bottom as
Element is respectively：X1、X2、X3、X8、X0、X4、X5、X6、X7, and corresponding pixel value is set；
Step 2：Circulation builds graph structure in certain sequence in selected neighborhood；
Step 3：According to the order of the graph structure size of compared pixels value successively, and use binary representation in the direction of arrows；
Step 4：The characteristic value of center pixel in neighborhood is calculated according to obtained binary numeral.
2. face recognition features' extraction algorithm according to claim 1 based on local circulation graph structure, it is characterised in that：
Step 2 graph structure be in the following order structure and it is indicated with arrows：
X0→X8→X1→X0→X2→X3→X0→X4→X5→X0→X6→X7→X0。
3. face recognition features' extraction algorithm according to claim 1 based on local circulation graph structure, it is characterised in that：
The step 3 is with the mode of binary representation：When pixel value becomes big between two pixels of arrow instruction, the binary system is set to
1, when pixel value change is small, then binary system is set to 0.
4. face recognition features' extraction algorithm according to claim 1 based on local circulation graph structure, it is characterised in that：
The specific method of the step 4 is：By binary numeral according to X0→X8→X1→X0→X2→X3→X0→X4→X5→X0→X6
→X7→X0Direction arrangement, then turn to decimal number, obtain final pixel value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710362000.XA CN107358147A (en) | 2017-05-22 | 2017-05-22 | Face recognition features' extraction algorithm based on local circulation graph structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710362000.XA CN107358147A (en) | 2017-05-22 | 2017-05-22 | Face recognition features' extraction algorithm based on local circulation graph structure |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107358147A true CN107358147A (en) | 2017-11-17 |
Family
ID=60272208
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710362000.XA Pending CN107358147A (en) | 2017-05-22 | 2017-05-22 | Face recognition features' extraction algorithm based on local circulation graph structure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107358147A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108090460A (en) * | 2017-12-29 | 2018-05-29 | 天津科技大学 | Expression recognition feature extraction algorithm based on multi-direction description of weber |
| CN108446686A (en) * | 2018-05-28 | 2018-08-24 | 天津科技大学 | A kind of face recognition features' extraction algorithm based on image local graph structure |
| CN108446679A (en) * | 2018-05-07 | 2018-08-24 | 天津科技大学 | Facial expression recognition feature extraction algorithm based on central symmetry partial gradient coding |
| CN108509927A (en) * | 2018-04-09 | 2018-09-07 | 中国民航大学 | A kind of finger venous image recognition methods based on Local Symmetric graph structure |
| CN108596126A (en) * | 2018-04-28 | 2018-09-28 | 中国民航大学 | A kind of finger venous image recognition methods based on improved LGS weighted codings |
| CN109598295A (en) * | 2018-11-23 | 2019-04-09 | 贵州宇鹏科技有限责任公司 | A kind of learning method for image characteristics extraction |
| CN111931590A (en) * | 2020-07-15 | 2020-11-13 | 杭州电子科技大学 | Balanced face feature extraction method of self-adaptive four-corner star local graph structure |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102163283A (en) * | 2011-05-25 | 2011-08-24 | 电子科技大学 | Method for extracting face characteristic based on local three-value mode |
| CN102262773A (en) * | 2010-05-29 | 2011-11-30 | 深圳宝嘉电子设备有限公司 | Dual-threshold image lossless data embedding method |
| CN106096557A (en) * | 2016-06-15 | 2016-11-09 | 浙江大学 | A kind of semi-supervised learning facial expression recognizing method based on fuzzy training sample |
| CN106529447A (en) * | 2016-11-03 | 2017-03-22 | 河北工业大学 | Small-sample face recognition method |
-
2017
- 2017-05-22 CN CN201710362000.XA patent/CN107358147A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102262773A (en) * | 2010-05-29 | 2011-11-30 | 深圳宝嘉电子设备有限公司 | Dual-threshold image lossless data embedding method |
| CN102163283A (en) * | 2011-05-25 | 2011-08-24 | 电子科技大学 | Method for extracting face characteristic based on local three-value mode |
| CN106096557A (en) * | 2016-06-15 | 2016-11-09 | 浙江大学 | A kind of semi-supervised learning facial expression recognizing method based on fuzzy training sample |
| CN106529447A (en) * | 2016-11-03 | 2017-03-22 | 河北工业大学 | Small-sample face recognition method |
Non-Patent Citations (1)
| Title |
|---|
| EIMAD E.A ET AL: "Face Recognition Using Local Graph Structure(LGS)", 《SPRINGER-VERLAG BERLIN HEIDELBERG 2011》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108090460A (en) * | 2017-12-29 | 2018-05-29 | 天津科技大学 | Expression recognition feature extraction algorithm based on multi-direction description of weber |
| CN108509927A (en) * | 2018-04-09 | 2018-09-07 | 中国民航大学 | A kind of finger venous image recognition methods based on Local Symmetric graph structure |
| CN108509927B (en) * | 2018-04-09 | 2021-09-07 | 中国民航大学 | Finger vein image identification method based on local symmetrical graph structure |
| CN108596126A (en) * | 2018-04-28 | 2018-09-28 | 中国民航大学 | A kind of finger venous image recognition methods based on improved LGS weighted codings |
| CN108596126B (en) * | 2018-04-28 | 2021-09-14 | 中国民航大学 | Finger vein image identification method based on improved LGS weighted coding |
| CN108446679A (en) * | 2018-05-07 | 2018-08-24 | 天津科技大学 | Facial expression recognition feature extraction algorithm based on central symmetry partial gradient coding |
| CN108446686A (en) * | 2018-05-28 | 2018-08-24 | 天津科技大学 | A kind of face recognition features' extraction algorithm based on image local graph structure |
| CN109598295A (en) * | 2018-11-23 | 2019-04-09 | 贵州宇鹏科技有限责任公司 | A kind of learning method for image characteristics extraction |
| CN111931590A (en) * | 2020-07-15 | 2020-11-13 | 杭州电子科技大学 | Balanced face feature extraction method of self-adaptive four-corner star local graph structure |
| CN111931590B (en) * | 2020-07-15 | 2023-09-29 | 杭州电子科技大学 | Balanced face feature extraction method of self-adaptive four-corner star-shaped partial graph structure |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107358147A (en) | Face recognition features' extraction algorithm based on local circulation graph structure | |
| Wang et al. | Tire defect detection using fully convolutional network | |
| CN113822209B (en) | Hyperspectral image recognition method and device, electronic equipment and readable storage medium | |
| CN109190752A (en) | The image, semantic dividing method of global characteristics and local feature based on deep learning | |
| CN110443239A (en) | The recognition methods of character image and its device | |
| CN114241522B (en) | Site operation safety wearing identification method, system, equipment and storage medium | |
| Arkin et al. | A survey of object detection based on CNN and transformer | |
| CN105426884A (en) | Fast document type recognition method based on full-sized feature extraction | |
| CN110751195B (en) | Fine-grained image classification method based on improved YOLOv3 | |
| CN108520215B (en) | Single-sample face recognition method based on multi-scale joint feature encoder | |
| CN110188217A (en) | Image duplicate checking method, apparatus, equipment and computer-readable storage media | |
| CN105160305B (en) | A kind of multi-modal Feature fusion of finger | |
| CN104463091B (en) | A kind of facial image recognition method based on image LGBP feature subvectors | |
| CN104794726B (en) | A kind of underwater picture Parallel segmentation method and device | |
| CN116051957A (en) | Personal protection item detection network based on attention mechanism and multi-scale fusion | |
| CN109086801A (en) | A kind of image classification method based on improvement LBP feature extraction | |
| CN105354547A (en) | Pedestrian detection method in combination of texture and color features | |
| CN108154130A (en) | A kind of detection method of target image, device and storage medium, robot | |
| Chen et al. | MSGC-YOLO: An Improved Lightweight Traffic Sign Detection Model under Snow Conditions | |
| CN105205476A (en) | Face recognition hardware framework based on LBP characteristics | |
| Luo et al. | Multi-scale face detection based on convolutional neural network | |
| Ouyang et al. | An anchor-free detector with channel-based prior and bottom-enhancement for underwater object detection | |
| Yuan et al. | RPN-FCN based Rust detection on power equipment | |
| CN104573663B (en) | A kind of English scene character recognition method based on distinctive stroke storehouse | |
| CN105894475A (en) | International phonetic symbol image character refining method |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171117 |