CN107169420A - A kind of eyeball center positioning method based on image gradient - Google Patents

A kind of eyeball center positioning method based on image gradient Download PDF

Info

Publication number
CN107169420A
CN107169420A CN201710256950.4A CN201710256950A CN107169420A CN 107169420 A CN107169420 A CN 107169420A CN 201710256950 A CN201710256950 A CN 201710256950A CN 107169420 A CN107169420 A CN 107169420A
Authority
CN
China
Prior art keywords
value
image
gradient
row
method based
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.)
Granted
Application number
CN201710256950.4A
Other languages
Chinese (zh)
Other versions
CN107169420B (en
Inventor
冯文廷
陈志�
岳文静
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Post and Telecommunication University
Nanjing University of Posts and Telecommunications
Original Assignee
Nanjing Post and Telecommunication University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing Post and Telecommunication University filed Critical Nanjing Post and Telecommunication University
Priority to CN201710256950.4A priority Critical patent/CN107169420B/en
Publication of CN107169420A publication Critical patent/CN107169420A/en
Application granted granted Critical
Publication of CN107169420B publication Critical patent/CN107169420B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06V40/18—Eye characteristics, e.g. of the iris
    • G06V40/193—Preprocessing; Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/66—Analysis of geometric attributes of image moments or centre of gravity
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • 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
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • 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
    • G06V10/56—Extraction of image or video features relating to colour
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30041—Eye; Retina; Ophthalmic

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Geometry (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Ophthalmology & Optometry (AREA)
  • Human Computer Interaction (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)
  • Eye Examination Apparatus (AREA)

Abstract

The invention discloses a kind of eyeball center positioning method based on image gradient.This method proposes a kind of Grad by calculating human eye area image, constructs and solves using Grad and pixel value as the optimal solution of the object function of variable, so as to position the coordinate at eyeball center in image.The characteristics of the inventive method has simple be easily achieved, has relatively low time complexity and stronger robustness by support of geometry teaching.

Description

A kind of eyeball center positioning method based on image gradient
Technical field
The present invention relates to technical field of image processing, particularly a kind of eyeball center positioning method based on image gradient.
Background technology
The complete image of one width, is made up of red, green, blue three passages, and the one of which of gray proces Method is exactly the triple channel value of image to be averaged so as to obtain the gray-scale map corresponding to original image, and each pixel of gray-scale map has 256 gray scales, are just avoided that the distortion of bands visible, and play simplified effect to the further analysis of image.
Gaussian filtering is a kind of linear smoothing filtering, it is adaptable to eliminate Gaussian noise, is widely used in subtracting for image procossing Make an uproar process, gaussian filtering is exactly that average process is weighted to entire image, the value of each pixel by itself and Other pixel values in field are obtained after being weighted averagely.There is eyeball centralized positioning calculating time complexity at this stage higher Shortcoming.
The content of the invention
The technical problems to be solved by the invention overcome the deficiencies in the prior art and provide a kind of based on image gradient Eyeball center positioning method, effectively and accurately to be positioned to eyeball center.
The present invention uses following technical scheme to solve above-mentioned technical problem：
According to a kind of eyeball center positioning method based on image gradient proposed by the present invention, comprise the following steps：
Step 1, to size be n rows n arrange eye image G carry out gray proces obtain gray level image G '；
Step 2, to image G ' carry out Gaussian Blur operations；It is specific as follows：
Step 21, ambiguity in definition radius r, r ＜ n and r are odd number, use Gaussian function to calculate fuzzy square of the size for r*r Battle array P；
Step 22, calculate P in element and sp, to each element divided by s in Pp, obtain normalized matrix P '；
Step 23, to define G ' (i, j) be the i-th row in G ', the element of jth row, from theRow starts continuously to take r rows, fromRow start continuously to take r to arrange, and constitute a submatrix, by each element of the submatrix and P ' in same a line same row Element multiplication and result is summed, obtained value is designated as H (i, j), i span fromArriveJ value Scope fromArriveRow；
Step 24, each H (i, the j) value obtained in step 23 is placed in the ith row and jth column of matrix H, obtain by Image array H after gaussian filtering；
Step 25, calculate H in element and sh, to each element divided by s in Hh, obtain normalized matrix H '；
Step 3, the gradients of calculating H ' (i, j) both horizontally and vertically, and the ladder of each pixel is calculated accordingly Direction value is spent, H ' (i, j) is the element of matrix H ' middle ith row and jth column；Comprise the following steps that：
Step 31, horizontal direction gradient dx (i, j)=H ' (i, the j+1)-H ' (i, j) for calculating pixel point coordinates (i, j)；
Step 32, vertical gradient dy (i, j)=H ' (i+1, j)-H ' (i, j) for calculating pixel point coordinates (i, j)；
Step 33, the gradient direction value for calculating each pixel
Step 4, objective functionThreshold valueTo every in H ' Individual element value is judged, if H ' (i, j) ＜ ε, then calculates the corresponding functional value f of H ' (i, j) according to current function；If H ' (i, J) >=ε, then the corresponding functional value f of H ' (i, j) are 0；Solving makes the coordinate (i of the element corresponding to target function value maximumc,jc), Should (ic,jc) be eyeball central point；Wherein, w1,w2For customized parameter；ic,jcThe horizontal seat of eyeball central point is corresponded to respectively Mark and ordinate.
It is described as a kind of further prioritization scheme of eyeball center positioning method based on image gradient of the present invention Step 1 is specific as follows：To each pixel in G, the average value of the Red Green Blue corresponding to it is calculated, the average value is For the gray value of the element.
It is used as a kind of further prioritization scheme of eyeball center positioning method based on image gradient of the present invention, input Eye image pass through gaussian filtering process.
It is described as a kind of further prioritization scheme of eyeball center positioning method based on image gradient of the present invention Blur radius r=3 in step 2, α=1.5.
It is described as a kind of further prioritization scheme of eyeball center positioning method based on image gradient of the present invention In step 4, regulation parameter w1=1.5, w2=1.
The present invention uses above technical scheme compared with prior art, with following technique effect：
1) present invention proposes a kind of eyeball center positioning method based on image gradient, and its complete procedure includes pair Image carries out gray proces and gaussian filtering process, calculates the gradient direction value of each pixel in image, constructs with gradient side To the object function of value and image pixel value for independent variable, by solving the maximum of object function and then obtaining corresponding eye Ball center's coordinate；
2) present invention can effectively reduce the influence of noise of image by carrying out gaussian filtering process to image, improve positioning The degree of accuracy；
3) present invention carries out geometrical analysis using the feature of image gradient to image, and the problem of looking for eyeball central point converts The problem of to solve object function maximum, the time complexity that effectively reduction is calculated improves the degree of accuracy of Ins location.
Brief description of the drawings
Fig. 1 is the eyeball center positioning method flow based on image gradient.
Fig. 2 is eye image matrix example.
Fig. 3 is the gaussian filtering template that blur radius is 3.
Embodiment
Technical scheme is described in further detail below in conjunction with the accompanying drawings：
Image gradient equivalent to the derivation to two-dimensional discrete function, gradient information can reflect image pixel change direction and Rate of change, present invention utilizes this feature of image gradient, and combines the priori of eye image, show that eyeball center must Positioned at the place that Grad is smaller and color is deeper, obtained by the two factors of simultaneous construction object function and solution in eyeball The coordinate value of the heart, is yielded good result.
The corresponding flow chart of eyeball center positioning method based on image gradient shown in 1, human eye shown in accompanying drawing 2 with reference to the accompanying drawings Image array example, blur radius shown in accompanying drawing 3 is 3 gaussian filtering template, and the specific embodiment of the invention is：
1) the eye image matrix G arranged as shown in Figure 2 for 4 rows 4, the figure for the image obtain after gray processing processing As matrix is as shown below:
2) accompanying drawing 3 show the gaussian filtering template that blur radius is 3, and center point coordinate is (0,0).With top left co-ordinate Exemplified by (- 1,1), the coordinate and central point (0,0) abscissa direction apart from x=1, ordinate direction apart from y=1, its is right The Gaussian function numerical value answered isSimilarly calculate the Gaussian function of other points Value, calculating obtains fuzzy matrix P and is：
21) in calculating matrix P element and sp, to each element divided by s in Pp, obtain normalized matrix P ' as follows：
23) it is the i-th row in G ', the element of jth row, from the to define G ' (i, j)Row starts continuously to take r rows, fromRow start continuously to take r to arrange, and constitute a submatrix, by the element in the element of submatrix and P ' in same a line same row It is multiplied and result is summed, obtained value is designated as H (i, j).By taking G ' (2,2) as an example, the pixel is carried out using matrix P ' high This filtering process:
H (2,2)=123*0.094+144*0.118+256*0.094+85*0.118+206*0.148+136*0.1 18+ 120*0.094+76*0.118+43*0.094=277.9.Matrix H is obtained after similarly carrying out gaussian filtering to G ' other elements:
24) calculate element in H and sh=2287.67, to each element divided by s in Hh, obtain normalized matrix H′：
3) gradients of H ' (i, j) both horizontally and vertically are calculated, and calculate the gradient side of each pixel accordingly To value.Comprise the following steps that：
31) horizontal direction gradient dx (i, j)=H ' (i, j+1)-H ' (i, j) of pixel point coordinates (i, j) is calculated.Calculate It is as follows to horizontal direction gradient matrix：
32) vertical gradient dy (i, j)=H ' (i+1, j)-H ' (i, j) of pixel point coordinates (i, j) are calculated.Calculate It is as follows to vertical gradient matrix：
33) the gradient direction value of pixel is calculatedCalculating obtains ladder Spend direction value and gradient direction is as follows：
4) objective functionThreshold valueTo in H Each element value is judged, if H ' (i, j) ＜ ε, then the corresponding functional value f of calculating elements H ' (i, j)；If H ' (i, j) >=ε, Then the corresponding functional value f of H ' (i, j) are 0.It is as follows that calculating obtains the corresponding functional value of the middle each elements of H '：
41) from matrix f, the corresponding coordinate of function maxima is (2,2), so coordinate points (2,2) are eyeball Center point coordinate.
The present invention is not only effectively reduced the influence of picture noise, and takes full advantage of the feature of image gradient, will look for The problem of the problem of eyeball central point is converted into solution object function maximum, the time complexity that effectively reduction is calculated is improved The degree of accuracy of Ins location.

Claims (5)

1. a kind of eyeball center positioning method based on image gradient, it is characterised in that comprise the following steps：
Step 1, to size be n rows n arrange eye image G carry out gray proces obtain gray level image G '；
Step 2, to image G ' carry out Gaussian Blur operations；It is specific as follows：
Step 21, ambiguity in definition radius r, r ＜ n and r are odd number, use Gaussian function to calculate fuzzy matrix P of the size for r*r；
Step 22, calculate P in element and sp, to each element divided by s in Pp, obtain normalized matrix P '；
Step 23, to define G ' (i, j) be the i-th row in G ', the element of jth row, from theRow starts continuously to take r rows, fromRow start continuously to take r to arrange, and constitute a submatrix, by each element of the submatrix and P ' in same a line same row Element multiplication and result is summed, obtained value is designated as H (i, j), i span fromArriveJ value Scope fromArriveRow；
Step 24, each H (i, the j) value obtained in step 23 is placed in the ith row and jth column of matrix H, obtained by Gauss Filtered image array H；
Step 25, calculate H in element and sh, to each element divided by s in Hh, obtain normalized matrix H '；
Step 3, the gradients of calculating H ' (i, j) both horizontally and vertically, and the gradient side of each pixel is calculated accordingly To value, H ' (i, j) is the element of matrix H ' middle ith row and jth column；Comprise the following steps that：
Step 31, horizontal direction gradient dx (i, j)=H ' (i, the j+1)-H ' (i, j) for calculating pixel point coordinates (i, j)；
Step 32, vertical gradient dy (i, j)=H ' (i+1, j)-H ' (i, j) for calculating pixel point coordinates (i, j)；
Step 33, the gradient direction value for calculating each pixel
Step 4, objective functionThreshold valueTo each member in H ' Plain value is judged, if H ' (i, j) ＜ ε, then calculates the corresponding functional value f of H ' (i, j) according to current function；If H ' (i, j) >= ε, then the corresponding functional value f of H ' (i, j) are 0；Solving makes the coordinate (i of the element corresponding to target function value maximumc,jc), should (ic,jc) be eyeball central point；Wherein, w1,w2For customized parameter；ic,jcThe abscissa of eyeball central point is corresponded to respectively And ordinate.
2. a kind of eyeball center positioning method based on image gradient according to claim 1, it is characterised in that the step Rapid 1 is specific as follows：To each pixel in G, the average value of the Red Green Blue corresponding to it is calculated, the average value is The gray value of the element.
3. a kind of eyeball center positioning method based on image gradient according to claim 1, it is characterised in that input Eye image passes through gaussian filtering process.
4. a kind of eyeball center positioning method based on image gradient according to claim 1, it is characterised in that the step Blur radius r=3, α=1.5 in rapid 2.
5. a kind of eyeball center positioning method based on image gradient according to claim 1, it is characterised in that the step In rapid 4, regulation parameter w1=1.5, w2=1.
CN201710256950.4A 2017-04-19 2017-04-19 Eyeball center positioning method based on image gradient Active CN107169420B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710256950.4A CN107169420B (en) 2017-04-19 2017-04-19 Eyeball center positioning method based on image gradient

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710256950.4A CN107169420B (en) 2017-04-19 2017-04-19 Eyeball center positioning method based on image gradient

Publications (2)

Publication Number Publication Date
CN107169420A true CN107169420A (en) 2017-09-15
CN107169420B CN107169420B (en) 2020-04-21

Family

ID=59813201

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710256950.4A Active CN107169420B (en) 2017-04-19 2017-04-19 Eyeball center positioning method based on image gradient

Country Status (1)

Country Link
CN (1) CN107169420B (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105550657A (en) * 2015-12-23 2016-05-04 北京化工大学 Key point based improved SIFT human face feature extraction method

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105550657A (en) * 2015-12-23 2016-05-04 北京化工大学 Key point based improved SIFT human face feature extraction method

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
FABIAN TIMM等: "Accurate Eye Centre Localisation by Means of Gradients", 《VISAPP 2011 - PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON COMPUTER VISION》 *
ROBERTO VALENTI等: "Accurate eye center location and tracking using isophote curvature", 《2008 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 *
刘军: "复杂环境下驾驶员眼睛定位及眼睛状态识别算法研究", 《中国优秀硕士论文全文数据库信息科技辑》 *
李亚利等: "基于改进型抛物线Hough变换的眼睛特征提取", 《清华大学学报（自然科学版）》 *
王静等: "基于梯度和模板二次匹配的人眼定位", 《计算机技术与发展》 *

Also Published As

Publication number Publication date
CN107169420B (en) 2020-04-21

Similar Documents

Publication Publication Date Title
CN103914699B (en) A kind of method of the image enhaucament of the automatic lip gloss based on color space
CN109145927A (en) The target identification method and device of a kind of pair of strain image
CN106339707B (en) A kind of gauge pointer image-recognizing method based on symmetric characteristics
CN109584251A (en) A kind of tongue body image partition method based on single goal region segmentation
CN107154030A (en) Image processing method and device, electronic equipment and storage medium
CN103702015B (en) Exposure control method for human face image acquisition system under near-infrared condition
CN110570480B (en) Sketch drawing method of drawing robot, electronic equipment and storage medium
CN108682017A (en) Super-pixel method for detecting image edge based on Node2Vec algorithms
CN105893944B (en) Depth information static gesture dividing method
CN106803257A (en) The dividing method of scab in a kind of crop disease leaf image
CN106127740B (en) One kind being based on the associated profile testing method of the more orientation of sensory field of visual pathway
CN103903256B (en) Depth estimation method based on relative height-depth clue
CN101510255A (en) Method for identifying and positioning human face, apparatus and video processing chip
CN109767459A (en) Novel ocular base map method for registering
CN103440633A (en) Digital image automatic speckle-removing method
CN105761292A (en) Image rendering method based on color shift and correction
CN106204453A (en) The interpolation method of a kind of image and device
CN116721099B (en) Image segmentation method of liver CT image based on clustering
CN102567969B (en) Color image edge detection method
CN106683043A (en) Parallel image stitching method and device for multi-channel optical detection system
CN108596992A (en) A kind of quickly real-time lip gloss cosmetic method
CN107169420A (en) A kind of eyeball center positioning method based on image gradient
CN117197064A (en) Automatic non-contact eye red degree analysis method
CN108256475B (en) Bill image inversion detection method
CN116823648A (en) Image enhancement method and system based on Zero-DCE network

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
GR01 Patent grant
GR01 Patent grant