CN110543843B - Human eye positioning and size calculating algorithm based on forward oblique projection and backward oblique projection - Google Patents
Human eye positioning and size calculating algorithm based on forward oblique projection and backward oblique projection Download PDFInfo
- Publication number
- CN110543843B CN110543843B CN201910782472.XA CN201910782472A CN110543843B CN 110543843 B CN110543843 B CN 110543843B CN 201910782472 A CN201910782472 A CN 201910782472A CN 110543843 B CN110543843 B CN 110543843B
- Authority
- CN
- China
- Prior art keywords
- projection
- eye
- oblique projection
- image
- backward
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 claims abstract description 12
- 238000012216 screening Methods 0.000 claims abstract description 6
- 210000001747 pupil Anatomy 0.000 claims description 28
- 210000003786 sclera Anatomy 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 10
- 210000003128 head Anatomy 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 230000000149 penetrating effect Effects 0.000 claims 1
- 238000005286 illumination Methods 0.000 description 13
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 210000003491 skin Anatomy 0.000 description 3
- 230000006870 function Effects 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
Abstract
The application provides a human eye positioning and size calculating algorithm based on forward oblique projection and backward oblique projection. The method comprises the steps of firstly carrying out forward oblique projection and backward oblique projection on a human eye image, generating a plurality of projection lines after carrying out forward oblique projection and backward oblique projection, providing a projection line screening method based on relative linear density, selecting one forward oblique projection line and one backward oblique projection line, selecting the position of eyes at the intersection point of the two projection lines, respectively selecting projection lines at two ends of the projection line with the minimum relative linear density in the forward oblique projection line and the backward oblique projection line as candidate distances of the eye sizes after determining the positions of the eyes, solving the average value of the two distances to determine the size (in units of pixels) of the eyes in the image, and acquiring the difference of the sizes of the eyes in the image due to the different distances between the user and a computer screen, so that the eye sizes calculated in the image are required to be converted from the actual eye sizes.
Description
Technical Field
The application provides a method for positioning human eyes and judging the sizes of the eyes based on forward oblique projection and backward oblique projection, which comprises the steps of firstly carrying out forward oblique projection and backward oblique projection on human eye images, generating a plurality of projection lines after carrying out forward oblique projection and backward oblique projection, selecting a forward oblique projection line and a backward oblique projection line based on relative linear density, selecting the position of the eyes at the intersection point of the two projection lines, respectively selecting projection lines at the two ends of the projection line with the minimum relative linear density in the forward oblique projection line and the backward oblique projection line as candidate distances of the sizes of the eyes after determining the positions of the eyes, solving the average value of the two distances to be determined as the sizes (the units are pixels) of the eyes in the images, and acquiring the difference between the sizes of the eyes in the images due to different distances between users and computer screens, so that the sizes of the eyes calculated in the images are required to be converted from the actual sizes of the eyes.
Background
The measurement of the size of the eyes and the positions of the eyes needs to be carried out by using professional equipment, such as an eye tracker, a common eye tracker cannot support the measurement of the size of the eyes and can only judge the positions of the eyes, so that the size of the eyes needs to be measured by using a scientific eye tracker, and the TobiiX2\X3 series and the Spectrum series support the measurement of the sizes of the eyes, but the cost is high, the use scene is limited, and the detection of the sizes and the positions of the eyes by using a common camera can effectively reduce the economic cost.
The Tobii series eye tracker adopts an infrared camera to detect the eye part, meanwhile, the distance between the user and the camera can be obtained, the Tobii eye tracker can record the image of the eye, and then the recorded eye image is used for calculating the model of the eye. The Tobii eye tracker uses the eye model to provide eye-to-sensor distance data, calculates the diameter of the pupil on the image by eye tracker firmware, and multiplies the calculated pupil diameter by a conversion factor to calculate the pupil size. The distance information of the eye distance sensor cannot be obtained by the common camera, so that the calculation of the size of the eye by using the monocular camera has a great challenge.
Because the shape of the eyes is round, the detection of the eyes by using the Hough algorithm and the corresponding improved algorithm becomes a common method for positioning the eyes and outputting the sizes of the eyes, but the Hough algorithm has high requirements on the complexity of illumination and background, lacks robustness, and meanwhile, the Hough algorithm has high calculation complexity, needs a lot of memory space for calculation and has high memory expense. Therefore, developing an algorithm with high robustness to illumination and background becomes a key for eye positioning and size detection.
Disclosure of Invention
The structure of the eye part is shown in fig. 1, the central part of the eye is a pupil area, the outside of the pupil area is an iris area, the outside of the iris area is a sclera area, the color of the pupil area is the darkest, and the color of the eye in the sclera area is the lightest, so when the eye area is projected, the effect of surrounding complexion can be reduced, and therefore, the effect of complexion can be reduced becomes the key of eye positioning, therefore, the application provides an eye positioning algorithm based on orthographic projection and retrooblique projection, the rough positioning is carried out on the eye area, and the calculation mode of the orthographic projection algorithm and the retroprojection algorithm is as follows:
formula (3) is to calculate forward oblique projection, formula (4) is to calculate backward oblique projection, data is image data of an eye part, x is row pixels of an image, y is column pixels of the image, row is the number of rows of the image, col is the number of columns of the image, the area of the eye is positioned through the forward oblique projection and the backward oblique projection, fig. 2 is a forward oblique projection result, fig. 3 is a backward oblique projection result, as can be found through fig. 2 and 3, the eye is of a symmetrical structure, the image pixel value of a pupil area is the lowest, so that when projection is performed, a peak point and a trough point of the image can appear, the projection line passing through the pupil area corresponds to the two peak points, the sum of the peak point and the trough point of an algorithm provided by the patent is certain to be an odd number, so that the method provided by the patent has interpretation on the positioning result of the eye image, the projection line can be seen through the forward oblique projection and the backward oblique projection result, the projection line roughly divides the eye area, and the projection line area effectively reduces the interference of the skin area; the result after coarse positioning is utilized to find that a plurality of projection lines appear no matter the projection lines are forward-oblique projection or backward-oblique projection, and the projection result of the projection lines has certain regularity, namely, two projection lines respectively penetrate through skin and iris areas, one projection line penetrates through skin, sclera and pupil area, and the pixel value of the pupil area is the lowest, therefore, the patent provides a projection line screening method based on relative linear density, which screens the projection lines, namely, the image forward (backward) oblique projection gray pixel value on unit length, and the relative linear density is calculated as follows:
wherein Z is the sum of the forward-oblique projection gray scale pixel values, F is the sum of the backward-oblique projection gray scale pixel values (Xend, yend) is the pixel position of the image end point, and (Xstart, ystart) is the pixel position of the image start point, namely the function of denominator is to calculate the Euclidean distance between the oblique projection start point and the end point. rld _Z is the forward projection relative linear density value and rld _F is the reverse projection relative linear density value. The results of screening the projection lines with relative line densities are shown in fig. 4 and 5. The position of the intersection of the forward projection line and the reverse projection line is the position of the pupil, and the result of positioning the pupil position is shown in fig. 7. By analyzing the projection line results, two projection lines closest to the projection line from two sides of the projection line are selected by utilizing the projection line screened by the relative line density, the distance between the two projection lines is the diameter of an eye, the distance between the peak point of the projection line and the trough point is the projection line moving from the edge of the sclera to the inside of the sclera, when the projection line moves to the pupil position, the projection line is the trough at the moment, then the projection line moves from the pupil area to the edge of the sclera, and the calculation mode of the eye size is as follows according to the waveform change from the trough to the crest of the projection curve result:
where (X0, Y0) is the start point of the projection line, A, B, C is the coefficient of the projection line ax+by+c=0, respectively, and R calculated at this time is the size of the eye region in the image. The method provided by the application has robustness to illumination change, because the eye area is small, the effect of illumination on the eye area is the same, the illumination intensity of an original image is assumed to be M, when the illumination is changed, the illumination intensity N which is the same as the increase or decrease of the original illumination intensity is equivalent to that of the illumination intensity N, namely, the illumination intensity change is changed from M to M+N, and the image is equivalent to that of linear transformation, so that the projection result is unchanged, and the method has good robustness to illumination change.
The distance between the user and the computer screen can change the size of the eye area of the acquired image, and in order to eliminate calculation errors caused by different distances between the head and the screen, the application provides a compensation method for calculating the size of eyes based on different distances between the head and the screen, wherein the different distances between the head and the computer screen are shown in fig. 10. The areas S of the eye area at different distances L from the screen are collected respectively, and fitting is performed on S and L, it can be found that the distance from the screen and the change of the eye area show linear changes, and the result is shown in fig. 11. The fitting equation at this time, i.e., s=al+b, is obtained, and therefore when the size of the eye is measured, it is known that the diameter size of the eye in the image is RP when the distance is L, and when the measurement result of the eye diameter in the image is RC, Δ=rc-RP is the change amount of the eye diameter, i.e., the change amount of the eye size is
Fig. 9 is a program interface for real-time detection of eye position and size using the algorithm of the present application.
Drawings
FIG. 1 is an original eye image;
FIG. 2 is a front oblique projection result;
FIG. 3 is a back oblique projection result;
FIG. 4 is a positive oblique projection line screening result;
FIG. 5 is a reverse bias projection line screening result;
FIG. 6 is a view of the intersection of forward and reverse oblique projection lines;
FIG. 7 is an eye positioning result;
FIG. 8 is a graph showing the eye size determination;
FIG. 9 is an eye positioning and size determination GUI;
FIG. 10 is an area of eyes at various distances from a screen
FIG. 11 is a graph showing the relationship between the difference of the eye distance and the screen and the change of the eye area
Detailed Description
The location of eyes area and the detection of eyes size can receive the influence of illumination, and the method that this patent provided can overcome illumination effectively and to the influence of eyes size and location, compares in other algorithms, and the method that this patent provided is consistent with the structure of eyes, has fine interpretation.
The eye consists of a sclera, an iris and a pupil from the outside to the inside, wherein the sclera is a white area, the iris is a brown area and the pupil is a black area. When the projection line passes through the iris region from the sclera region, the trough appears due to the dark iris color, and when the projection line passes through the pupil region from the iris region, the trough appears due to the dark pupil region color, and meanwhile, compared with the iris region, the color of the pupil region is darker, the trough value is smaller, and the projection result shows symmetry due to the symmetrical structure of the eye. As the projection line moves from the iris region to the pupil region and from the pupil region to the iris region, a global minimum occurs at this point because the pupil assumes a circular shape, and the projection line results in passing through the pupil center point.
The result corresponding to the global minimum point of the back-oblique projection also passes through the center point of the pupil, as in the front-oblique projection.
The position of the eye cannot be uniquely determined by a plurality of projection lines, since the pupil part is a black area, the pixel values on each projection line are summed,
argmin Z&&argmin F
the projection line with the smallest sum of pixel values is the candidate projection line which passes through the center point of the eye, but the sum of projection lines is normalized because the length of each projection line is not uniform, i.e.
argmin rld_Z&&ar gmin rld_F
The two projection lines include the iris region of the eye, and because the projection lines are parallel to each other and the iris is approximately circular, the size of the eye region is obtained by calculating the distance between the projection lines, using
The size of the eye is calculated.
Claims (1)
1. The human eye positioning and size calculating method based on the forward oblique projection and the backward oblique projection is characterized by comprising the following steps of:
coarse positioning is carried out on the eye area, and the calculation mode of the forward and backward oblique projection algorithm is as follows:
wherein Z is the sum of the forward-oblique projection gray scale pixel values, and F is the sum of the reverse-oblique projection gray scale pixel values; data is image data of an eye part, x is row pixels of the image, y is column pixels of the image, row is the number of rows of the image, col is the number of columns of the image, the region of the eye is positioned by forward oblique projection and backward oblique projection,
because the eyes are of symmetrical structures, the pixel values of the images in the pupil areas are the lowest, when the images are projected, the peaks and the trough points of the images appear, projection lines penetrating through the pupil areas are correspondingly arranged between the two peak points, and the eye areas are roughly divided by the projection lines as seen through the forward oblique projection and the backward oblique projection results;
screening the projection lines, and calculating relative linear density, namely the pixel value of the forward and backward oblique projection gray scale of the image on the unit length, wherein the relative linear density is shown in the following formula:
the formula (3) is used for calculating forward oblique projection, the formula (4) is used for calculating backward oblique projection, the (Xend, yeend) is used for calculating pixel positions of an image end point, the (Xstart, yestart) is used for calculating Euclidean distance between an oblique projection start point and an image end point, and the function of denominator is used for calculating the Euclidean distance between the start point and the end point of oblique projection; rld _Z is the relative linear density value of the forward oblique projection, rld _F is the relative linear density value of the backward oblique projection;
selecting two projection lines closest to the projection lines from two sides of the projection line by utilizing projection lines screened by relative linear density, wherein the distance between the two projection lines is the diameter of an eye, the distance between the peak point and the trough point of the projection line is that the projection line moves from the edge of a sclera to the inside of the sclera, when the projection line moves to a pupil position, the projection line is the trough at the moment, then the projection line moves from the pupil area to the edge of the sclera, and the calculation mode of the eye size is as follows according to the waveform change from the trough to the crest of a projection curve result:
wherein (X) 0 ，Y 0 ) As the starting point of the projection line, A, B, C is the coefficient of the projection line ax+by+c=0, respectively, and the calculated R is the size of the eye region in the image;
the distance between the user and the computer screen can enable the size of the eye area of the acquired image to be changed, so that calculation errors caused by different distances between the head and the screen can be eliminated, and the eye size calculation is compensated based on different distances between the head and the screen: the areas S of the eye areas at different distances L from the screen are respectively acquired, and the S and the L are simulatedThe fitting equation at this time, i.e., s=al+b, is obtained, and therefore, when the size of the eye is measured, it is known that the diameter size of the eye in the image is RP when the distance is L, and when the measurement result of the diameter of the eye in the image is RC, Δ=rc-RP is the change amount of the diameter of the eye, i.e., the change amount of the size of the eye is
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910782472.XA CN110543843B (en) | 2019-08-23 | 2019-08-23 | Human eye positioning and size calculating algorithm based on forward oblique projection and backward oblique projection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910782472.XA CN110543843B (en) | 2019-08-23 | 2019-08-23 | Human eye positioning and size calculating algorithm based on forward oblique projection and backward oblique projection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110543843A CN110543843A (en) | 2019-12-06 |
| CN110543843B true CN110543843B (en) | 2023-12-15 |
Family
ID=68711893
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910782472.XA Active CN110543843B (en) | 2019-08-23 | 2019-08-23 | Human eye positioning and size calculating algorithm based on forward oblique projection and backward oblique projection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110543843B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1475961A (en) * | 2003-07-14 | 2004-02-18 | 中国科学院计算技术研究所 | Human eye location method based on GaborEge model |
| CN106066696A (en) * | 2016-06-08 | 2016-11-02 | 华南理工大学 | The sight tracing compensated based on projection mapping correction and point of fixation under natural light |
| CN109034051A (en) * | 2018-07-24 | 2018-12-18 | 哈尔滨理工大学 | Human-eye positioning method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105320919A (en) * | 2014-07-28 | 2016-02-10 | 中兴通讯股份有限公司 | Human eye positioning method and apparatus |
-
2019
- 2019-08-23 CN CN201910782472.XA patent/CN110543843B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1475961A (en) * | 2003-07-14 | 2004-02-18 | 中国科学院计算技术研究所 | Human eye location method based on GaborEge model |
| CN106066696A (en) * | 2016-06-08 | 2016-11-02 | 华南理工大学 | The sight tracing compensated based on projection mapping correction and point of fixation under natural light |
| CN109034051A (en) * | 2018-07-24 | 2018-12-18 | 哈尔滨理工大学 | Human-eye positioning method |
Non-Patent Citations (3)
| Title |
|---|
| A model of the entrance pupil of the human eye;Geoffrey K. Aguirre;《Scientific Reports》;第1-10页 * |
| 基于边缘检测和灰度投影的人眼定位;周立宇等;《广西科学 》;第242-246页 * |
| 积分投影与连通域法结合的人眼定位;崔庆华等;《计算机与数字工程》;第949-953页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110543843A (en) | 2019-12-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11769265B2 (en) | Skin assessment using image fusion | |
| US9025826B2 (en) | Formation of a time-varying signal representative of at least variations in a value based on pixel values | |
| CN112106105B (en) | Method and system for generating three-dimensional image of object | |
| JP4473136B2 (en) | Acquisition of 3D images by active stereo technology using local unique patterns | |
| JP6793325B2 (en) | Skin diagnostic device and skin diagnostic method | |
| US6504942B1 (en) | Method of and apparatus for detecting a face-like region and observer tracking display | |
| US9900485B2 (en) | Image generation device | |
| AU2007324081B2 (en) | Focus assist system and method | |
| EP1958158B1 (en) | Method for detecting streaks in digital images | |
| JP2006285310A (en) | Evaluation method of canopy of forest, and its canopy evaluation program | |
| US8503734B2 (en) | Detecting image detail level | |
| US11010877B2 (en) | Apparatus, system and method for dynamic in-line spectrum compensation of an image | |
| JP2011077764A (en) | Multidimensional image processing device, multidimensional image photographing system, multidimensional image printed matter and multidimensional image processing method | |
| Curt et al. | Optimal digital color image correlation | |
| CN110543843B (en) | Human eye positioning and size calculating algorithm based on forward oblique projection and backward oblique projection | |
| JP7234086B2 (en) | Biometric information acquisition device and program | |
| CN110889820B (en) | Image quality assessment | |
| JP2002286422A (en) | Displacement sensor | |
| EP2693397B1 (en) | Method and apparatus for noise reduction in an imaging system | |
| KR101334029B1 (en) | Method for extracting muscle of head and neck and method of ＲＯＩ analysis | |
| JP2017191082A (en) | Bright-spot image acquisition apparatus and bright-spot image acquisition method | |
| Humblot-Renaux | Implementation of HDR for image acquisition on a finger vein scanner | |
| JP2017191083A (en) | Bright-spot image acquisition apparatus and bright-spot image acquisition method | |
| CN114707529A (en) | Image quality evaluation method and system in focusing process of linear array camera | |
| JP2020160681A (en) | Image processing device, image processing method and program |
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 |