CN105224285A - Eyes open and-shut mode pick-up unit and method - Google Patents
Eyes open and-shut mode pick-up unit and method Download PDFInfo
- Publication number
- CN105224285A CN105224285A CN201410227002.4A CN201410227002A CN105224285A CN 105224285 A CN105224285 A CN 105224285A CN 201410227002 A CN201410227002 A CN 201410227002A CN 105224285 A CN105224285 A CN 105224285A
- Authority
- CN
- China
- Prior art keywords
- eyes
- ocular
- state
- eye
- shut mode
- 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
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a kind of eyes open and-shut mode pick-up unit and method.Described eyes open and-shut mode pick-up unit comprises: ocular detecting unit, detects the ocular comprising eyes from target image; Eyes cutting unit, splits the ocular detected, to extract eye image; Based on the grey scale change in eye image, status determining unit, determines that eyes are in the state of opening or closed-eye state.
Description
Technical field
The application relates to a kind of eyes open and-shut mode pick-up unit and method.More particularly, relate to a kind of by determining to detect pick-up unit and the method that eyes open and-shut modes detects nictation.
Background technology
Eyes are primary window that the mankind obtain external world information, and the information of its acceptance accounts for brain accepts full detail 80% ~ 90%.The research of human eye sight tracking technique is of far-reaching significance in fields such as cognitive science, psychology, man-machine interactions.
Blink detection technology refers to by detecting the image of some characteristics in human eye movement's process, estimates the state that current eyes are in eye closing and still open and the degree that eyes are opened.Along with the universal development of Video processing and Video Supervision Technique, blink detection becomes indispensable pith in eye image analytic process.Such as, in sight line estimating system, need to carry out blink detection to eye image in real time, filter out eye closing and micro-image of opening, avoiding sight line to estimate there is larger noise and fluctuation in focus; In driving monitoring, blink detection and frequency of wink statistical study are carried out, to fatigue driving situation and alarm to driver; In interactive process, by trigger event of having blinked, realize the interactive experience of simpler and easy hommization.
The blink detection method of current view-based access control model is mainly divided into:
1) template matching method: the features such as the original image information of eye image that selection is normally opened or the edge of extraction are as template, and other frame of video are mated and calculate correlativity.The eye image correlativity opening state is high, and the correlativity of closed-eye state is lower, therefore by arranging threshold value, eye state is divided into eyes-open state, closed-eye state and intermediateness.But the method shortcoming is to need manually to choose the eye image normally opened.
2) eyelid detection method: the characteristic utilizing lower eyelid area image uniform gray level by secondary light source or under natural light, and lower eyelid on the feature direct-detections such as eyelid moves up and down in nictation process, according to the Distance Judgment eyes eye closing degree at upper lower eyelid edge.In the method, the utilization of secondary light source can produce restriction to use scenes, on the other hand, because the segmentation contrast locating of eyelid is more difficult, usually needs in sequence frame image, add up the region moved up and down and finally determines eyelid, cause computation complexity to increase.
Therefore, a kind of device and method can determining eyes open and-shut mode is accurately and quickly needed.
Summary of the invention
Other aspect and/or advantage will partly be set forth in the following description, also have part will be clearly from description, or are learned by enforcement of the present invention.
According to an aspect of the present invention, provide a kind of eyes open and-shut mode pick-up unit, comprising: ocular detecting unit, from target image, detect the ocular comprising eyes; Eyes cutting unit, splits the ocular detected, to extract eye image; Based on the grey scale change in eye image, status determining unit, determines that eyes are in the state of opening or closed-eye state.
Described ocular can be the rectangular area comprising eyes.
When determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, the angle rolling target image that ocular detecting unit can tilt based on eyes, and can again detect from postrotational target image the ocular comprising eyes.
When determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, ocular detecting unit can correct described rectangle ocular, makes the eyes in the rectangle ocular of overcorrect be in horizontality relative to the rectangle ocular after correction.
Machine learning algorithm can be used to train described ocular detecting unit, make, when ocular detecting unit processes target image, ocular to be detected from target image.
By using Gauss complexion model, described eyes cutting unit determines whether each pixel in ocular belongs to skin area, and can by skin area in ocular with the image zooming-out of exterior domain for eye image.
Can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from least one position four of an ocular corner location.
Can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from the top of described ocular or bottom.
By the distribution of the gray-scale value detecting all pixels in eye image, described status determining unit determines that eyes are in the state of opening or closed-eye state.
Described status determining unit can calculate the mean square deviation of the gray-scale value of all pixels of eye image, when the mean square deviation calculated is more than or equal to predetermined threshold value, described status determining unit determination eyes are in the state of opening, when the mean square deviation calculated is less than described predetermined threshold value, described status determining unit determination eyes are in closed-eye state.
Described status determining unit can carry out transverse axis projection to eye image, the column average gray scale of the often row pixel on the direction that in calculating eye image, edge is vertical with eyes to form column average grey scale curve, and is in the state of opening or closed-eye state based on the intensity of variation determination eyes of the column average grey scale curve formed.
When the intensity of variation of described column average grey scale curve exceedes predetermined threshold, described status determining unit can determine that eyes are in the state of opening, when the intensity of variation of described column average grey scale curve is lower than described predetermined threshold, described status determining unit can determine that eyes are in closed-eye state.
The average gray row vector of column average grey scale curve and edge detection template function also can be carried out convolution by described status determining unit, and are in the state of opening or closed-eye state based on the curve determination eyes that convolution obtains.When the amplitude of the crest in described convolution graph and trough exceedes predetermined amplitude, described status determining unit can determine that eyes are in the state of opening; When the amplitude of the crest in described convolution graph and trough is lower than described predetermined amplitude, described status determining unit can determine that eyes are in closed-eye state.
According to another aspect of the present invention, provide a kind of eyes open and-shut mode detection method, comprising: from target image, detect the ocular comprising eyes; The ocular detected is split, to extract eye image; Determine that eyes are in the state of opening or closed-eye state based on the grey scale change in eye image.
Described ocular can be the rectangular area comprising eyes.
The step detecting ocular also can comprise: when determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, based on the angle rolling target image that eyes tilt, and again detect the ocular comprising eyes from postrotational target image.
The step detecting ocular also can comprise: when determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, described rectangle ocular is corrected, makes the eyes in the rectangle ocular of overcorrect be in horizontality relative to described rectangle ocular.
The step detecting ocular can use machine learning algorithm to detect ocular from target image.
By using Gauss complexion model, the step extracting eye image determines whether each pixel in ocular belongs to skin area, and by skin area in ocular with the image zooming-out of exterior domain for eye image.
Can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from least one position four of a described ocular corner location.
Can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from the top of described ocular or bottom.
Determine that eyes are in the state of opening or the step of closed-eye state can comprise: determine that eyes are in the state of opening or closed-eye state by the distribution of the gray-scale value detecting all pixels in eye image.
Determine that eyes are in the state of opening or the step of closed-eye state can comprise: the mean square deviation calculating the gray-scale value of all pixels of eye image, when the mean square deviation calculated is more than or equal to predetermined threshold value, determine that eyes are in the state of opening, when the mean square deviation calculated is less than described predetermined threshold value, determine that eyes are in closed-eye state.
Determine that eyes are in the state of opening or the step of closed-eye state can comprise: transverse axis projection is carried out to eye image, the column average gray scale of the often row pixel on the direction that in calculating eye image, edge is vertical with eyes to form column average grey scale curve, and is in the state of opening or closed-eye state based on the intensity of variation determination eyes of the column average grey scale curve formed.
When the intensity of variation of described column average grey scale curve exceedes predetermined threshold, can determine that eyes are in the state of opening, when the intensity of variation of described column average grey scale curve is lower than described predetermined threshold, can determine that eyes are in closed-eye state.
Determine that the average gray row vector of column average grey scale curve and edge detection template function also can be carried out convolution by the step of eye state, and be in the state of opening or closed-eye state based on the curve determination eyes that convolution obtains.When the amplitude of the crest in described convolution graph and trough exceedes predetermined amplitude, can determine that eyes are in the state of opening; When the amplitude of the crest in described convolution graph and trough is lower than described predetermined amplitude, can determine that eyes are in closed-eye state.
Beneficial effect
Based on eyes open and-shut mode pick-up unit of the present invention and method, do not need manually to select eye opening template to determine the open and-shut mode of eyes, it also avoid in prior art the difficulty of eyelid segmentation when determining eyes open and-shut mode, therefore, it is possible to determine eyes open and-shut mode accurately and quickly, be particularly suitable for single-frame images and detect
Accompanying drawing explanation
By the detailed description of carrying out exemplary embodiment below in conjunction with accompanying drawing, above and other object of the present invention and feature will become apparent, wherein:
Fig. 1 is the block diagram of the configuration of the eyes open and-shut mode pick-up unit illustrated according to exemplary embodiment of the present invention;
Fig. 2 illustrates that the eyes cutting unit according to exemplary embodiment of the present invention extracts the schematic diagram of the design of eye image;
Fig. 3 is the schematic diagram of the adjustment ocular of the ocular adjustment unit illustrated according to exemplary embodiment of the present invention;
Fig. 4 is the schematic representation of the status determining unit determination eyes open and-shut mode illustrated according to exemplary embodiment of the present invention;
Fig. 5 illustrates that use detects the simulation result figure of the opening/closing state of eyes according to an embodiment of the invention.
Fig. 6 is the schematic representation of the opening/closing state of status determining unit determination eyes illustrated according to another exemplary embodiment of the present invention.
Fig. 7 illustrates the process flow diagram according to exemplary embodiment eyes open and-shut mode detection method of the present invention.
In the accompanying drawings, identical drawing reference numeral will be understood to mean identical parts.
Embodiment
There is provided the description carried out referring to accompanying drawing to help complete understanding by the exemplary embodiment of the present invention of claim and equivalents thereof.Described description comprises various specific detail to help to understand, but these details are considered to be only exemplary.Therefore, those of ordinary skill in the art will recognize: without departing from the scope and spirit of the present invention, can make various changes and modifications the embodiments described herein.In addition, for clarity and conciseness, the description of known function and structure can be omitted.
Fig. 1 is the block diagram of the configuration of the eyes open and-shut mode pick-up unit 100 illustrated according to exemplary embodiment of the present invention.
With reference to Fig. 1, the eyes open and-shut mode pick-up unit 100 according to exemplary embodiment of the present invention can comprise: ocular detecting unit 110, eyes cutting unit 120 and status determining unit 130.
According to exemplary embodiment of the present invention, ocular detecting unit 110 can be used for from target image, detect the ocular comprising eyes.The ocular detected can be the rectangular area comprising eyes, and this rectangular area is usually slightly large than the region shared by whole eyes and comprise whole eyes (rectangular area such as, shown in Fig. 2).
Only exemplarily, described ocular detecting unit 110 can use machine learning algorithm to detect ocular from target image.Specifically, in an exemplary embodiment of the present invention, the machine learning methods such as Adboost or randomforest can be used to train described ocular detecting unit 110, make, when ocular detecting unit 110 pairs of target images process, ocular to be detected from target image.When training eye region detection unit 110, a large amount of positive samples and negative sample can be used to train it.Positive sample refers to the image of eyes or the image comprising eyes based on eyes.By mark in the video pictures image of the different attitude of different people different scale and the image intercepting the ocular of people to obtain multiple positive sample.Negative sample refers to the image not comprising eyes, obtains by the region random acquisition beyond eyes from image.Generally speaking, when training ocular detecting unit 110, use negative sample quantity far more than positive sample quantity (such as, the quantity of negative sample is 2 to 3 times of the quantity of positive sample), be conducive to the ocular that ocular detecting unit 110 identifies eyes place exactly in the picture.
To those skilled in the art, easy understand and know the expectation object how utilizing machine learning algorithm to detect such as eyes from image, therefore for the sake of simplicity, will no longer be explained in more detail at this.
In addition, although be also to be understood that in superincumbent description and use machine learning to detect the ocular comprising eyes from image, the present invention is not limited thereto, additive method known in the art also can be used to determine the ocular in image, the such as method such as template matches, rim detection.
After ocular detecting unit 110 detects ocular, can be split the ocular detected by eyes cutting unit 120, to extract eye image.
Specifically, what ocular detecting unit 110 detected comprises the ocular of eyes normally to the rectangular area of eyes coarse positioning, therefore, in described ocular, except the image of eyes, also can comprise the image of such as skin.In order to (namely the open and-shut mode of eyes can be judged exactly, eyes are opened or are closed), usual needs go part in addition to the eye (such as, removing the skin area of around eyes) from the image of ocular, thus obtain the eye image only comprising eyes.
What comprise the skin of eyes and around eyes below with ocular is assumed to be example to explain that eyes cutting unit 120 extracts the process of eye image.
Fig. 2 illustrates that the eyes cutting unit 120 according to exemplary embodiment of the present invention extracts the schematic diagram of the design of eye image.
With reference to Fig. 2, in order to extract eye image from ocular, first need to identify skin area from ocular.In an exemplary embodiment of the present invention, by using Gauss complexion model, eyes cutting unit 120 determines whether each pixel in ocular belongs to skin area, and by skin area in ocular with the image zooming-out of exterior domain for eye image.Consider the shape of eyes and the shape of ocular, indicated by the label 210 in Fig. 2 (a), four corner location places of ocular are generally skin, therefore can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from least one position four of a described ocular corner location.Selectively, also can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from the top of ocular and/or bottom.
For the dermatological specimens chosen, train Gauss's complexion model by formula (1) below:
Wherein, x is the color vector of each pixel of dermatological specimens, and μ is the mean value of the color vector of all dermatological specimens, and Σ is the covariance matrix of the color vector of all dermatological specimens, matrix (x-μ)
tit is the transposition of matrix (x-μ).P (x|skin) represents the probable value that this pixel is skin.
Above-mentioned formula is applied to all pixels of ocular and Threshold segmentation is carried out to the P (x|skin) of each pixel, remove skin area, the eye image (as Suo Shi Fig. 2 (b)) of mainly sclera, iris and pupil can be extracted.
In the above description for eyes cutting unit 120, should be appreciated that, when the eyes comprised in the rectangle ocular that ocular detecting unit 110 detects are in non-horizontal directions relative to described rectangle ocular, because the uncertainty of eyes position in ocular may cause also comprising eye image in the dermatological specimens selected, the possibility of result makes eyes segmentation result not accurate enough, therefore, for making last eye state judged result more accurate, the eyes in ocular should be made to be in horizontal direction as far as possible.
In an embodiment of the present invention, when determining that the eyes in rectangle ocular are in non-standard state relative to described rectangle ocular (ocular as indicated by 310 in Fig. 3), the angle rolling target image that ocular detecting unit 110 can tilt based on eyes, and the ocular comprising eyes is again detected from postrotational target image, or, ocular detecting unit 110 can correct (such as to described ocular, prune), the eyes in the rectangle ocular of overcorrect are made to be in horizontality (as shown in the rectangular area that 310 ' in Fig. 3 indicates) relative to the rectangle ocular after correction.
In an embodiment of the present invention, various method can be used to determine whether eyes are in horizontality relative to rectangle ocular.Such as, as shown in Figure 3, when ocular detecting unit 110 detects two corresponding with two, the left and right eyes of same people respectively rectangle oculars 310, the center by connecting these two regions obtains the baseline 320 connecting left and right eyes.The degree of tilt of described baseline 320 is similar to the rotational angle of face plane, therefore can according to the approximate inclination angle between the degree of tilt determination eyes of baseline 320 and horizontal direction.Then, ocular detecting unit 110 can detect again based on this inclination angle image rotating, or correct accordingly ocular, thus the eyes that ocular detecting unit 110 is finally outputted in the ocular image of eyes cutting unit 120 are levels relative to described ocular.But, this is only determine whether eyes are in an example of horizontality relative to described rectangle ocular, the present invention is not limited thereto, such as, also first determine the rotational angle of face plane by other modes such as active shape models, thus determine approximate inclination angle between eyes to horizontal direction and carry out corresponding adjustment.
After obtaining eye image, can determine that eyes are in the state of opening or closed-eye state by status determining unit 130 according to an embodiment of the invention based on the grey scale change in eye image.
The exemplary operation of status determining unit 130 is described in detail below with reference to Fig. 4.Fig. 4 is the schematic representation that the status determining unit 130 illustrating according to exemplary embodiment of the present invention determines the opening/closing state of eyes.
First status determining unit 130 can carry out transverse axis projection to eye image according to an embodiment of the invention, the column average gray scale of the often row pixel on the direction that in calculating eye image, edge is vertical with eyes is to form column average grey scale curve, and the intensity of variation then based on the column average grey scale curve formed determines that eyes are in the state of opening or closed-eye state.
Specifically, as shown in Figure 4, a left side of Fig. 4 (a) illustrates the eyes being in the state of opening, and the right side illustrates the eyes being in closed-eye state.(namely the eyes that a left side of Fig. 4 (b) illustrates opening state carry out the result of transverse axis projection, by calculating the curve formed along the column average gray scale of the often row pixel on the direction vertical with eyes in eye image), the right side of Fig. 4 (b) illustrates the result of the eyes of closed-eye state being carried out to transverse axis projection.Eyes are opened to the image of state, due to the strong contrast of iris and sclera, have violent change (shown in figure as left in Fig. 4 (b), significantly rising and falling appears in column average grey scale curve) in its juncture area column average gray scale; And for the image of eyes closed-eye state, because iris, pupil, sclera are covered by eyelid, the inequality of eyelashes only can bring the faint change of average gray (shown in figure as right in Fig. 4 (b), obvious fluctuating does not appear in column average grey scale curve).Therefore, by the change of the column average grey scale curve of acquisition after the transverse axis projection of eye image, can judge that eyes are in the state of opening or closed-eye state.
In an embodiment of the present invention, when the intensity of variation of described column average grey scale curve exceedes predetermined threshold, status determining unit 130 can determine that eyes are in the state of opening, when the intensity of variation of described column average grey scale curve is lower than described predetermined threshold (described predetermined threshold can be determined in advance according to many experiments or user experience), status determining unit 130 can determine that eyes are in closed-eye state.Described predetermined threshold can pre-set according to many experiments or rule of thumb.
Preferably, in order to make it possible to the state judging eyes more accurately, after the column average grey scale curve obtaining eye image, described status determining unit 130 also can by the average gray row vector of column average grey scale curve and edge detection template function (such as, Sigmoid stencil function) carry out convolution, amplify the sharp regions of column average grey scale curve, to judge that eyes are in the state of opening or closed-eye state more accurately based on the curve after convolution.
A left side of Fig. 4 (c) illustrates the result of the column average grey scale curve under eyes-open state being carried out to convolution, and the right side of Fig. 4 (c) illustrates the result of the column average grey scale curve under closed-eye state being carried out to convolution.Can find out from Fig. 4 (c), convolution is carried out to the column average grey scale curve under eyes-open state and has occurred obvious crest and trough in the curve obtained, the column average grey scale curve under closed-eye state is carried out then not having obvious crest and trough in the curve of convolution acquisition.In this case, status determining unit 130 can judge eye state according to the amplitude of the Wave crest and wave trough in the curve of convolution acquisition and/or quantity.Such as, when the amplitude of the crest in the curve obtained by described convolution and trough exceedes predetermined amplitude (described predetermined amplitude can be determined in advance according to many experiments or user experience), described status determining unit 130 can determine that eyes are in the state of opening, when the amplitude of the crest in described curve and trough is lower than described predetermined amplitude, described status determining unit 130 can determine that eyes are in closed-eye state.Selectively, when the crest in the curve that described convolution obtains and trough total quantity a predetermined level is exceeded (such as, two) time, described status determining unit 130 can determine that eyes are in the state of opening, when the total quantity of the crest in described curve and trough is lower than described predetermined quantity, described status determining unit 130 can determine that eyes are in closed-eye state.Or selectively, when the amplitude in the curve obtained by described convolution exceedes the quantity a predetermined level is exceeded of predetermined amplitude (described predetermined amplitude can be determined in advance according to many experiments or user experience) crest and/or trough, status determining unit 130 can determine that eyes are in the state of opening, otherwise status determining unit 130 can determine that eyes are in closing device.
Fig. 5 illustrates to use simulation result figure according to an embodiment of the invention.Fig. 5 illustrates the testing result (that is, detecting the open and-shut mode of the eyes in the multiple image obtained continuously for single people) of single multiple image.
In Figure 5, transverse axis represents frame index number, and the longitudinal axis represents the projection response of X-direction (that is, horizontal direction), that is, use the result of horizontal projection (that is, the column average grey scale curve) convolution of sigmoid function and eye image.Although shown in the left figure of Fig. 4 (c) with sigmoid convolution of functions after there are two extreme points (namely, correspond respectively to the white of the eye in column average grey scale curve to the minimum value of the trough of the stepped portion of pupil and to correspond in column average grey scale curve pupil to the maximal value of the crest of the stepped portion of the white of the eye), but this only illustrates for convenience of explanation, can only get one of these two extreme points Detection results is described.In Figure 5, use and correspond to pupil in column average grey scale curve detects eyes open and-shut mode to the maximal value of the crest of the stepped portion of the white of the eye
As shown in Figure 5, " O " represents the frame by the eye closing detected of the eyes open and-shut mode pick-up unit 100 of exemplary embodiment according to the present invention, and " X " represents the frame of actual true eye closing, and "+" represents close to closed situation.What arrange in test in Figure 5 is " 3000 " for the threshold value determining eyes open and-shut mode.But should be appreciated that, the setting of the size of this threshold value is determined by many experiments repeatedly and contrast, and is not limited to certain concrete numerical value.
As can be seen from Figure 5, detect with actual, the frame of eye closing occurs that the frame of closing one's eyes is basically identical by the eyes open and-shut mode pick-up unit 100 of exemplary embodiment according to the present invention, there is good detection accuracy.
Although in the description of above composition graphs 4, use the mode of transverse axis projection to detect the opening/closing state of eyes, the present invention is not limited thereto.According to another exemplary embodiment of the present invention, after eyes cutting unit 120 is partitioned into eye image, by the distribution of the gray-scale value detecting all pixels in eye image, status determining unit 130 determines that eyes are in the state of opening or closed-eye state.The method determining the opening/closing state of eyes according to the status determining unit 130 of another exemplary embodiment of the present invention is described below with reference to Fig. 6.In figure 6, transverse axis represents gray-scale value, and the longitudinal axis represents the quantity of pixel.
Specifically, when eyes are opened, gray-scale value in eye image characterizes primarily of the gray-scale value of the gray-scale value of the pixel at pupil position and the pixel at white of the eye position, and during eyes closed, the gray-scale value in eye image characterizes primarily of the gray-scale value of the pixel at eyelashes position.Therefore, status determining unit 130 can determine the opening/closing state of eyes based on the distribution of gray-scale value.As shown in Figure 6, when eyes are opened, in the color histogram of the distribution of the grey scale pixel value of statistics eye image, gray-scale value is mainly distributed in black region (corresponding to pupil position, gray-scale value is close to 0) and white portion (corresponding to white of the eye position, gray-scale value is close to 255) two regions, grey scale change is obvious; And when the eyes are occluded, in described color histogram, the gray-scale value of pixel is mainly distributed in black region (corresponding to eyelashes position, gray-scale value is close to 0), grey scale change is not obvious, therefore, and can easily by the opening/closing state of such grey value profile determination eyes.Selectively, status determining unit 130 also can calculate the mean square deviation of the gray-scale value of all pixels of eye image.As shown in Figure 6, when eyes are opened, in eye image, the gray-scale value of the pixel at the white of the eye and pupil position differs greatly usually, and in this case, the mean square deviation of the gray-scale value of all pixels in eye image is larger; And during eyes closed, a line of only ciliation formation in eye image, the gray-scale value of the pixel therefore in eye image is close, and the mean square deviation of the gray-scale value of all pixels in eye image is less, even close to 0.Therefore, when the mean square deviation calculated is more than or equal to predetermined threshold value (determining this threshold value in advance by experiment or experience), described status determining unit 130 can determine that eyes are in the state of opening, when the mean square deviation calculated is less than described predetermined threshold value, described status determining unit 130 can determine that eyes are in closed-eye state.
Fig. 7 illustrates the process flow diagram according to exemplary embodiment eyes open and-shut mode detection method of the present invention.
With reference to Fig. 7, in step 710, can detect by the ocular detecting unit 110 in eyes open and-shut mode pick-up unit 100 according to an embodiment of the invention the ocular comprising eyes from target image.In an embodiment of the present invention, described ocular can be the rectangular area comprising eyes, and machine learning algorithm can be used to train described ocular detecting unit 110, make, when ocular detecting unit 110 pairs of target images process, ocular to be detected from target image.
Preferably, in step 710, when the eyes in the rectangle ocular that described ocular detecting unit 110 detects are in non-standard state relative to described rectangle ocular, the angle rolling target image that ocular detecting unit 110 can tilt based on eyes, and the ocular comprising eyes is again detected from postrotational target image, or selectively, described ocular detecting unit 110 can correct described rectangle ocular, the eyes in the rectangle ocular of overcorrect are made to be in horizontality relative to the rectangle ocular after correction.Like this, ocular detecting unit 110 finally outputs to the eyes in the ocular image of eyes cutting unit 120 can be level relative to described ocular.
Describe the operation of ocular detecting unit 110 in the preceding article in detail, therefore for the sake of simplicity, no longer will carry out unnecessary description at this.
In step 730, the ocular detected can be split, to extract eye image by the eyes cutting unit 120 in eyes open and-shut mode pick-up unit 100 according to an embodiment of the invention.
Only exemplarily, in step 730, described eyes cutting unit 120 can use Gauss's complexion model to determine whether each pixel in ocular belongs to skin area, and by skin area in ocular with the image zooming-out of exterior domain for eye image.Can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from least one position four of a described ocular corner location.Selectively, also can the region of pre-sizing be selected originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from the top of described ocular or bottom.
Describe the operation of eyes cutting unit 120 in the preceding article with reference to Fig. 2 in detail, therefore for the sake of simplicity, no longer will carry out unnecessary description at this.
In step 750, can determine that eyes are in the state of opening or closed-eye state by the status determining unit 130 in eyes open and-shut mode pick-up unit 100 according to an embodiment of the invention based on the grey scale change in eye image.
Such as, status determining unit 130 can carry out transverse axis projection to the eye image extracted, the column average gray scale of the often row pixel on the direction that in calculating eye image, edge is vertical with eyes to form column average grey scale curve, and is in the state of opening or closed-eye state based on the intensity of variation determination eyes of the column average grey scale curve formed.
When the intensity of variation of described column average grey scale curve exceedes predetermined threshold, described status determining unit 130 can determine that eyes are in the state of opening, when the intensity of variation of described column average grey scale curve is lower than described predetermined threshold, described status determining unit 130 can determine that eyes are in closed-eye state.
Preferably, the average gray row vector of column average grey scale curve and edge detection template function also can be carried out convolution by described status determining unit 130, to amplify the sharp regions of column average grey scale curve, and be in the state of opening or closed-eye state based on the curve determination eyes that convolution obtains.In this case, when the amplitude of the crest in the curve obtained by described convolution and trough exceedes predetermined amplitude, described status determining unit 130 can determine that eyes are in the state of opening, when the amplitude of the crest in described curve and trough is lower than described predetermined amplitude, described status determining unit 130 can determine that eyes are in closed-eye state.Selectively, when the crest in the curve that described convolution obtains and trough total quantity a predetermined level is exceeded (such as, two) time, described status determining unit 130 can determine that eyes are in the state of opening, when the total quantity of the crest in described curve and trough is lower than described predetermined quantity, described status determining unit 130 can determine that eyes are in closed-eye state.Or selectively, when the amplitude in the curve obtained by described convolution exceedes the quantity a predetermined level is exceeded of predetermined amplitude (described predetermined amplitude can be determined in advance according to many experiments or user experience) crest and/or trough, status determining unit 130 can determine that eyes are in the state of opening, otherwise status determining unit 130 can determine that eyes are in closing device.
As another example, by the distribution of the gray-scale value detecting all pixels in eye image, status determining unit 130 also determines that eyes are in the state of opening or closed-eye state.Specifically, if the gray-scale value of the pixel in eye image be mainly distributed in gray-scale value close to 0 black region and gray-scale value close to 255 white portion two regions, then status determining unit 130 can determine that eyes are in the state of opening.If the gray-scale value of the pixel in eye image be mainly distributed in gray-scale value close to 0 black region, then status determining unit 130 can determine that eyes are in closure state.Selectively, described status determining unit 130 also can calculate the mean square deviation of the gray-scale value of all pixels of eye image, when the mean square deviation calculated is more than or equal to predetermined threshold value, described status determining unit determination eyes are in the state of opening, when the mean square deviation calculated is less than described predetermined threshold value, described status determining unit determination eyes are in closed-eye state.
Based on eyes open and-shut mode pick-up unit of the present invention and method, do not need manually to select eye opening template to determine the open and-shut mode of eyes, it also avoid in prior art the difficulty of eyelid segmentation when determining eyes open and-shut mode, therefore, it is possible to determine eyes open and-shut mode accurately and quickly, be particularly suitable for single-frame images and detect
In addition, although in superincumbent description, eyes open and-shut mode pick-up unit of the present invention and method are applied to determines that eyes are in the state of opening or closed-eye state, but application of the present invention is not limited thereto, such as, when status determining unit 130 of the present invention determines eye state, can need eye state to be divided in more detail the state of opening by arranging different threshold values, micro-ly to open the more multimode such as state, closed-eye state according to user.In addition, when using eyes open and-shut mode pick-up unit of the present invention and method to detect continuously the multiple successive image frames in video, by adding up the eye state in every frame, detected object can be determined (such as, people) number of winks, in a short time every or frequency of wink, and further to operate based on these results.
Exemplary embodiment of the present can be embodied as the computer-readable code on computer readable recording medium storing program for performing.Computer readable recording medium storing program for performing is the arbitrary data memory storage that can store the data that can be read by computer system thereafter.The example of computer readable recording medium storing program for performing comprises: ROM (read-only memory) (ROM), random access memory (RAM), CD-ROM, tape, floppy disk, optical data storage devices and carrier wave (such as transmitting through the data of wired or wireless transmission path by internet).Computer readable recording medium storing program for performing also can be distributed in the computer system of interconnection network, thus computer-readable code stores in a distributed manner and performs.In addition, complete function program of the present invention, code and code segment can easily be explained within the scope of the present invention by the ordinary programmers in field related to the present invention.
Although specifically show with reference to its exemplary embodiment and describe the present invention, but it should be appreciated by those skilled in the art, when not departing from the spirit and scope of the present invention that claim limits, the various changes in form and details can be carried out to it.
Claims (26)
1. an eyes open and-shut mode pick-up unit, comprising:
Ocular detecting unit, detects the ocular comprising eyes from target image;
Eyes cutting unit, splits the ocular detected, to extract eye image;
Based on the grey scale change in eye image, status determining unit, determines that eyes are in the state of opening or closed-eye state.
2. eyes open and-shut mode pick-up unit as claimed in claim 1, it is characterized in that, described ocular is the rectangular area comprising eyes.
3. eyes open and-shut mode pick-up unit as claimed in claim 2, it is characterized in that, when determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, the angle rolling target image that ocular detecting unit tilts based on eyes, and the ocular comprising eyes is again detected from postrotational target image.
4. eyes open and-shut mode pick-up unit as claimed in claim 2, it is characterized in that, when determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, ocular detecting unit corrects described rectangle ocular, makes the eyes in the rectangle ocular of overcorrect be in horizontality relative to the rectangle ocular after correction.
5. eyes open and-shut mode pick-up unit as claimed in claim 1, it is characterized in that, use machine learning algorithm to train described ocular detecting unit, make, when ocular detecting unit processes target image, from target image, to detect ocular.
6. eyes open and-shut mode pick-up unit as claimed in claim 1, it is characterized in that, by using Gauss complexion model, described eyes cutting unit determines whether each pixel in ocular belongs to skin area, and by skin area in ocular with the image zooming-out of exterior domain for eye image.
7. eyes open and-shut mode pick-up unit as claimed in claim 6, is characterized in that, select the region of pre-sizing originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from least one position four corner location of ocular.
8. eyes open and-shut mode pick-up unit as claimed in claim 6, is characterized in that, selects the region of pre-sizing originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from the top of described ocular or bottom.
9. eyes open and-shut mode pick-up unit as claimed in claim 1, is characterized in that, by the distribution of the gray-scale value detecting all pixels in eye image, described status determining unit determines that eyes are in the state of opening or closed-eye state.
10. eyes open and-shut mode pick-up unit as claimed in claim 9, is characterized in that, described status determining unit calculates the mean square deviation of the gray-scale value of all pixels of eye image,
Wherein, when the mean square deviation calculated is more than or equal to predetermined threshold value, described status determining unit determination eyes are in the state of opening,
When the mean square deviation calculated is less than described predetermined threshold value, described status determining unit determination eyes are in closed-eye state.
11. eyes open and-shut mode pick-up units as claimed in claim 1, it is characterized in that, described status determining unit carries out transverse axis projection to eye image, the column average gray scale of the often row pixel on the direction that in calculating eye image, edge is vertical with eyes to form column average grey scale curve, and is in the state of opening or closed-eye state based on the intensity of variation determination eyes of the column average grey scale curve formed.
12. eyes open and-shut mode pick-up units as claimed in claim 11, it is characterized in that, when the intensity of variation of described column average grey scale curve exceedes predetermined threshold, described status determining unit determination eyes are in the state of opening, when the intensity of variation of described column average grey scale curve is lower than described predetermined threshold, described status determining unit determination eyes are in closed-eye state.
13. eyes open and-shut mode pick-up units as claimed in claim 11, it is characterized in that, the average gray row vector of column average grey scale curve and edge detection template function are also carried out convolution by described status determining unit, and be in the state of opening or closed-eye state based on the curve determination eyes that convolution obtains
Wherein, when the amplitude of the crest in described convolution graph and trough exceedes predetermined amplitude, described status determining unit determination eyes are in the state of opening, when the amplitude of the crest in described convolution graph and trough is lower than described predetermined amplitude, described status determining unit determination eyes are in closed-eye state.
14. 1 kinds of eyes open and-shut mode detection methods, comprising:
The ocular comprising eyes is detected from target image;
The ocular detected is split, to extract eye image;
Determine that eyes are in the state of opening or closed-eye state based on the grey scale change in eye image.
15. eyes open and-shut mode detection methods as claimed in claim 14, it is characterized in that, described ocular is the rectangular area comprising eyes.
16. eyes open and-shut mode detection methods as claimed in claim 15, it is characterized in that, the step detecting ocular also comprises: when determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, based on the angle rolling target image that eyes tilt, and again detect the ocular comprising eyes from postrotational target image.
17. eyes open and-shut mode detection methods as claimed in claim 15, it is characterized in that, the step detecting ocular also comprises: when determining that the eyes in described rectangle ocular are in non-standard state relative to described rectangle ocular, described rectangle ocular is corrected, makes the eyes in the rectangle ocular of overcorrect be in horizontality relative to described rectangle ocular.
18. eyes open and-shut mode detection methods as claimed in claim 14, is characterized in that, the step detecting ocular uses machine learning algorithm to detect ocular from target image.
19. eyes open and-shut mode detection methods as claimed in claim 14, it is characterized in that, by using Gauss complexion model, the step extracting eye image determines whether each pixel in ocular belongs to skin area, and by skin area in ocular with the image zooming-out of exterior domain for eye image.
20. eyes open and-shut mode detection methods as claimed in claim 19, it is characterized in that, select the region of pre-sizing originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from least one position four corner location of described ocular.
21. eyes open and-shut mode detection methods as claimed in claim 19, is characterized in that, select the region of pre-sizing originally to train Gauss's complexion model as the skin-like of Gauss's complexion model from the top of described ocular or bottom.
22. eyes open and-shut mode detection methods as claimed in claim 14, it is characterized in that, determine that eyes are in the state of opening or the step of closed-eye state comprises: determine that eyes are in the state of opening or closed-eye state by the distribution of the gray-scale value detecting all pixels in eye image.
23. eyes open and-shut mode detection methods as claimed in claim 22, is characterized in that, determine that eyes are in the state of opening or the step of closed-eye state comprises: the mean square deviation calculating the gray-scale value of all pixels of eye image,
Wherein, when the mean square deviation calculated is more than or equal to predetermined threshold value, determine that eyes are in the state of opening,
When the mean square deviation calculated is less than described predetermined threshold value, determine that eyes are in closed-eye state.
24. eyes open and-shut mode detection methods as claimed in claim 14, it is characterized in that, determine that eyes are in the state of opening or the step of closed-eye state comprises: transverse axis projection is carried out to eye image, the column average gray scale of the often row pixel on the direction that in calculating eye image, edge is vertical with eyes to form column average grey scale curve, and is in the state of opening or closed-eye state based on the intensity of variation determination eyes of the column average grey scale curve formed.
25. eyes open and-shut mode detection methods as claimed in claim 24, it is characterized in that, when the intensity of variation of described column average grey scale curve exceedes predetermined threshold, determine that eyes are in the state of opening, when the intensity of variation of described column average grey scale curve is lower than described predetermined threshold, determine that eyes are in closed-eye state.
26. eyes open and-shut mode detection methods as claimed in claim 24, it is characterized in that, determine that the average gray row vector of column average grey scale curve and edge detection template function are also carried out convolution by the step of eye state, and be in the state of opening or closed-eye state based on the curve determination eyes that convolution obtains
Wherein, when the amplitude of the crest in described convolution graph and trough exceedes predetermined amplitude, determine that eyes are in the state of opening, when the amplitude of the crest in described convolution graph and trough is lower than described predetermined amplitude, determine that eyes are in closed-eye state.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410227002.4A CN105224285A (en) | 2014-05-27 | 2014-05-27 | Eyes open and-shut mode pick-up unit and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410227002.4A CN105224285A (en) | 2014-05-27 | 2014-05-27 | Eyes open and-shut mode pick-up unit and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105224285A true CN105224285A (en) | 2016-01-06 |
Family
ID=54993288
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410227002.4A Pending CN105224285A (en) | 2014-05-27 | 2014-05-27 | Eyes open and-shut mode pick-up unit and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105224285A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292251A (en) * | 2017-06-09 | 2017-10-24 | 湖北天业云商网络科技有限公司 | A kind of Driver Fatigue Detection and system based on human eye state |
| CN107358151A (en) * | 2017-06-02 | 2017-11-17 | 广州视源电子科技股份有限公司 | A kind of eye motion detection method and device and vivo identification method and system |
| CN107610499A (en) * | 2016-07-11 | 2018-01-19 | 富士通株式会社 | Detection method, detection means and the electronic equipment of parking stall state |
| CN107943527A (en) * | 2017-11-30 | 2018-04-20 | 西安科锐盛创新科技有限公司 | The method and its system of electronic equipment is automatically closed in sleep |
| CN108090450A (en) * | 2017-12-20 | 2018-05-29 | 深圳和而泰数据资源与云技术有限公司 | Face identification method and device |
| CN108629293A (en) * | 2018-04-16 | 2018-10-09 | 西安交通大学 | A kind of adaptive near-infrared iris image acquiring method with feedback mechanism |
| CN108764121A (en) * | 2018-05-24 | 2018-11-06 | 释码融和（上海）信息科技有限公司 | Method, computing device and readable storage medium storing program for executing for detecting live subject |
| CN108921117A (en) * | 2018-07-11 | 2018-11-30 | 北京市商汤科技开发有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN109543629A (en) * | 2018-11-26 | 2019-03-29 | 平安科技（深圳）有限公司 | A kind of blink recognition methods, device, equipment and readable storage medium storing program for executing |
| CN109740512A (en) * | 2018-12-29 | 2019-05-10 | 山东财经大学 | A kind of method for recognizing human eye state for fatigue driving judgement |
| CN109847168A (en) * | 2019-01-22 | 2019-06-07 | 中国科学院苏州生物医学工程技术研究所 | Wearable fatigue detecting and interfering system |
| CN109963068A (en) * | 2017-12-25 | 2019-07-02 | 深圳市优必选科技有限公司 | A kind of image processing method, system and terminal device |
| CN110352033A (en) * | 2017-02-27 | 2019-10-18 | 托比股份公司 | Eyes degree of opening is determined with eye tracks device |
| CN111091058A (en) * | 2019-11-18 | 2020-05-01 | 京东方科技集团股份有限公司 | Eye state detection method, device, equipment and storage medium |
| CN111263065A (en) * | 2020-02-18 | 2020-06-09 | 珠海市它物云科技有限公司 | Eyeball detection prompting method, computer device and computer readable storage medium |
| EP3809310A4 (en) * | 2018-07-16 | 2021-09-01 | Honor Device Co., Ltd. | Method and electronic device for detecting open and closed states of eyes |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101127076A (en) * | 2007-09-27 | 2008-02-20 | 上海交通大学 | Human eye state detection method based on cascade classification and hough circle transform |
| CN101923645A (en) * | 2009-06-09 | 2010-12-22 | 黑龙江大学 | Iris splitting method suitable for low-quality iris image in complex application context |
| CN102831399A (en) * | 2012-07-30 | 2012-12-19 | 华为技术有限公司 | Method and device for determining eye state |
-
2014
- 2014-05-27 CN CN201410227002.4A patent/CN105224285A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101127076A (en) * | 2007-09-27 | 2008-02-20 | 上海交通大学 | Human eye state detection method based on cascade classification and hough circle transform |
| CN101923645A (en) * | 2009-06-09 | 2010-12-22 | 黑龙江大学 | Iris splitting method suitable for low-quality iris image in complex application context |
| CN102831399A (en) * | 2012-07-30 | 2012-12-19 | 华为技术有限公司 | Method and device for determining eye state |
Non-Patent Citations (1)
| Title |
|---|
| 李莹莹: "基于计算机图像处理技术的人眼变化识别系统", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107610499A (en) * | 2016-07-11 | 2018-01-19 | 富士通株式会社 | Detection method, detection means and the electronic equipment of parking stall state |
| CN110352033A (en) * | 2017-02-27 | 2019-10-18 | 托比股份公司 | Eyes degree of opening is determined with eye tracks device |
| CN110352033B (en) * | 2017-02-27 | 2023-04-28 | 托比股份公司 | Determining eye opening with an eye tracking device |
| CN107358151A (en) * | 2017-06-02 | 2017-11-17 | 广州视源电子科技股份有限公司 | A kind of eye motion detection method and device and vivo identification method and system |
| CN107292251A (en) * | 2017-06-09 | 2017-10-24 | 湖北天业云商网络科技有限公司 | A kind of Driver Fatigue Detection and system based on human eye state |
| CN107292251B (en) * | 2017-06-09 | 2020-08-28 | 湖北天业云商网络科技有限公司 | Driver fatigue detection method and system based on human eye state |
| CN107943527A (en) * | 2017-11-30 | 2018-04-20 | 西安科锐盛创新科技有限公司 | The method and its system of electronic equipment is automatically closed in sleep |
| CN108090450A (en) * | 2017-12-20 | 2018-05-29 | 深圳和而泰数据资源与云技术有限公司 | Face identification method and device |
| CN108090450B (en) * | 2017-12-20 | 2020-11-13 | 深圳和而泰数据资源与云技术有限公司 | Face recognition method and device |
| CN109963068A (en) * | 2017-12-25 | 2019-07-02 | 深圳市优必选科技有限公司 | A kind of image processing method, system and terminal device |
| CN108629293A (en) * | 2018-04-16 | 2018-10-09 | 西安交通大学 | A kind of adaptive near-infrared iris image acquiring method with feedback mechanism |
| CN108629293B (en) * | 2018-04-16 | 2020-10-27 | 西安交通大学 | Self-adaptive near-infrared iris image acquisition method with feedback mechanism |
| CN108764121A (en) * | 2018-05-24 | 2018-11-06 | 释码融和（上海）信息科技有限公司 | Method, computing device and readable storage medium storing program for executing for detecting live subject |
| CN108764121B (en) * | 2018-05-24 | 2021-03-02 | 释码融和（上海）信息科技有限公司 | Method for detecting living object, computing device and readable storage medium |
| CN108921117A (en) * | 2018-07-11 | 2018-11-30 | 北京市商汤科技开发有限公司 | Image processing method and device, electronic equipment and storage medium |
| EP3809310A4 (en) * | 2018-07-16 | 2021-09-01 | Honor Device Co., Ltd. | Method and electronic device for detecting open and closed states of eyes |
| US11430264B2 (en) | 2018-07-16 | 2022-08-30 | Honor Device Co., Ltd. | Eye open or closed state detection method and electronic device |
| CN109543629A (en) * | 2018-11-26 | 2019-03-29 | 平安科技（深圳）有限公司 | A kind of blink recognition methods, device, equipment and readable storage medium storing program for executing |
| CN109740512A (en) * | 2018-12-29 | 2019-05-10 | 山东财经大学 | A kind of method for recognizing human eye state for fatigue driving judgement |
| CN109847168A (en) * | 2019-01-22 | 2019-06-07 | 中国科学院苏州生物医学工程技术研究所 | Wearable fatigue detecting and interfering system |
| CN109847168B (en) * | 2019-01-22 | 2022-03-08 | 中国科学院苏州生物医学工程技术研究所 | Wearable fatigue detection and intervention system |
| CN111091058A (en) * | 2019-11-18 | 2020-05-01 | 京东方科技集团股份有限公司 | Eye state detection method, device, equipment and storage medium |
| CN111091058B (en) * | 2019-11-18 | 2024-05-17 | 京东方科技集团股份有限公司 | Eye state detection method, device, equipment and storage medium |
| CN111263065A (en) * | 2020-02-18 | 2020-06-09 | 珠海市它物云科技有限公司 | Eyeball detection prompting method, computer device and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105224285A (en) | Eyes open and-shut mode pick-up unit and method | |
| CN106803067B (en) | Method and device for evaluating quality of face image | |
| EP3355104B1 (en) | Method and device and computer program for determining a representation of a spectacle glass rim | |
| US9733703B2 (en) | System and method for on-axis eye gaze tracking | |
| Drutarovsky et al. | Eye blink detection using variance of motion vectors | |
| CN109343700B (en) | Eye movement control calibration data acquisition method and device | |
| CN106228293A (en) | teaching evaluation method and system | |
| WO2021077382A1 (en) | Method and apparatus for determining learning state, and intelligent robot | |
| CN102567734B (en) | Specific value based retina thin blood vessel segmentation method | |
| CN107895157B (en) | Method for accurately positioning iris center of low-resolution image | |
| CN105373767A (en) | Eye fatigue detection method for smart phones | |
| KR102442220B1 (en) | Living-body detection method and apparatus for face, electronic device ad computer readable medium | |
| CN109165630A (en) | A kind of fatigue monitoring method based on two-dimentional eye recognition | |
| CN108742656A (en) | Fatigue state detection method based on face feature point location | |
| KR20130043366A (en) | Gaze tracking apparatus, display apparatus and method therof | |
| CN103778406A (en) | Object detection method and device | |
| CN114708658A (en) | Online learning concentration degree identification method | |
| CN101642376A (en) | Device and method for detecting fatigue | |
| Naveed et al. | Eye tracking system with blink detection | |
| CN105279764B (en) | Eye image processing apparatus and method | |
| US20170309040A1 (en) | Method and device for positioning human eyes | |
| CN104408409A (en) | Pupil location method suitable for astigmatic lens environment | |
| CN103136511B (en) | Behavioral value method and device | |
| CN109034117A (en) | A kind of video extraction method and device | |
| Zhao et al. | Fast localization algorithm of eye centers based on improved hough transform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20210312 |
|
| AD01 | Patent right deemed abandoned |