CN104921697B - Method for quickly measuring longitudinal distances of sight of human eyes - Google Patents

Abstract

The invention relates to a method for measuring the longitudinal distance of the line of sight of human eyes. The specific steps include: 1. Build an eccentric camera measurement device, including a camera, infrared light source, infrared filter, telescope, face support, infrared cut-off filter, etc.; 2. Pupil boundary detection, use the run-length algorithm to obtain the pupil area, Then use equation fitting to obtain the boundary; 3. Calculation of the distribution of brightness and darkness in the pupil, the average value of the gray value of the entire pupil area is used as the threshold, and the ratio of the length of the longitudinal length of the pupil area greater than the threshold to the length of the entire longitudinal axis is calculated; 4. The measurement of the longitudinal distance , according to the calibration parameters before the measurement and the current distribution of pupil brightness and darkness, calculate the longitudinal distance of the line of sight. The method can obtain the longitudinal distance information in line-of-sight tracking, not only can realize real three-dimensional line-of-sight tracking, but also can be used to improve the precision of traditional two-dimensional line-of-sight tracking system. At the same time, the calculation complexity of this method is low, and the effect of fast measurement and real-time tracking can be achieved.

Description

A Quick Measuring Method of Longitudinal Distance of Human Eye's Line of Sight

technical field

The invention relates to the technical field of measuring the visual distance of the human eye, in particular to a method for quickly measuring the longitudinal distance of the human visual line.

Background technique

At present, the measurement of the visual distance of human eyes is mainly based on the binocular parallax method, and the accuracy of such measurement is very poor. When people's eyes are focused on the same point, because there is a distance (about 3-5cm) between the eyes, the line-of-sight vectors of the two eyes do not completely coincide. Therefore, the binocular parallax method is to measure the sight direction vector of each eye separately, and the intersection of the two vectors is the focus of human eye observation. However, the measurement error of the human eye's line of sight direction vector is relatively large, and the secondary error caused by using it to calculate the focus position of the human eye will be very large. Moreover, some studies have shown that the position of the intersection point of the line of sight direction vector is not necessarily the position of the human eye. The focus of observation, the focus of observation of the human eye will appear within a certain distance before and after the intersection of the line of sight direction vector. This will cause the traditional binocular parallax method to have relatively low measurement accuracy.

Contents of the invention

The purpose of the present invention is to propose a method for quickly measuring the longitudinal distance of the line of sight of human eyes. The pupil image of the human eye is captured by the camera, the pupil light and dark distribution features are extracted from the image, and the human eye line-of-sight distance is calculated. We have improved the optical path of the infrared eccentric photography method, and added appropriate optical aids (infrared filter, infrared cut-off filter, infrared light source, telescope system), so that the measurement process has less restrictions on people and less time spent.

The purpose of the present invention is achieved through the following technical solutions.

A method for measuring the longitudinal distance of the line of sight of the human eye, comprising the steps of:

(1) Build an eccentric camera measurement device on the optical bench, including a camera, an infrared light source, an infrared filter, a telescope, a face support, and an infrared cut-off filter; along the direction of the optical path, the infrared cut-off filter, the telescope and The camera is coaxial with the same height, and the face bracket is used to support the face of the testee; the infrared cut-off filter bends the measuring light path to the right side of the face bracket; the telescope is used to shorten the shooting light path; the camera is at the far right end of the light path, and the telescope A diaphragm and an infrared light source are set between the camera and the pupil image with clear boundaries can be obtained by using the infrared light source. It is also connected to the computer through the USB interface, and the image processing and calculation are performed on the computer using the MATLAB software platform;

(2) Pupil boundary detection, process and obtain pupil area with run-length algorithm, then obtain pupil boundary with equation fitting method;

(3) Calculation of the light and dark distribution of the pupil area. Since the diaphragm of the device blocks the lower part of the camera, the light and dark distribution of the captured pupil image changes vertically. The diameter in the vertical direction can be selected to calculate the light and dark distribution characteristics. The average value of the gray value of the entire pupil area is used as the threshold, and the ratio of the length greater than the threshold to the pupil diameter on the diameter in the vertical direction is calculated;

(4) The measurement of the longitudinal distance, the longitudinal distance of the line of sight is calculated by the calibration parameters before the measurement and the distribution of the current pupil brightness; the calibration parameters include the pupil radius R, the distance A between the camera and the pupil of the human eye, the pupil darkness Area to diameter ratio k.

Further optimally, the above method uses the improved infrared eccentric photography method to directly measure the longitudinal distance of the human eye's line of sight. The specific improvement method is to change the correction of the human eye's diopter to the human eye's focal length into the correction of the diopter to the object distance, and then obtain Calculation expression for the longitudinal distance of line of sight.

The invention adopts the telephoto system to shorten the shooting optical path, and weakens the influence of the center of the pupil and the infrared light source being out of axis. Use infrared light source and infrared filter to reduce the interference of ambient stray light.

To further optimize, the above method uses the method of image processing to calculate the distribution of brightness and darkness of the pupil image. Firstly, the run length algorithm is used to obtain the pupil area from the human eye image; Correct it into an ellipse; then perform a stretching transformation to stretch the pupil area into a circle, and finally take the vertical diameter of the circle to calculate the light and dark distribution characteristics of the pupil area.

Further optimally, the above-mentioned method uses an infrared cut-off filter to make the measurement optical path perpendicular to the optical path of the human eye's line of sight, and the measurement process will not affect the visual field of the human eye.

Further optimally, the specific calculation expression of the longitudinal distance of line of sight obtained by the calculation in step (4) is:

S is the longitudinal distance of the line of sight, R is the radius of the pupil, k is the ratio of the dark area of the pupil to the diameter, A is the distance between the camera and the pupil, and E is the distance from the light source to the optical axis. The ratio k of the dark area of the pupil to the diameter can be obtained from the pupil image through image processing Obtained by technology, the parameters E, A, and R are obtained through the calibration process; for each frame of human eye image, calculate the ratio k of the pupil dark area to the diameter and substitute it into the above expression to calculate the longitudinal distance of the current human eye line of sight .

Compared with the prior art, the beneficial effects of the present invention include the following points:

1) The use of infrared light source simplifies the image processing process and improves the measurement accuracy.

2) Use the lens group of the telescope system to shorten the optical path, weaken the influence of the misalignment between the center of the pupil and the infrared light source, and reduce the restrictions on people.

3) The time for each measurement is short, which can meet the requirements of real-time measurement and tracking measurement.

4) Using an infrared cut-off filter to reflect infrared light and transmit visible light, the measurement process will not affect the field of vision of the testee.

Description of drawings

Fig. 1a and Fig. 1b are structural schematic diagrams of the eccentric photogrammetry device in the example.

Figure 2 is an effect diagram of the distribution of brightness and darkness of the pupil image.

Figure 3 is a schematic diagram of the infrared off-center photography method.

Figure 4 is the actual human eye observation light path diagram.

Fig. 5 is a measurement diagram of pupil light and dark distribution.

Fig. 6 is the fitting curve diagram of the experimental parameters.

detailed description

The specific implementation of the present invention will be further described below in conjunction with the accompanying drawings and examples, but the implementation and protection of the present invention are not limited thereto.

The problems to be solved in this example can be mainly divided into the following points: From the captured eye image, accurately detect the boundary of the pupil. Determines the distribution of light and dark in the pupil area. According to the calibration process, empirical parameters are obtained. Actual measurement, use the formula of empirical parameters to calculate the line-of-sight distance.

1. Coaxial adjustment of the measuring system

In conjunction with Fig. 1, the entire device is fixed on the optical bench, and the face support 1 is used to support the face of the subject; the infrared cut-off filter 2 bends the measuring light path to the right side of the face support; the telescope 3 is used to In order to shorten the shooting light path; the camera 5 is at the far right end, and between the telescope 3 and the camera 5 is a diaphragm and an infrared light source, and the pupil image with a clear boundary can be obtained by using an infrared light source; an infrared filter 4 is arranged in front of the camera to reduce the interference of visible light in the environment influences. The adjustment process is as follows: First, adjust the coaxial contours of the camera 5, the telescope 3, the hollow tube and the infrared cut filter 2 on the optical bench. Then, the camera 5 shoots the eye image, which is transmitted to the computer through the USB interface. View the real-time eye image, adjust the height of the face bracket 1 (due to individual differences in human eyes), so that the eye area is included in the image.

2. Pupil Region Segmentation

This step is to segment the pupil area from the human eye image, and stretch the pupil area into a circular area as shown in Figure 2. Since the pupil is sensitive to infrared light, the pupil can be distinguished in image processing by letting the infrared light source shine on the eyeball. The infrared light source is installed in the center of the aperture, and the brightness can be adjusted. The image processing process is divided into the following steps:

(1) Use the histogram of the eye image to obtain a suitable threshold that can distinguish the pupil from other parts.

(2) Using the obtained threshold value, perform threshold value processing on the eye image to divide different regions.

(3) Use the run-length coding algorithm and the judgment of the adjacency vector to number each region.

(4) Count the number of points contained in each area, and select the area with the largest area as the pupil area.

(5) and select the head and tail ends of each column vector in the region to extract the boundary of the pupil.

(6) Use the obtained boundary to fit the ellipse. Based on the idea of least square method, the fitting equation of the ellipse is calculated by using the singular value decomposition algorithm. Obtain the final boundary and pupil region.

(7) Stretch the pupil area obtained above with respect to the minor axis of the ellipse, so that the pupil area becomes a circular area.

3. Measurement of pupil light and dark distribution ratio

Combined with the principle diagram of the infrared off-center photography method in Figure 3, the off-center photography method is a method for measuring the diopter of the human eye. The pupil area Figure 2 obtained above is obtained by shooting with the optical path shown in Figure 3. The expression of diopter D can be deduced from the principle of eccentric photography combined with the light path diagram:

Wherein D is a diopter, E is the distance from the infrared light source 302 shown in Figure 3 to the edge of the diaphragm 301, A is the distance from the camera to the pupil 303, R is the radius of the pupil, and k is the ratio of the dark area of the pupil to the diameter (this will be mentioned later. arrive). In conjunction with Fig. 3 and Fig. 4, the adjustment of the diopter to the focal length of the eye (eyeball 305) is added into it to become a normal eye, which is imaged on the retina 304. The Gaussian imaging formula of this optical path is:

Transforming the subtransposition of the above formula, we can get:

The meaning of deformation here is not trivial, and it can be understood in this way: the adjustment of the diopter to the focal length of the human eye is transformed into the adjustment of the diopter to the object distance. If this process is considered that the object point is imaged on the normal retina, then the distance S from the object point to the pupil, that is, the longitudinal distance of the line of sight satisfies:

From the above relationship, substituting the expression of diopter D into the expression of sight depth S:

Among them, S is the longitudinal distance of the line of sight, R is the radius of the pupil, k is the ratio of the dark area of the pupil to the diameter, A is the distance between the camera and the pupil, and E is the distance from the infrared light source to the edge of the diaphragm.

From the above principle derivation, it can be seen that the light and dark distribution characteristics of the pupil (ratio K of the dark area of the pupil to the diameter) are different at different line-of-sight distances, and the ratio of the dark area increases as the distance increases. Since the diaphragm of the device blocks the lower part of the camera, the brightness and darkness distribution of the captured pupil image changes vertically. Therefore, the diameter in the vertical direction is simply selected to calculate the brightness and darkness distribution characteristics. The average value of the gray value of the entire pupil area As a threshold, the ratio of the length of the diameter in the vertical direction greater than the threshold to the pupil diameter is calculated. The measurement of a picture is shown in Figure 5.

4. Calibration process

A circular target with a black dot in the center is used as a calibration object, and the human eye is fixed on the bull's-eye. In the direction of the eye line of sight, between 50cm-300cm, do a calibration measurement every 50cm. Obtain the line-of-sight distance s and the corresponding ratio k of the dark area of the pupil to the entire vertical axis, substitute the number into the theoretically derived formula, and obtain the calibration parameters by fitting the curve. The fitting diagram is shown in Figure 6. , the calibrated formula can be used for actual measurement. The vertical expression of the line of sight after scaling is:

The calibration expression obtained above is used to calculate the result shown in Fig. 5, where k=0.21, which is substituted into the above formula: the longitudinal distance of line of sight S=0.873m.

Claims (4)

1. a kind of measuring method of human eye sight fore-and-aft distance, it is characterised in that comprise the steps：

(1) build eccentric image measuring device on optical bench, including camera, infrared light supply, infrared filter, telescope, Face support, cutoff filter；Cutoff filter, telescope and camera on optical path direction is coaxially contour, Face support is used to support testee face；Optical path is folded into face support side by cutoff filter；Look in the distance Mirror is used to shorten shooting light path；Camera sets diaphragm and infrared light supply in the low order end of light path between telescope and camera, make The pupil image of sharpness of border can be obtained with infrared light supply, diaphragm shelters from the latter half of camera；It is provided with before camera The infrared filter, for reducing the impact of ambient visible light；Camera also connects computer by USB interface, on computers With carrying out image procossing and calculating on MATLAB software platforms；

(2) pupil boundary detection, with the distance of swimming algorithm process pupil region is obtained, then obtains pupil boundary with equation model method；

(3) the bright dark distribution of pupil region is calculated, because the diaphragm of device is the latter half of blocking camera, the pupillogram of shooting The bright dark distribution of picture is vertical change, can choose the diameter of vertical direction to calculate bright dark distribution characteristics, whole pupil region Gray value mean value calculates the ratio of the length with PD for being diametrically more than threshold value of vertical direction as threshold value；

(4) measurement of fore-and-aft distance, by the scaling parameter before measuring and the bright dark distribution situation of current pupil, calculates and obtains sight line Fore-and-aft distance；The scaling parameter includes that pupil radium R, camera and people are oculopupillary apart from A, pupil dark areas and diameter ratio Value k；The calculating obtains the concrete calculation expression of sight line fore-and-aft distance：

$S=\frac{2R\·A\·k}{E+2R\·k},$

S is sight line fore-and-aft distance, and R is pupil radium, and k is pupil dark areas and diameter ratio, and A is camera and interpupillary distance, E It is that light source deviates optical axis distance, wherein pupil dark areas can be obtained from pupil image with diameter ratio k by image processing techniques, Parameter E, A and R are obtained by calibration process；For each frame eye image, pupil dark areas and diameter ratio k are calculated Substituting into expression above can just calculate the fore-and-aft distance of current human eye sight.

2. the measuring method of a kind of human eye sight fore-and-aft distance according to claim 1, it is characterised in that with improved The fore-and-aft distance of infrared eccentric Photographic technique direct measurement human eye sight, concrete improved method is for human eye by human eye diopter The amendment of focal length, becomes to be amendment of the diopter to object distance, and then obtains the calculation expression of sight line fore-and-aft distance.

3. the measuring method of a kind of human eye sight fore-and-aft distance according to claim 1, it is characterised in that at image The method of reason calculates the bright dark distribution situation of pupil image, and pupil region is obtained from eye image initially with distance of swimming algorithm； Then pupil center is obtained to the method for zone boundary ellipse fitting, and border is modified to into ellipse；Then flexible change is carried out Change, pupil region is stretched circular, finally take the diameter of circular vertical to calculate the bright dark distribution characteristics of pupil region.

4. the measuring method of a kind of human eye sight fore-and-aft distance according to claim 1, it is characterised in that using infrared section Only optical filter causes optical path vertical with human eye sight light path, and measurement process does not interfere with human eye field range.