RFID dual-tag eye movement detection method

Abstract

The present invention discloses an RFID dual-tag eye movement detection method: Step 1, select a reference tag based on the RSSI difference and reading rate difference between the contact lens tag and the commercial tag; Step 2, arrange two antennas, a reference tag, and a contact lens tag; Step 3, filter the four phase information obtained in Step 2, and calculate the phase difference between the contact lens tag and the reference tag in each antenna and the first-order difference value of the phase difference of antenna 1 based on the result of the filtering process, thereby detecting the blinking action; Step 4, remove the blinking information from the phase difference vector obtained in Step 3, and use the peak method to detect the eye movement direction. Since the present invention uses commercial RFID equipment, the difficulty of operation is greatly reduced, thereby greatly reducing the detection cost. At the same time, the present invention takes into account the influence of the surrounding environment and body activities on the signal when judging eye movements, thereby improving the accuracy of eye movement judgment.

Description

Technical Field

[0001] The present invention belongs to the field of wireless sensing technology, and in particular relates to an RFID double-tag eye movement detection method. Background Art

[0002] The human eye is a vital sensory organ and one of the primary tools for humans to gather information about their surroundings. Eye movements are closely linked to brain health, and eye movement patterns can indicate this state. Studies have shown that analyzing eye movement patterns can aid in the early diagnosis of neurological disorders such as ADHD, autism, and Parkinson's disease. Furthermore, the detection of eye movements and blinks is widely used in human-computer interaction, providing new avenues for interaction and significantly improving its convenience. For example, eye movement detection can serve as an effective human-computer interaction solution for individuals with paralysis, helping them communicate with others. In scenarios where verbal or physical communication is inconvenient, eye movement interaction can improve communication efficiency while protecting user privacy.

[0003] Existing eye movement detection methods can be roughly divided into two categories: contact eye movement detection methods and non-contact detection methods.

[0004] Contact eye movement detection methods require hardware to be connected to the human eye. For example, connecting multi-channel EOG electrodes to the eye can achieve high-precision eye movement tracking, but this method is invasive and poses potential harm to the human eye. It also affects human observation and makes people less sensitive to their surroundings. To make EOG measurements less invasive, related research suggests integrating EOG sensors into headphones or glasses. However, because EOG sensors are very sensitive to the number of channels, the sensing channels provided by headphones or glasses are very limited. Therefore, integrating EOG sensors into other parts of the human body will significantly reduce detection accuracy.

[0005] Non-contact methods rely primarily on cameras to capture eye movements. Undoubtedly, this method can generally provide high-precision eye tracking under suitable lighting conditions. However, slight changes in light intensity can significantly affect tracking accuracy, and users need to keep light intensity within a certain range at all times. Secondly, camera-based eye movement tracking methods can easily lead to privacy issues. At the same time, camera-based solutions have high requirements for deployment angles and distances, requiring the camera to be placed very close to the eyeballs. This method undoubtedly limits its use scenarios, and users cannot use it in moving scenes. Summary of the Invention

[0006] The purpose of the present invention is to provide an RFID dual-tag eye movement detection method to solve the problems in the prior art of low detection accuracy due to environmental influences and only applicable to close distances.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0008] An RFID dual-tag eye movement detection method comprises the following steps:

[0009] Step 1: Select a reference tag based on the RSSI difference and reading rate difference between the contact lens tag and the commercial tag;

[0010] Step 2: Arrange two antennas, a reference tag, and a contact lens tag, use a commercial RFID reader to obtain the phase and timestamp information of the contact lens tag and the reference tag received by the two antennas, perform time domain interpolation processing, and update the phase vector based on the interpolation results;

[0011] Step 3: Filter the four phase information obtained in step 2, and calculate the phase difference between the contact lens tag and the reference tag in each antenna and the first-order difference of the phase difference of antenna 1 based on the filtering results, so as to detect the blinking action;

[0012] Step 4: After removing the blink information from the phase difference phase1 vector and the phase difference phase2 vector obtained in step 3, the eye movement direction is detected using a peak method.

[0013] Furthermore, the step 1 includes the following sub-steps:

[0014] Step S10: Arrange an antenna connected to a commercial RFID reader in the application scenario, place the contact lens tag at a certain distance from the antenna, and record the contact lens tag reading rate V lens , whose RSSI is recorded as RSS lens ;

[0015] Step S11: In the same application scenario as step S10, select a commercial RFID tag A and place tag A in the same position as the contact lens tag in step S10, and record the reading rate V of tag A. ref , whose RSSI is recorded as RSS ref ; Calculate the difference between label A and the contact lens label using the following formula:

[0016] Diff rate =(V lens -V ref ) / V lens

[0017] Diff rss =RSS lens -RSS ref

[0018] Among them, Diff rateis the difference in reading rate between the contact lens tag and tag A, Diff rss is the difference in RSSI between the contact lens tag and tag A;

[0019] If tag A satisfies both of the following formulas, tag A is used as the reference tag. Otherwise, step S10 is executed to select another commercial RFID tag A until the following two formulas are satisfied:

[0020] Diff rate ≤ε rate

[0021] Diff rss ≤ε rss

[0022] Among them, ε rate is the threshold of the difference in reading rate between the contact lens tag and the reference tag, ε rss is the RSSI difference threshold between the contact lens tag and the reference tag.

[0023] Furthermore, in step 10, the contact lens label is obtained by connecting a three-turn spiral copper coil to the RFID chip, and wrapping the copper coil and the RFID chip with a PDMS substrate.

[0024] Furthermore, in step S11, ε rate =0.08;ε rss =1.

[0025] Furthermore, step 2 includes the following sub-steps:

[0026] Step S20, deploying two antennas and a reference tag;

[0027] In step S21, a commercial RFID reader is used to simultaneously collect the echo signals of the two tags received by antenna 1 and antenna 2, and extract the phase information and timestamp information of antenna 1 and antenna 2 respectively; the phase information and timestamp information within a certain time period are saved in the form of vectors:

[0028] Antenna 1: contact lens tag phase information lens1 , contact lens label timestamp information time lens1 , reference tag phase information phase ref1 , reference tag timestamp information time ref1 ;

[0029] Antenna 2: contact lens tag phase information lens2 , contact lens label timestamp information time lens2 , reference tag phase information phase ref2, reference tag timestamp information time ref2 ;

[0030] Step S22, expand the timestamp vector time lens1 , time lens2 , time ref1 , time ref2 Specifically, the start time is obtained according to the timestamp information in step S21 start and end time end , the timestamp vector time lens1 , time lens2 , time ref1 , time ref2 All updated to time start :time gap :time end , where time gap is the time interval;

[0031] Start time start =min{time lens1 , time lens2 , time ref1 , time ref2},

[0032] End time end =max{time lens1 , time lens2 , time ref1 , time ref2};

[0033] Step S23, the phase vector obtained in step S21 is used as a standard based on the timestamp information obtained in step S22, and the phase vector phase lens1 , phase lens2 , phase ref1 , phase ref2 Perform three hermite interpolations in the time domain and use the interpolated results to update the phase vector phase lens1 , phase lens2 , phase ref1 , phase ref2 .

[0034] Furthermore, the operation of step S20 is as follows:

[0035] Two antennas with a gain of 10 dBi were connected to the two ports of a commercial RFID reader, and the two antennas were marked as antenna 1 and antenna 2 respectively. Antenna 1 was placed on the left side of the subject, and antenna 2 was placed perpendicular to antenna 1, that is, above the subject. The subject wore a contact lens label and the reference label was attached to the forehead or head.

[0036] Furthermore, step 3 includes the following sub-steps:

[0037] Step S30: Perform filtering on the four phase vectors obtained after the expansion in step S23 as shown in the following equations, and use the filtered results to update the phase vector phase lens1 , phase lens2 , phase ref1 , phase ref2 :

[0038] phase i+1 =phase i +KGain i+1 *(phase i+1 -phase i )

[0039] Among them, phase i is the i-th value in the corresponding phase vector, for phase lens1 During filtering, phase i Phase lens1 The i-th value of the vector, KGain i is the Kalman gain;

[0040] Step S31, calculating the phase difference phase1 of the phase information between the contact lens tag and the reference tag collected by antenna 1 in step 30 and the phase difference phase2 of the phase information between the contact lens tag and the reference tag collected by antenna 2, using the following calculation formula:

[0041] phase1=phase lens1 -phase ref1

[0042] phase2=phase lens2 -phase ret2

[0043] Step S32, calculate the first-order difference value phasediff1 of the phase difference of antenna 1 i , the calculation formula is as follows:

[0044] phasediff1 i =phase1 i+step -phase1i

[0045] Among them, phase1 i is the i-th element value of the phase difference phase1 vector, and step is the difference step size;

[0046] Step S32, traverse the phasediff1 vector from front to back, if the following conditions are met:

[0047] For positive integers j, a, b, c, d, we have:

[0048]

[0049] It is considered that there is a blinking action between the time corresponding to the j index as the starting point and the time corresponding to the j+d index as the end point, where diff_max is the maximum value of the first-order difference vector phasediff1, and the sign is positive; diff_min is the minimum value of the first-order difference vector phasediff1, and the sign is negative.

[0050] Furthermore, in step S30, the Kalman gain KGaini+1 is calculated by the following formula:

[0051] KGain i+1 =p i+1 / (p i+1 +r)

[0052] Where r is the noise covariance of the measurement process r = 0.03, p i+1 is the estimated error covariance in the phase vector, which is calculated as follows:

[0053] p i+1 =(1-KGain i )*p i +q

[0054] Where q is the prediction process noise covariance q = 0.001, p i The initial value of p0 = 10, KGain i The initial value KGain0=0.

[0055] Furthermore, step 4 includes the following sub-steps:

[0056] Step S40, replacing the element value corresponding to the blink action determined in step S32 in phase1 and phase2 obtained in step 31 with the phase information before the blink action, to obtain new phase1 and phase2; the specific operation is: replacing the j-th element to the j+d-th element phase1[j:jd] of the phase1 vector obtained in step 3 with the j-1-th element phase1[j-1] of the phase1 vector, and replacing the j-th element to the j+d-th element phase2[j:jd] of the phase2 vector obtained in step 31 with the j-1-th element phase2[j-1] of phase2;

[0057] Step S41, traverse the phase1 vector and phase2 vector obtained in step S40 simultaneously, for index j and index j+e, where e is the index interval, determine the peak direction in each vector and obtain the eye movement direction corresponding to the peak direction:

[0058] If the phase1 vector peak direction is upward and the phase2 vector peak direction is upward, then the eye movement direction is upward;

[0059] If the phase1 vector peak direction is downward and the phase2 vector peak direction is downward, the eye movement direction is downward;

[0060] If the phase1 vector peak direction is downward and the phase2 vector peak direction is upward, the eye movement direction is left;

[0061] If the phase1 vector peak direction is upward and the phase2 vector peak direction is downward, the eye movement direction is right.

[0062] Furthermore, in step S41, the peak direction determination method is:

[0063] If less than If the phase value of the index points is greater than the median value, and the difference between the maximum phase value and the minimum phase value is greater than 0.5, the peak direction is considered to be upward;

[0064] If less than If the phase value of the index points is less than the median value, and the difference between the maximum phase value and the minimum phase value is greater than 0.5, the peak direction is considered to be downward;

[0065] In step S41, the index interval e is a variable value between 150 and 600.

[0066] Compared with the prior art, the technical effects of the present invention are as follows:

[0067] 1. Improved detection accuracy in sports or other dynamic scenes. This invention uses the change in the phase difference between the contact lens tag and the reference tag to distinguish the type and direction of eye movement. Since the reference tag is used to assess environmental changes, the phase difference between the contact lens tag and the reference tag is only affected by eye movement. Therefore, the detection rate in dynamic scenes is relatively high, effectively reducing the impact of dynamic environmental changes.

[0068] 2. Improved detection distance. The present invention uses radio frequency RFID signals to detect eye movements and transmit data, capable of achieving a detection distance of over 1.5 meters. This overcomes the drawback of existing methods that use NFC or Bluetooth to sense and transmit data, resulting in shorter distances. The present invention significantly improves the detection distance.

[0069] 3. Simple deployment and low cost. The deployment process of the present invention does not need to consider the influence of lighting and other factors. It only needs to deploy the antenna in a suitable position to perform detection. There is no need to limit the angle to a small range. As long as the general direction is correct, the deployment is convenient and fast.

[0070] 4. Good privacy. The present invention adopts RFID technology, so it does not involve transmission methods such as sound and light that may infringe on user privacy, which greatly guarantees the privacy of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 It is a schematic diagram of an application scenario of the present invention;

[0072] Figure 2 This is the result diagram of recognition accuracy of different eye movement directions;

[0073] Figure 3 This is the detection accuracy result diagram at different distances;

[0074] Figure 4 This is a graph of detection accuracy results in a general office scenario;

[0075] Figure 5 This is the detection accuracy result graph in the driving scene;

[0076] Figure 6 This is the detection accuracy result diagram in the treadmill scene;

[0077] Figure 7 This is the result graph of eye movement direction detection accuracy at different eye movement speeds;

[0078] Figure 8 This is the detection accuracy result diagram when multiple targets are detected simultaneously.

[0079] The present invention is further described in detail below with reference to the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0080] The human eye is a vital sensory organ and one of the primary tools for humans to gather information about their surroundings. Eye movements can be used to indicate brain health or improve human-computer interaction. However, previous approaches have drawbacks such as being invasive, requiring short distances, and requiring static detection.

[0081] The present invention provides a new eye movement detection method using dual RFID tags. The method utilizes the signal difference between the dual tags to eliminate the influence of the environment on the detection accuracy, realizes detection in dynamic scenes, and at the same time improves the detection accuracy and detection distance.

[0082] 1. Detailed steps of the method of the present invention

[0083] Step 1: Select a reference tag based on the RSSI difference and read rate difference between the contact lens tag and the commercial tag. This includes the following sub-steps:

[0084] Step S10: Arrange an antenna connected to a commercial RFID reader in the application scenario, place the contact lens tag at a distance of 80 cm from the antenna, and record the contact lens tag reading rate V lens , whose RSSI is recorded as RSS lens ;

[0085] A contact lens tag with an RFID chip was made. The shape of the contact lens tag is similar to that of Google contact lenses. The contact lens tag of the present invention uses a three-turn spiral copper coil as an antenna, which is connected to the RFID chip, and a PDMS substrate is used to wrap the copper coil and RFID chip.

[0086] Step S11: In the same application scenario as step S10, select a commercial RFID tag A and place tag A in the same position as the contact lens tag in step S10, and record the reading rate V of tag A. ref , whose RSSI is recorded as RSS ref ; Calculate the difference between label A and the contact lens label using the following formula:

[0087] Diff rate =(V lens -V ref ) / V lens (Formula 1)

[0088] Diff rss =RSS lens -RSS ref (Formula 2)

[0089] Among them, Diff rate is the difference in reading rate between the contact lens tag and tag A, Diff rssis the difference in RSSI between the contact lens tag and tag A.

[0090] If tag A satisfies both formula 3 and formula 4, tag A is used as a reference tag. Otherwise, step S10 is executed to reselect other commercial RFID tags A until the conditions are met.

[0091] Diff rate ≤ε rate (Formula 3)

[0092] Diff rss ≤ε rss (Formula 4)

[0093] Among them, ε rate is the threshold of the difference in reading rate between the contact lens tag and the reference tag, ε rss is the RSSI difference threshold between the contact lens tag and the reference tag.

[0094] Better, reading rate difference threshold ε rate =0.08; RSSI difference threshold ε rss =1.

[0095] Step 2: Arrange two antennas, a reference tag, and a contact lens tag. Use a commercial RFID reader to obtain the phase and timestamp information of the contact lens tag and reference tag received by the two antennas, perform time domain interpolation processing, and update the phase vector based on the interpolation results. This includes the following sub-steps:

[0096] Step S20: Deploy the antenna and reference tag in appropriate locations. Specifically, connect two antennas with a gain of 10dBi to the two ports of a commercial RFID reader, and label them as antenna 1 and antenna 2. Place antenna 1 on the left side of the subject, and antenna 2 on the vertical plane of antenna 1, i.e. above the subject. The subject wears a contact lens tag, and the reference tag is attached to the forehead or head, as shown in the figure. Figure 1 .

[0097] In step S21, a commercial RFID reader simultaneously collects the echo signals from both tags received by antennas 1 and 2, and extracts the phase information and timestamp information from antennas 1 and 2, respectively. The phase information and timestamp information for several minutes (preferably one minute) are stored as vectors. The stored information is shown in Table 1.

[0098] Table 1: Antenna and tag correspondence information

[0099]

[0100] Step S22, expand the timestamp vector time lens1, time lens2 , time ref1 , time ref2 Specifically: get the start time time according to the timestamp information in step S21 start and end time end , the timestamp vector time lens1 , time lens2 , time ref1 , time ref2 All updated to time start :time gap :time end , where time gap is the time interval;

[0101] Start time start =min{time lens1 , time lens2 , time ref1 , time ref2},

[0102] End time end =max{time lens1 , time lens2 , time ref1 , time ref2};

[0103] Better, time interval time gap =1ms.

[0104] The purpose of this step is to make the time interval between adjacent elements in the timestamp vector smaller;

[0105] Step S23, the phase vector obtained in step S21 is used as a standard based on the timestamp information obtained in step S22, and the phase vector phase lens1 , phase lens2 , phase ref1 , phase ref2 Perform three hermite interpolations in the time domain and use the interpolated results to update the phase vector phase lens1 , phase lens2 , phase ref1 , phase ref2 .

[0106] Step 3: Filter the four phase information obtained in step 2, and calculate the phase difference between the contact lens tag and the reference tag in each antenna and the first-order difference of the phase difference of antenna 1 based on the filtering results, so as to detect the blinking action. The following sub-steps are included:

[0107] Step S30, the four expanded phase vectors obtained in step S23 are filtered as shown in formula 5, and the phase vector phase is updated using the filtered results. lens1 , phase lens2 , phase ref1 , phase ref2 :

[0108] phase i+1 =phase i +KGain i+1 *(phase i+1 -phase i ) (Formula 5)

[0109] Among them, phase i is the i-th value in the corresponding phase vector, for example, lens1 During filtering, phase i Phase lens1 The i-th value of the vector, KGain i is the Kalman gain.

[0110] Kalman gain KGain i+1 Calculated by the following formula:

[0111] KGain i+1 =p i+1 / (p i+1 +r) (Formula 11)

[0112] Where r is the noise covariance of the measurement process r = 0.03, p i+1 is the estimated error covariance in the phase vector, which is calculated as follows:

[0113] p i+1 =(1-KGain i )*p i +q (Formula 12)

[0114] Where q is the prediction process noise covariance q = 0.001, p i The initial value of p0 = 10, KGain i The initial value KGain0=0.

[0115] Step S31, calculating the phase difference phase1 of the phase information between the contact lens tag and the reference tag collected by antenna 1 in step 30 and the phase difference phase2 of the phase information between the contact lens tag and the reference tag collected by antenna 2:

[0116] phase1=phase lens1 -phase ref1 (Formula 6)

[0117] phase2=phase lens2 -phase ref2 (Formula 7)

[0118] Step S32, calculate the first-order difference value phasediff1 of the phase difference of antenna 1 i :

[0119] phasediff1 i =phase1 i+step -phase1 i (Formula 8)

[0120] Among them, phase1 i is the i-th element value of the phase difference phase1 vector, and step is the difference step size. Preferably, the difference step size is 300.

[0121] Step S32, traverse the phasediff1 vector from front to back, if the following conditions are met:

[0122] For positive integers j, a, b, c, d, we have

[0123]

[0124] It is considered that there is a blinking action between the time corresponding to the i index as the starting point and the time corresponding to the j+d index as the end point, where diff_max is the maximum value of the first-order difference vector phasediff1, and the sign is positive; diff_min is the minimum value of the first-order difference vector phasediff1, and the sign is negative.

[0125] Step 4: After removing the blink information from the phase difference phase1 vector and the phase difference phase2 vector obtained in step 3, the eye movement direction is detected using the peak value method. This includes the following sub-steps:

[0126] In step S40 , the element value corresponding to the blinking action determined in step S32 is replaced in phase 1 and phase 2 obtained in step 31 by the phase information before the blinking action, to obtain new phase 1 and phase 2.

[0127] The specific operation is: replace the j-th element to the j+d-th element phase1[j:jd] of the phase1 vector obtained in step 3 with the j-1-th element phase1[j-1] of the phase1 vector, and replace the j-th element to the j+d-th element phase2[j:jd] of the phase2 vector obtained in step 31 with the j-1-th element phase2[j-1] of phase2.

[0128] Step S41: traverse the phase1 vector and phase2 vector obtained in step S40 simultaneously. For index j and index j+e, determine the peak direction in each vector and look up the table below to obtain the eye movement direction corresponding to the peak direction.

[0129] Table 2: Eye movement direction judgment

[0130] Phase1 vector peak direction Phase2 vector peak direction Eye movement direction superior superior superior Down Down Down Down superior Left superior Down right

[0131] Where e is the index interval.

[0132] Specifically, the peak direction determination method is:

[0133] If less than If the phase value of the index points is greater than the median value, and the difference between the maximum phase value and the minimum phase value is greater than 0.5, the peak direction is considered to be upward.

[0134] If less than If the phase value of the index points is less than the median value, and the difference between the maximum phase value and the minimum phase value is greater than 0.5, the peak direction is considered to be downward.

[0135] Preferably, the index interval e is a variable value between 150 and 600, with a default value of 300.

[0136] 2. Experimental Verification

[0137] In order to demonstrate the feasibility and effectiveness of the method of the present invention, the applicant conducted the following experimental verification.

[0138] 1. Recognition accuracy under static conditions:

[0139] The bionic eye was placed on a table and performed preset eye movements to simulate human eye movements. The receiving antenna was placed between 50cm and 150cm. A set of experiments was performed every 20cm. Each set of experiments included 100 blinks and 300 eye movements. The results were judged according to the method described in the present invention. Figure 2 、 Figure 3 , at a distance of 150cm, it can still achieve an accuracy rate of more than 81%, among which the accuracy of eye movement direction is the lowest at 93%.

[0140] 2. Different scenarios:

[0141] To verify the impact of different environments and different movement amplitudes on detection accuracy, the applicant wore bionic glasses on the head and attached a reference label to the forehead. The above static experiments were repeated in normal office scenes (the subject was almost motionless), driving scenes (the subject's upper body moved), and treadmill speed running scenes (the subject moved significantly). The results were as follows: Figure 4 、 Figure 5 、 Figure 6 ,The recognition accuracy decreases as the movement amplitude increases, but it can still reach an accuracy of more than 85% within a distance of 130cm in a running situation.

[0142] The impact of eye movement speed:

[0143] To verify the effect of different eye movement speeds on detection accuracy, the present invention conducted comparative experiments on rapid eye movement (65ms / 30°), medium eye movement (300ms / 30°), and slow eye movement (1s / 30°). The results are as follows: Figure 7 , the recognition accuracy rate is over 85% for low-speed and medium-speed eye movements, and it decreases slightly for high-speed eye movements. However, high-speed eye movements are a speed that is difficult for our human eyes to reach, so the system meets the general speed requirements.

[0144] 3. Impact between multiple labels:

[0145] To verify the impact of multiple contact lens tags working simultaneously on the detection accuracy, we used one contact lens as the target tag and the other tag worked normally next to it to calculate the recognition accuracy. The result is: Figure 8 , the accuracy rate is still above 80% at a distance of 140cm from the receiving antenna.

Claims

1. A RFID dual-tag eye movement detection method, characterized in that: The steps include: Step 1: Select a reference tag based on the RSSI difference and reading rate difference between the contact lens tag and the commercial tag; Step 2: Arrange two antennas, a reference tag, and a contact lens tag, use a commercial RFID reader to obtain the phase and timestamp information of the contact lens tag and the reference tag received by the two antennas, perform time domain interpolation processing, and update the phase vector based on the interpolation results; Step 3: Filter the four phase information obtained in step 2, and calculate the phase difference between the contact lens tag and the reference tag in each antenna and the first-order difference of the phase difference of antenna 1 based on the filtering results, so as to detect the blinking action; Step 4: After removing the blink information from the phase difference phase1 vector and the phase difference phase2 vector obtained in step 3, the eye movement direction is detected using a peak method.

2. The RFID dual-tag eye movement detection method according to claim 1, wherein: The step 1 includes the following sub-steps: Step S10: Arrange an antenna connected to a commercial RFID reader in the application scenario, place the contact lens tag at a certain distance from the antenna, and record the contact lens tag reading rate V lens , whose RSSI is recorded as RSS lens ; Step S11: In the same application scenario as step S10, select a commercial RFID tag A and place tag A in the same position as the contact lens tag in step S10, and record the reading rate V of tag A. ref , whose RSSI is recorded as RSS ref ; Calculate the difference between label A and the contact lens label using the following formula: Diff rate =(V lens -V ref ) / V lens Diff rss =RSS lens -RSS ref Among them, Diff rate is the difference in reading rate between the contact lens tag and tag A, Diff rss is the difference in RSSI between the contact lens tag and tag A; If tag A satisfies both of the following formulas, tag A is used as the reference tag. Otherwise, step S10 is executed to select another commercial RFID tag A until the following two formulas are satisfied: Diff rate ≤ε rate Diff rss ≤ε rss Among them, ε rate is the threshold of the difference in reading rate between the contact lens tag and the reference tag, ε rss is the RSSI difference threshold between the contact lens tag and the reference tag.

3. The RFID dual-tag eye movement detection method according to claim 2, wherein: In step S10, the contact lens label is obtained by connecting a three-turn spiral copper coil to an RFID chip, and wrapping the copper coil and the RFID chip with a PDMS substrate.

4. The RFID dual-tag eye movement detection method according to claim 2, wherein: In the step S11, ε rate =0.08;ε rss =1.

5. The RFID dual-tag eye movement detection method according to claim 1, wherein: The step 2 includes the following sub-steps: Step S20, deploying two antennas and a reference tag; In step S21, a commercial RFID reader is used to simultaneously collect the echo signals of the two tags received by antenna 1 and antenna 2, and extract the phase information and timestamp information of antenna 1 and antenna 2 respectively; the phase information and timestamp information within a certain time period are saved in the form of vectors: Antenna 1: contact lens tag phase information lens1 , contact lens label timestamp information time lens1 , reference tag phase information phase ref1 , reference tag timestamp information time ref1 ; Antenna 2: contact lens tag phase information lens2 , contact lens label timestamp information time lens2 , reference tag phase information phase ref2 , reference tag timestamp information time ref2 ; Step S22, expand the timestamp vector time lens1 ,time lens2 ,time ref1 ,time ref2 Specifically, the start time is obtained according to the timestamp information in step S21 start and end time end , the timestamp vector time lens1 ,time lens2 ,time ref1 ,time ref2 All updated to time start :time gap :time end , where time gap is the time interval; Start time start =min{time lens1 ,time lens2 ,time ref1 ,time ref2 }, End time end =max{time lens1 ,time lens2 ,time ref1 ,time ref2 }; Step S23, the phase vector obtained in step S21 is used as a standard based on the timestamp information obtained in step S22, and the phase vector phase lens1 ,phase lens2 ,phase ref1 ,phase ref2 Perform three hermite interpolations in the time domain and use the interpolated results to update the phase vector phase lens1 ,phase lens2 ,phase ref1 ,phase ref2 .

6. The RFID dual-tag eye movement detection method according to claim 5, wherein: The operation of step S20 is as follows: Two antennas with a gain of 10 dBi were connected to the two ports of a commercial RFID reader, and the two antennas were marked as antenna 1 and antenna 2 respectively. Antenna 1 was placed on the left side of the subject, and antenna 2 was placed perpendicular to antenna 1, that is, above the subject. The subject wore a contact lens label and the reference label was attached to the forehead or head.

7. The RFID dual-tag eye movement detection method according to any one of claims 5 to 6, characterized in that: The step 3 includes the following sub-steps: Step S30: Perform filtering on the four phase vectors obtained after the expansion in step S23 as shown in the following equations, and use the filtered results to update the phase vector phase lens1 ,phase lens2 ,phase ref1 ,phase ref2 : phase i+1 =phase i +KGain i+1 *(phase i+1 -phase i ) Among them, phase i is the i-th value in the corresponding phase vector, for phase lens1 During filtering, phase i Phase lens1 The i-th value of the vector, KGain i is the Kalman gain; Step S31, calculating the phase difference phase1 of the phase information between the contact lens tag and the reference tag collected by antenna 1 in step 30 and the phase difference phase2 of the phase information between the contact lens tag and the reference tag collected by antenna 2, using the following calculation formula: phase1=phase lens1 -phase ref1 phase2=phase lens2 -phase ref2 Step S32, calculate the first-order difference value phasediff1 of the phase difference of antenna 1 i , the calculation formula is as follows: phasediff1 i =phase1 i+step -phase1 i Among them, phase1 i is the i-th element value of the phase difference phase1 vector, and step is the difference step size; Step S32, traverse the phasediff1 vector from front to back, if the following conditions are met: For positive integers j, a, b, c, d, we have: It is considered that there is a blinking action between the time corresponding to the j index as the starting point and the time corresponding to the j+d index as the end point, where diff_max is the maximum value of the first-order difference vector phasediff1, and the sign is positive; diff_min is the minimum value of the first-order difference vector phasediff1, and the sign is negative.

8. The RFID dual-tag eye movement detection method according to claim 7, wherein: In step S30, the Kalman gain KGain i+1 Calculated by the following formula: KGain i+1 =p i+1 / (p i+1 +r) Where r is the noise covariance of the measurement process r = 0.03, p i+1 is the estimated error covariance in the phase vector, which is calculated as follows: p i+1 =(1-KGain i )*p i +q Where q is the prediction process noise covariance q = 0.001, p i The initial value of p0 = 10, KGain i The initial value KGain0=0.

9. The RFID dual-tag eye movement detection method according to claim 7, wherein: The step 4 includes the following sub-steps: Step S40: Replace the element value corresponding to the blink action determined in step S32 in phase1 and phase2 obtained in step 31 with the phase information before the blink action, to obtain new phase1 and phase2. The specific operation is: replace the j-th element to the j+d-th element phase1[j:jd] of the phase1 vector obtained in step 3 with the j-1-th element phase1[j-1] of the phase1 vector, and replace the j-th element to the j+d-th element phase2[j:jd] of the phase2 vector obtained in step 31 with the j-1-th element phase2[j-1] of phase2. Step S41, traverse the phase1 vector and phase2 vector obtained in step S40 simultaneously, for index j and index j+e, where e is the index interval, determine the peak direction in each vector and obtain the eye movement direction corresponding to the peak direction: If the phase1 vector peak direction is upward and the phase2 vector peak direction is upward, then the eye movement direction is upward; If the phase1 vector peak direction is downward and the phase2 vector peak direction is downward, the eye movement direction is downward; If the phase1 vector peak direction is downward and the phase2 vector peak direction is upward, the eye movement direction is left; If the phase1 vector peak direction is upward and the phase2 vector peak direction is downward, the eye movement direction is right.

10. The RFID dual-tag eye movement detection method according to claim 9, wherein: In step S41, the peak direction determination method is: If less than If the phase value of the index points is greater than the median value, and the difference between the maximum phase value and the minimum phase value is greater than 0.5, the peak direction is considered to be upward; If less than If the phase value of the index points is less than the median value, and the difference between the maximum phase value and the minimum phase value is greater than 0.5, the peak direction is considered to be downward; In step S41, the index interval e is a variable value between 150 and 600.

Images