CN106963386B - eye blinking recognition method and device - Google Patents

Abstract

The present disclosure relates to a method and device for eye blink recognition. The method includes: sampling a signal that satisfies a preset sampling condition among voltage signals output by a body motion detection chip to obtain a sampling data sequence; selecting any sampling point from untraversed sampling points as a target sampling point; When the point satisfies the preset effective condition of the blink sampling point, create a cluster for the target sampling point; add the target sampling point and the sampling points that meet the clustering conditions between the target sampling point and the target sampling point to the cluster; repeat execution never A step of selecting any sampling point among the traversed sampling points as a target sampling point until all the sampling points in the sampling data sequence have been traversed; identifying blink parameters according to the established clusters. Through the above technical solution, the calculation amount of data can be reduced, the complexity of the algorithm can be reduced, possible errors in the calculation process can be effectively avoided, and the accuracy rate is greatly improved.

Description

Eye blink recognition method and device

technical field

The present disclosure relates to the field of eye blink detection, and in particular, to a method and device for eye blink recognition.

Background technique

The body motion detection chip integrates a whole set of circuits with electromagnetic wave transmitting and receiving functions into one chip, which is used to detect human body movements. Its advantages are small size, low power consumption, and easy to use. Its basic principle is to emit electromagnetic waves to the outside world and detect the reflected electromagnetic waves, and output the detection results in the form of voltage. When the body motion detection chip is working, transmitting and receiving electromagnetic waves is a continuous process, and its output voltage is also a process of continuous change. When no human motion is detected, the output voltage is stable within an extremely small range of values; when human motion is detected, the output voltage fluctuates, and the amplitude of the fluctuation corresponds to the amplitude of the detected human motion. Frequency corresponds to the frequency of human movement. The detection result output by the body motion detection chip is an analog signal with continuous voltage changes, which is output to the outside world through the chip pins.

In the prior art, the method of identifying and detecting eye blinking based on the voltage signal output by the body motion detection chip needs to continuously calculate the distance between points and store a large amount of data. Therefore, during the implementation of the above method, the requirements for the memory and performance of the hardware are relatively high, which increases the overall cost of the product. In addition, the long-term heavy-duty calculation and storage processing operations also cause a relatively large loss of hardware, which increases the failure rate of the product and greatly reduces the service life.

Contents of the invention

The purpose of the present disclosure is to provide an efficient and accurate eye blink recognition method and device.

In order to achieve the above object, according to the first aspect of the present disclosure, a method for eye blink recognition is provided, the method includes: sampling the signal satisfying the preset sampling condition among the voltage signals output by the body motion detection chip, and obtaining the sampled A data sequence, wherein the body movement detection chip is used to detect eye blinking; any sampling point is selected as a target sampling point from untraversed sampling points; when the target sampling point satisfies the preset effective conditions for blinking sampling points , creating a cluster for the target sampling point; adding the target sampling point and the sampling points satisfying the clustering condition between the target sampling point and the target sampling point to the cluster to complete the establishment of the cluster, wherein , the established cluster corresponds to an eye blink; repeat the step of selecting any sampling point from the untraversed sampling points as the target sampling point until all the sampling points in the sampling data sequence have been traversed; according to the Created clusters, identifying blink parameters.

Optionally, the preset sampling condition is: a voltage change exceeds a preset voltage change threshold.

Optionally, the preset valid condition of the eye blink sampling point includes: in the first area formed with the target sampling point as the center and a preset radius parameter as the radius, all but the target sampling point The number of sampling points is not less than the preset domain density threshold; or in the first area formed with the target sampling point as the center and the preset radius parameter as the radius, except for the target sampling point When the number of sampling points is less than the preset domain density threshold, the voltage value of the target sampling point is less than the nearest peak value and greater than the maximum idle peak value, and the first area except the target The voltage values of sampling points other than the sampling point are all smaller than the peak value and larger than the maximum idle peak value, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters.

Optionally, the method further includes: identifying the target sampling point as a noise point when the target sampling point does not satisfy the preset valid condition of eye blink sampling points, and Sampling points not included in other clusters are identified as noise points; and the noise points are removed.

Optionally, the sampling points satisfying the clustering condition between the target sampling point and the target sampling point are determined in the following manner: in the first area formed with the target sampling point as the center and a preset radius parameter as the radius 1. Add sampling points other than the target sampling point to the set of candidate sampling points; repeatedly execute the traversal of the sampling points in the set of candidate sampling points, and take the traversed sampling point as the center and the preset radius In the second area formed by the parameter radius, if the number of sampling points other than the traversed sampling points is not less than the preset domain density threshold, the second area except the traversed sampling points The step of adding sampling points outside the set to the set of candidate sampling points until all the sampling points in the set of candidate sampling points have been traversed; the sampling points not included in other clusters in the set of candidate sampling points It is determined to be a sampling point that satisfies the clustering condition with the target sampling point.

Optionally, the identifying blink parameters according to the established clusters includes at least one of the following: identifying the total number of established clusters as the number of eye blinks; for each cluster, determining the maximum time difference of cluster members It is identified as a single blink time; for two adjacent clusters, the difference between the adjacent time of the two adjacent clusters is identified as a blink time interval; the sum of the blink time intervals and the voltage signal The ratio between the total times is identified as the blink duty cycle.

Optionally, before the step of identifying blinking parameters according to the established clusters, the method further includes: deleting members in each cluster whose voltage value is less than a maximum idle peak value, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters.

According to the second aspect of the present disclosure, there is provided an eye blink recognition device, the device includes: a sampling module, configured to sample a signal that satisfies a preset sampling condition among the voltage signals output by a body motion detection chip, and obtain the sampled Data sequence, wherein, the body movement detection chip is used to detect eye blinking; the target sampling point selection module is used to select any sampling point from the sampling points obtained by the sampling module and not traversed as the target sampling point; cluster Create a module for creating a cluster for the target sampling point when the target sampling point selected by the target sampling point selection module satisfies the preset effective condition of the eye blink sampling point; a cluster establishment module for the said target sampling point The target sampling point selected by the target sampling point selection module and the sampling points satisfying the clustering condition between the target sampling point and the target sampling point are added to the cluster created by the cluster creation module to complete the establishment of the cluster , wherein the established cluster corresponds to an eye blinking action; the cluster establishment module is also used to re-trigger the target sampling point selection module to select any sampling point from the untraversed sampling points acquired by the sampling module as The target sampling point, until all the sampling points in the sampling data sequence have been traversed; the parameter identification module is configured to identify blink parameters according to the clusters established by the cluster establishment module.

Optionally, the preset sampling condition is: a voltage change exceeds a preset voltage change threshold.

Optionally, the preset valid condition of the eye blink sampling point includes: in the first area formed with the target sampling point as the center and a preset radius parameter as the radius, all but the target sampling point The number of sampling points is not less than the preset domain density threshold; or in the first area formed with the target sampling point as the center and the preset radius parameter as the radius, except for the target sampling point When the number of sampling points is less than the preset domain density threshold, the voltage value of the target sampling point is less than the nearest peak value and greater than the maximum idle peak value, and the first area except the target The voltage values of sampling points other than the sampling point are all smaller than the peak value and larger than the maximum idle peak value, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters.

Optionally, the device further includes: a noise point identification module, configured to select the target sampling point when the target sampling point selected by the target sampling point selection module does not meet the preset valid condition of the eye blink sampling point. The sampling point is identified as a noise point, and the sampling points in the first region that are not included in other clusters are identified as noise points; the noise point removal module is used to remove all the noise points identified by the noise point identification module noise point.

Optionally, the cluster establishment module includes: a candidate sampling point set establishment submodule, which is used to form the target sampling point selected by the target sampling point selection module as the center of the circle and the preset radius parameter as the radius. The sampling points in the first region except the target sampling point are added to the set of candidate sampling points; the candidate sampling point traversal submodule is used to traverse the sampling points in the set of candidate sampling points, and in the set In the second area formed by the traversed sampling point as the center and the preset radius parameter as the radius, when the number of sampling points other than the traversed sampling point is not less than the preset domain density threshold, all the In the second area, sampling points other than the traversed sampling points are added to the set of candidate sampling points until all the sampling points in the set of candidate sampling points have been traversed; the cluster building submodule is used to In the set of candidate sampling points, the sampling points not included in other clusters are determined to be the sampling points satisfying the clustering condition with the target sampling point.

Optionally, the parameter identification module includes at least one of the following: an eye blink identification submodule, configured to identify the total number of clusters established by the cluster establishment module as the number of eye blinks; a single blink time identification submodule A module, for each cluster established by the cluster establishment module, identifying the maximum time difference of the cluster members as a single blink time; a blink interval identification submodule, used for the establishment of the cluster establishment module Two adjacent clusters, the difference between the adjacent time of the two adjacent clusters is identified as the blink time interval; the blink duty cycle identification submodule is used to identify the blink interval identification submodule The ratio between the sum of the blink time intervals and the total time of the voltage signal acquired in the acquisition module is identified as the blink duty cycle.

Optionally, the device further includes: a deletion module, configured to delete parameters in each cluster whose voltage value is less than the maximum idle time peak members, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters.

Through the above technical solution, the identification and clustering of the sampling points belonging to a blink can be realized, and the start and end of the blink can be accurately determined according to the clustering results, thereby accurately completing the identification of the blink parameters. In this process, it is not necessary to calculate the distance between sampling points in real time as in the existing blink recognition algorithm, but more to perform operations such as judgment and marking, which can greatly reduce the amount of data calculation and reduce the cost of the algorithm. The complexity can effectively avoid possible errors in the calculation process, and the accuracy rate is greatly improved.

Other features and advantages of the present disclosure will be described in detail in the detailed description that follows.

Description of drawings

The accompanying drawings are used to provide a further understanding of the present disclosure, and constitute a part of the description, together with the following specific embodiments, are used to explain the present disclosure, but do not constitute a limitation to the present disclosure. In the attached picture:

FIG. 1 is a flow chart of an eye blink recognition method provided according to an embodiment of the present disclosure;

Fig. 2 is the schematic diagram of the blink voltage analog signal detected by the body motion detection chip;

Fig. 3 is a schematic diagram of the discrete data sequence collected according to the blink voltage analog signal detected by the body movement detection chip;

Fig. 4 is a flow chart of the steps of determining the sampling points satisfying the clustering condition between the target sampling points and the target sampling points in the eye blink recognition method provided according to another embodiment of the present disclosure;

Fig. 5 is a block diagram of an eye blink recognition device provided according to an embodiment of the present disclosure;

Fig. 6 is a block diagram of a cluster building module in an eye blink recognition device according to another embodiment of the present disclosure.

Detailed ways

Specific embodiments of the present disclosure will be described in detail below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described here are only used to illustrate and explain the present disclosure, and are not intended to limit the present disclosure.

According to a first aspect of the present disclosure, an eye blink recognition method is provided. FIG. 1 is a flow chart of an eye blink recognition method according to an embodiment of the present disclosure. As shown in Figure 1, the method includes:

In step S11, among the voltage signals output by the body motion detection chip, the signals that meet the preset sampling conditions are sampled to obtain a sampled data sequence, wherein the body motion detection chip is used to detect eye blinking, and the human eye position It emits electromagnetic waves and detects the electromagnetic waves reflected by the human eye, and finally outputs the detection results in the form of voltage.

FIG. 2 is a schematic diagram of the voltage signal output by the body motion detection chip. As shown in Figure 2, the voltage signal is a continuously changing analog signal. When blinking occurs, the voltage changes to a higher voltage level, and when no blinking occurs, the voltage is maintained at a lower voltage level. In addition, the voltage suddenly rises from no blinking to the start of the blink, and suddenly drops when the blink ends, and the voltage does not change much until the next blink. Therefore, in this disclosure, in order to avoid the impact of non-blinking motion data on the algorithm and reduce the amount of subsequent data processing, when sampling, the voltage signals output by the body motion detection chip satisfy the preset sampling conditions. sampling. Wherein, the preset sampling condition is: a voltage change exceeds a preset voltage change threshold. Sampling is performed according to this condition, and sampling points that may correspond to eye blinking actions can be obtained, as shown in FIG. 3 . Wherein, the abscissa is the time t, and the ordinate is the voltage v corresponding to the sampling point. These sample points form a sequence of sampled data.

In step S12, any sampling point is selected from the untraversed sampling points as the target sampling point. For example, in the initial stage, since none of the sampling points has been traversed, a sampling point may be randomly selected from these sampling points as the target sampling point. In each subsequent round of the cycle, that is, after step S14 mentioned below, it is determined that there are still untraversed sampling points in the sampled data sequence and when returning to the step S12, it is also a never-traversed sampling point Randomly select a sampling point as the new target sampling point.

In step S13, when the target sampling point satisfies the preset validity condition of the blinking sampling point, a cluster for the target sampling point is created.

Exemplarily, the preset valid condition of the eye blink sampling point may include the following two types. The first type: In the first area formed with the target sampling point as the center and the preset radius parameter as the radius, the number of sampling points other than the target sampling point is not less than the preset domain density threshold.

Wherein, the preset radius parameter is set according to the processing results of a large amount of experimental data of blinking process. For example, the duration of a single blink is obtained by analyzing the experimental data, and the radius parameter is set according to the duration. For example, the radius parameter can be set in the following manner: with the target sampling point as the center, the duration corresponding to the area formed by the radius parameter (that is, the time difference between the maximum time and the minimum time corresponding to the area) is smaller than the single The duration of each blink.

In addition, in order to avoid the impact of non-blinking motion data on the algorithm in step S11 and reduce the amount of subsequent data processing, when sampling, the voltage signals output by the body motion detection chip satisfy the preset sampling conditions. Sampling, so that the density of sampling points corresponding to the blinking action (that is, the density of the sampling points) is significantly greater than the density of sampling points corresponding to the non-blinking action. Therefore, the field density threshold can be set according to the number of sampling points belonging to a single blink action obtained from the analysis of a large amount of experimental data. When the number of sampling points other than the target sampling point in the first area is not less than the field density threshold, it means that the sampling points in this area are relatively dense, so the currently selected target sampling point is likely to correspond to the blinking The sampling point of the action. At this time, a cluster can be created for the target sampling point, and the establishment of the cluster can be completed through subsequent clustering operations.

As shown in FIG. 3 , it is assumed that the target sampling point is P1, the preset radius parameter is r, and the preset domain density threshold is 4. The first area formed by drawing a circle with P1 as the center and r as the radius is shown as circle O1. The number of sampling points included in O1 other than the target sampling point P1 is 4, which satisfies the condition not less than the preset domain density threshold. Therefore, it is determined that the target sampling point P1 satisfies the above effective conditions. At this time, Clusters can be built for P1.

The second type: in the first area formed with the target sampling point as the center and the preset radius parameter as the radius, the number of sampling points other than the target sampling point is less than the preset domain density threshold In this case, the voltage value of the target sampling point is less than the nearest peak value and greater than the maximum idle peak value, and the voltage values of the sampling points in the first area except the target sampling point are all less than the peak value and greater than the specified peak value. The maximum peak idle time is described. Wherein, the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters. Two adjacent clusters mean that the time interval between two clusters does not exceed the preset inter-cluster time threshold. The preset inter-cluster interval time threshold may be determined based on experimental data, for example, based on the time interval between two adjacent blinks in the experiment. When the time interval between the two created clusters exceeds the inter-cluster interval time threshold, it indicates that there may be other clusters between the two clusters, and at this time, the two clusters are not two adjacent clusters. This maximum peak idle time is initially 0 before the cluster is established. In addition, during the process of creating a cluster, the maximum idle time peak value may be continuously updated according to the established cluster.

Wherein, during a complete blinking process, the voltage waveform output by the body motion detection chip is a waveform that rises from a lower level voltage to a peak value, and then drops to the lower level voltage. When the number of sampling points other than the target sampling point in the first area is less than the preset domain density threshold, and the voltage value of the target sampling point is smaller than the nearest peak value and greater than the maximum idle peak value, and the first When the voltage values of the sampling points in the area except the target sampling point are less than the peak value and greater than the maximum idle peak value, it indicates that the position of the target sampling point may be in the process of voltage rising or falling. Therefore, the currently selected target sampling point may also be a sampling point corresponding to the blinking action. At this time, a cluster can be created for the target sampling point, and the establishment of the cluster can be completed through subsequent clustering operations.

As shown in Figure 3, suppose the target sampling point is P2, the preset radius parameter is r, and the preset domain density threshold is 4. The first area formed by drawing a circle with P2 as the center and r as the radius is shown as circle O2. The number of sampling points included in O2 except the target sampling point P2 is 2, which is less than the preset field density threshold of 4. The voltage corresponding to point M is the peak value closest to the target sampling point P2, and the voltage value of each sampling point in O2 except the target sampling point P2 is also smaller than the peak value. At this time, it is assumed that the maximum idle peak value is The initial value is 0, the target sampling point P2 in O2, and the voltage value of each sampling point except the target sampling point P2 are greater than the maximum idle peak value. Therefore, it is determined that the target sampling point P2 satisfies the above valid conditions. At this time , a cluster for P2 can be established.

After the cluster for the target sampling point is created, next, it is judged whether other sampling points near the target sampling point can be clustered with the target sampling point and merged into one cluster. That is, as shown in FIG. 1, in step S14, the target sampling point and the sampling points satisfying the clustering condition between the target sampling point and the target sampling point are added to the cluster to complete the establishment of the cluster, wherein the The created cluster corresponds to a blink action.

When determining the sampling point satisfying the clustering condition between the target sampling point and the target sampling point, it can be determined in the manner shown in FIG. 4 . First, in step S41, the sampling points other than the target sampling point in the first area formed with the target sampling point as the center and the preset radius parameter as the radius are added to the set of candidate sampling points.

As shown in Figure 3, assuming that the target sampling point is P2, all the sampling points in area O2 except the target sampling point P2 are added to the candidate sampling point set S. At this time, the set S includes two sampling points, respectively q1 and q2.

In step S42, traverse the sampling points in the set of candidate sampling points, and divide the traversed sampling points in the second area formed with the traversed sampling points as the center and the preset radius parameter as the radius When the number of other sampling points is not less than the preset domain density threshold, the sampling points in the second area other than the traversed sampling points are added to the set of candidate sampling points. Wherein, the preset radius parameter and domain density threshold in the second area mentioned here are the same as the preset radius parameter and domain density threshold in the first area described above.

As shown in Figure 3, assuming that the sampling point in the traversed set of candidate sampling points S is q1, then determine the number of sampling points other than q1 in the circle O4 formed with q1 as the center and r as the radius Whether the number is not less than the preset domain density threshold. As shown in Figure 3, the number of sampling points in O4 other than q1 is 4, which is not less than the preset domain density threshold of 4. Therefore, it is determined that these sampling points meet the clustering conditions. At this time, add the sampling points in O4 except the q1 point to the set S of candidate sampling points. It is worth noting that the sampling points already included in the candidate sampling point set S are not added any more.

In step S43, it is judged whether all the sampling points in the candidate sampling point set have been traversed, and if all the sampling points in the candidate sampling point set have been traversed, go to step S44; If all the points have not been traversed, go to step S42 and continue to traverse until no new members are added to the candidate sampling point set, and all the sampling points in the candidate sampling point set have been traversed.

In step S44, the sampling points in the set of candidate sampling points that are not included in other clusters are determined as the sampling points satisfying the clustering condition with the target sampling point.

After the sampling points satisfying the clustering condition between the target sampling point and the target sampling point are determined, these sampling points are added to the cluster created for the target sampling point to complete the establishment of the cluster.

In the above technical solution, according to the preset field density threshold, the sampling points belonging to a blink action can be aggregated into the same cluster. In the clustering process, the main operations are to judge and mark these operations, and there is no need to calculate the distance between each sampling point in real time. Therefore, this density-based aggregation method can ensure the accuracy of data aggregation of sampling points at the same time , can also reduce the complexity of data processing and improve the efficiency of data processing, thereby improving the efficiency and accuracy of blink parameter identification.

After the clusters for the target sampling points selected in step S12 are established, it may be determined whether there are still untraversed sampling points in the sampling data sequence. When there are still untraversed sampling points in the sampling data sequence, return to step S12, and re-execute the selection of any sampling point from the untraversed sampling points as the target sampling point and subsequent operations until the sampling point in the sampling data sequence All sampling points have been traversed.

In step S15, blink parameters are identified based on the established clusters.

In the present disclosure, the blink parameters may include at least one of the following: number of blinks; single blink time; blink time interval; blink duty cycle. Correspondingly, said identifying blink parameters according to the established clusters may include at least one of the following:

1) Identify the total number of clusters created as the number of eye blinks. As mentioned above, each time a cluster is established, the cluster corresponds to an eye blink action, therefore, the total number of established clusters is the number of eye blinks.

2) For each cluster, identify the maximum time difference of the cluster members as a single blink time. Among them, the maximum time difference is the time interval between the earliest sampling point and the latest sampling point in the cluster, as shown in Figure 3, taking the cluster created by the target sampling point P1 as an example, in this cluster The earliest sampling point corresponds to time t1 , and the latest sampling point corresponds to time t1 ′, so in the blinking action corresponding to this cluster, the single blinking time is T1=t1 ′−t1 . For another example, taking the cluster created by the target sampling point P2 as an example, the earliest sampling point in the cluster corresponds to time t2, and the latest sampling point corresponds to time t2', then in the blink action corresponding to the cluster, the single blink time T2=t2'-t2. Wherein, T1 and T2 may be the same or different.

3) For two adjacent clusters, the difference between the adjacent time of the two adjacent clusters is recognized as the blink time interval, wherein the adjacent time of the two adjacent clusters refers to the time difference between the two adjacent clusters in the previous cluster. The time interval between the later sample point and the earliest sample point in the next cluster. As shown in Figure 3, assuming that the cluster corresponding to point P1 and the cluster corresponding to point P2 are two adjacent clusters, the time interval between t1' and t2 represents the blink time interval. Wherein, the blink time interval between every two blink actions may be the same or different.

4) Identify the ratio between the obtained sum of blink time intervals and the total time of the voltage signal as the blink duty cycle.

Through the above technical solution, the identification and clustering of the sampling points belonging to a blink can be realized, and the start and end of the blink can be accurately determined according to the clustering results, thereby accurately completing the identification of the blink parameters. In this process, it is not necessary to calculate the distance between sampling points in real time as in the existing blink recognition algorithm, but more to perform operations such as judgment and marking, which can greatly reduce the amount of data calculation and reduce the cost of the algorithm. The complexity can effectively avoid possible errors in the calculation process, and the accuracy rate is greatly improved.

Optionally, the method may further include: when the target sampling point selected in step S12 does not meet the preset effective condition of the blinking sampling point, it indicates that the target sampling point has deviated from the blinking signal waveform at this time, therefore, the The target sampling point is identified as a noise point, and the sampling points not included in other clusters in the first area formed with the target sampling point as the center and the preset radius parameter as the radius are also identified as noise points . Afterwards, the noise points are removed.

For example, as shown in FIG. 3 , it is assumed that the preset radius parameter is r, and the domain density threshold is 4. If the randomly selected target sampling point is P3, it is judged whether P3 satisfies the preset effective condition of the blinking sampling point. In the circle O3 formed with P3 as the center and r as the radius, the number of sampling points other than the target sampling point P3 included is 1, which is less than the preset domain density threshold of 4. And at this time, the maximum idle peak value has been updated to the voltage value corresponding to point C, and the voltage values of the sampling points contained in area O3 do not meet the conditions that all of them are greater than the voltage value corresponding to point C, so it is determined that point P3 and area O3 Among the included sampling points, the members not included in other clusters are noise points, and the noise points are removed. Thus, on the one hand, the impact of noise points on eye blink recognition can be avoided, on the other hand, the number of sampling points to be processed can be reduced, and the processing amount of sampling point data can be reduced, thereby improving the recognition efficiency of eye blinking actions.

In addition, before step S15, the method may further include: deleting members in each cluster whose voltage value is smaller than the maximum idle peak value.

For example, as shown in FIG. 3 , after the clusters created with P1 and P2 as target sampling points are established, the maximum idle peak value is updated from an initial value of 0 to a voltage value corresponding to point C. Point q2 is a member of the cluster established for the target sampling point P2, however, the voltage value of the sampling point q2 is less than the corresponding voltage value of point C, therefore, delete q2 from the cluster established for the target sampling point P2 to update cluster members. Afterwards, blink parameters are identified based on the updated cluster membership.

Through this technical solution, the integrity of the blink process can be ensured without redundant boundary points, thereby improving the accuracy of blink parameter recognition.

According to a second aspect of the present disclosure, an eye blink recognition device is provided. FIG. 5 is a block diagram of an eye blink recognition device according to an embodiment of the present disclosure. As shown in Figure 5, the device 10 includes:

The sampling module 101 is configured to sample a signal that satisfies a preset sampling condition among the voltage signals output by the body motion detection chip, and obtain a sampled data sequence, wherein the body motion detection chip is used to detect eye blinking;

A target sampling point selection module 102, configured to select any sampling point as a target sampling point from the sampling points obtained by the sampling module 101 and not traversed;

A cluster creation module 103, configured to create a cluster for the target sampling point when the target sampling point selected by the target sampling point selection module 102 satisfies the preset effective condition of the eye blink sampling point;

The cluster establishment module 104 is used to add the target sampling point selected by the target sampling point selection module 102 and the sampling points satisfying the clustering condition between the target sampling point and the target sampling point to the cluster creation module 103. In the cluster, to complete the establishment of the cluster, wherein the established cluster corresponds to an eye blinking action; the cluster establishment module 104 is also used to re-trigger the target sampling point selection module 102 to execute from the sampling module 101 Selecting any sampling point from the obtained, untraversed sampling points as the target sampling point until all the sampling points in the sampling data sequence have been traversed;

The parameter identification module 105 is configured to identify blink parameters according to the clusters established by the cluster establishment module 104 .

Optionally, the preset sampling conditions are:

The change in voltage exceeds a preset voltage change threshold.

Optionally, the preset effective conditions for blinking sampling points include:

In the first area formed with the target sampling point as the center and a preset radius parameter as the radius, the number of sampling points other than the target sampling point is not less than the preset domain density threshold; or

In the first area formed with the target sampling point as the center and a preset radius parameter as the radius, the number of sampling points other than the target sampling point is less than the preset domain density threshold Next, the voltage value of the target sampling point is less than the nearest peak value and greater than the maximum idle peak value, and the voltage values of the sampling points in the first area except the target sampling point are all less than the peak value and greater than the maximum idle peak value, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters.

Optionally, the device 10 may also include:

A noise point identification module, configured to identify the target sampling point as a noise point when the target sampling point selected by the target sampling point selection module 102 does not meet the preset effective condition of the eye blink sampling point, and Identifying sampling points in the first region that are not included in other clusters as noise points;

A noise point removal module, configured to remove the noise points identified by the noise point identification module.

Optionally, as shown in FIG. 6 , it is a block diagram of the cluster building module 104 in the eye blink recognition device according to another embodiment of the present disclosure. As shown in Figure 6, the cluster building module 104 includes:

The candidate sampling point set establishment sub-module 1041 is used to divide the target sampling point selected by the target sampling point selection module 102 into the first area formed with the preset radius parameter as the radius, and divide the Sampling points other than the target sampling point are added to the set of candidate sampling points;

The candidate sampling point traversal sub-module 1042 is configured to traverse the sampling points in the candidate sampling point set, and in the second area formed with the traversed sampling point as the center and the preset radius parameter as the radius, except When the number of sampling points other than the traversed sampling points is not less than the preset field density threshold, add the sampling points in the second area except the traversed sampling points to the set of candidate sampling points , until all sampling points in the set of candidate sampling points have been traversed;

The cluster establishment sub-module 1043 is configured to determine the sampling points in the set of candidate sampling points that are not included in other clusters as the sampling points satisfying the clustering condition with the target sampling point.

Optionally, the parameter identification module 105 includes at least one of the following:

The number of blinks identification submodule is used to identify the total number of clusters established by the cluster building module 104 as the number of blinks;

The single blink time identification submodule is used for each cluster established by the cluster establishment module 104, identifying the maximum time difference of cluster members as the single blink time;

The blink time interval identification submodule is used to identify the difference between the adjacent time of the two adjacent clusters as the blink time interval for the two adjacent clusters established by the cluster establishment module 104;

The blink duty cycle identification submodule is configured to identify the ratio between the sum of the blink time intervals identified by the blink time interval identification submodule and the total time of the voltage signal as the blink duty cycle.

Optionally, the device 10 may also include:

The deletion module is used to delete the members whose voltage value is less than the maximum idle peak value in each cluster before the parameter identification module 105 identifies the blink parameter according to the cluster established by the cluster establishment module 104, wherein the maximum The peak value when idle is the maximum value among non-cluster members between two adjacent clusters.

Through the above technical solution, the identification and clustering of the sampling points belonging to a blink can be realized, and the start and end of the blink can be accurately determined according to the clustering results, thereby accurately completing the identification of the blink parameters. In this process, it is not necessary to calculate the distance between sampling points in real time as in the existing blink recognition algorithm, but more to perform operations such as judgment and marking, which can greatly reduce the amount of data calculation and reduce the cost of the algorithm. The complexity can effectively avoid possible errors in the calculation process, and the accuracy rate is greatly improved.

Regarding the apparatus in the foregoing embodiments, the specific manner in which each module executes operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

The preferred embodiments of the present disclosure have been described in detail above in conjunction with the accompanying drawings. However, the present disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present disclosure, various simple modifications can be made to the technical solutions of the present disclosure. These simple modifications all belong to the protection scope of the present disclosure.

In addition, it should be noted that the various specific technical features described in the above specific implementation manners may be combined in any suitable manner if there is no contradiction. In order to avoid unnecessary repetition, various possible combinations are not further described in this disclosure.

In addition, various implementations of the present disclosure can also be combined in any way, as long as they do not violate the idea of the present disclosure, they should also be regarded as the content disclosed in the present disclosure.

Claims (14)

1. A method for eye blink recognition, said method comprising: sampling a signal satisfying a preset sampling condition in a voltage signal output by a body motion detection chip, and obtaining a sampled data sequence, wherein said body motion detection chip uses For detecting blinking action, it is characterized in that, described method also comprises: Select any sampling point from the untraversed sampling points as the target sampling point; Create a cluster for the target sampling point when the target sampling point satisfies the preset effective condition of the eye blink sampling point; Adding the target sampling point and the sampling points satisfying the clustering condition between the target sampling point and the target sampling point to the cluster to complete the establishment of the cluster, wherein the established cluster corresponds to an eye blink; Repeating the step of selecting any sampling point from the untraversed sampling points as the target sampling point until all the sampling points in the sampled data sequence have been traversed; Based on the established clusters, blink parameters are identified. 2. method according to claim 1, is characterized in that, described preset sampling condition is: The change in voltage exceeds a preset voltage change threshold. 3. method according to claim 1, is characterized in that, described preset blink sampling point effective condition comprises: In the first area formed with the target sampling point as the center and a preset radius parameter as the radius, the number of sampling points other than the target sampling point is not less than the preset domain density threshold; or In the first area formed with the target sampling point as the center and a preset radius parameter as the radius, the number of sampling points other than the target sampling point is less than the preset domain density threshold Next, the voltage value of the target sampling point is less than the nearest peak value and greater than the maximum idle peak value, and the voltage values of the sampling points in the first area except the target sampling point are all less than the peak value and greater than the maximum idle peak value, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters. 4. method according to claim 3, is characterized in that, described method also comprises: When the target sampling point does not meet the preset effective condition of blinking sampling point, identify the target sampling point as a noise point, and select the sampling points in the first area that are not included in other clusters identified as noise points; remove the noise points. 5. method according to claim 1, is characterized in that, determine and satisfy the sampling point of clustering condition between described target sampling point by the following manner: Adding sampling points other than the target sampling point in the first area formed with the target sampling point as the center and a preset radius parameter as the radius to the set of candidate sampling points; Repeatedly traversing the sampling points in the set of candidate sampling points, and in the second area formed with the traversed sampling point as the center and the preset radius parameter as the radius, all but the traversed sampling points When the number of sampling points is not less than the preset domain density threshold, the step of adding sampling points in the second area other than the traversed sampling points to the set of candidate sampling points until the candidate Until all the sampling points in the sampling point set have been traversed; Determining the sampling points in the set of candidate sampling points that are not included in other clusters as the sampling points satisfying the clustering condition with the target sampling point. 6. The method according to any one of claims 1-5, wherein the identification of blink parameters according to the established cluster includes at least one of the following: identifying the total number of clusters created as the number of eye blinks; For each cluster, identify the maximum time difference of the cluster members as being a single blink time; For two adjacent clusters, the difference between the adjacent time of the two adjacent clusters is identified as the blink time interval; The ratio between the sum of the blink time intervals and the total time of the voltage signal is identified as the blink duty cycle. 7. The method according to claim 6, wherein, before the step of identifying blink parameters according to the established cluster, the method further comprises: Deleting members whose voltage value is less than the maximum idle peak value in each cluster, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters. 8. An eye blink recognition device, characterized in that the device comprises: The sampling module is used to sample the signal satisfying the preset sampling condition in the voltage signal output by the body motion detection chip, and obtain the sampling data sequence, wherein the body motion detection chip is used to detect eye blinking; A target sampling point selection module, configured to select any sampling point from the untraversed sampling points obtained by the sampling module as the target sampling point; A cluster creation module, configured to create a cluster for the target sampling point when the target sampling point selected by the target sampling point selection module satisfies the preset effective condition of the blinking sampling point; A cluster establishment module, configured to add the target sampling point selected by the target sampling point selection module and the sampling points satisfying the clustering condition between the target sampling point and the target sampling point to the cluster created by the cluster creation module , to complete the establishment of the cluster, wherein the established cluster corresponds to a blinking action; The cluster establishment module is also used to re-trigger the target sampling point selection module to select any sampling point from the untraversed sampling points acquired by the sampling module as the target sampling point until the sampling point in the sampling data sequence All sampling points have been traversed; The parameter identification module is used to identify blink parameters according to the clusters established by the cluster establishment module. 9. The device according to claim 8, wherein the preset sampling condition is: The change in voltage exceeds a preset voltage change threshold. 10. The device according to claim 8, characterized in that, the preset valid conditions for blinking sampling points include: In the first area formed with the target sampling point as the center and a preset radius parameter as the radius, the number of sampling points other than the target sampling point is not less than the preset domain density threshold; or In the first area formed with the target sampling point as the center and a preset radius parameter as the radius, the number of sampling points other than the target sampling point is less than the preset domain density threshold Next, the voltage value of the target sampling point is less than the nearest peak value and greater than the maximum idle peak value, and the voltage values of the sampling points in the first area except the target sampling point are all less than the peak value and greater than the maximum idle peak value, wherein the maximum idle peak value is the maximum value among non-cluster members between two adjacent clusters. 11. The device according to claim 10, further comprising: A noise point identification module, configured to identify the target sampling point as a noise point when the target sampling point selected by the target sampling point selection module does not meet the preset effective condition of the eye blink sampling point, and Sampling points not included in other clusters in the first region are identified as noise points; A noise point removal module, configured to remove the noise points identified by the noise point identification module. 12. The device according to claim 8, wherein the cluster building module comprises: The candidate sampling point set building sub-module is used to divide the target sampling point from the first area formed by taking the target sampling point selected by the target sampling point selection module as the center of the circle and taking the preset radius parameter as the radius. The sampling points other than the sampling points are added to the set of candidate sampling points; The candidate sampling point traversal submodule is used to traverse the sampling points in the candidate sampling point set, and divide all When the number of sampling points other than the traversed sampling points is not less than the preset field density threshold, add the sampling points in the second area except the traversed sampling points to the set of candidate sampling points , until all sampling points in the set of candidate sampling points have been traversed; The cluster establishment submodule is configured to determine the sampling points in the set of candidate sampling points that are not included in other clusters as the sampling points satisfying the clustering condition with the target sampling point. 13. The device according to any one of claims 8-12, wherein the parameter identification module includes at least one of the following: The number of blinks identification submodule is used to identify the total number of clusters established by the cluster building module as the number of blinks; The single blink time identification submodule is used to identify the maximum time difference of cluster members as the single blink time for each cluster established by the cluster establishment module; The blink time interval identification submodule is used to identify the difference between the adjacent time of the two adjacent clusters as the blink time interval for the two adjacent clusters established by the cluster establishment module; The blink duty cycle identification submodule is configured to identify the ratio between the sum of the blink time intervals identified by the blink time interval identification submodule and the total time of the voltage signal as the blink duty cycle. 14. The device according to claim 13, further comprising: A deletion module, configured to delete, before the parameter identification module identifies blink parameters according to the clusters established by the cluster establishment module, delete members in each cluster whose voltage value is less than the peak value of the maximum idle time, wherein the maximum idle time The peak is the maximum value among non-cluster members between two adjacent clusters.