Eyeball tracking process positioning optimization method and system

Abstract

The invention relates to an eyeball tracking process positioning optimization method and system, and belongs to the technical field of eyeball recognition. Specifically, a pupil center point is determined through an image preprocessing and visual analysis technology, then a spatial positioning grid graph is constructed, and the direction of the pupil center point is associated with a positioning grid; screening effective condition factors related to pupil center point pointing change height, and analyzing time sequence change characteristics of the effective condition factors; establishing an incidence relation library of the reference condition factor change feature group and the reference positioning grid mapping change; and performing dual feature retrieval on the real-time condition factor and the real-time positioning grid mapping change by using the relation library, selecting a reference positioning grid mapping change for correction, and realizing optimization correction on the real-time pupil center pointing change, and the eyeball tracking accuracy and stability are improved through the technical scheme.

Description

Technical Field

[0001] The invention relates to an eye tracking process positioning optimization method and system, belonging to the technical field of eye recognition. Background Art

[0002] In the field of eye tracking technology, achieving accurate and stable positioning has always been a challenging technical problem. Traditional eye tracking methods often rely on a single image processing technology, which obtains eye dynamic video data and performs image preprocessing to determine the pupil center point, thereby achieving eye tracking. However, this method is often affected by multiple factors such as the environment, lighting, and individual differences in actual applications, resulting in inaccurate positioning and unstable tracking.

[0003] Therefore, it is urgent to design an eye tracking method and system that can solve the above technical problems. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system capable of optimizing eye tracking positioning.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: One of the purposes of the present invention is to provide a method for optimizing positioning during eye tracking, which is characterized in that it comprises the following steps: Acquire eye dynamic video data, perform image preprocessing on each frame of the eye dynamic video data, and use visual analysis technology to determine the pupil center point; According to the eyeball positioning requirements, a spatial positioning grid is constructed. Each positioning grid on the spatial positioning grid is set with a specific spatial coordinate interval, and each direction of the pupil center is associated with the positioning grid to construct a pupil center direction ~ positioning grid correspondence table; Acquire several situation factors corresponding to the eyeball positioning requirement, the situation factors are correlated with the change in the direction of the pupil center point, and select several valid situation factors whose correlation with the change in the direction of the pupil center point is greater than or equal to a preset value; Combine the effective situation factors belonging to the same time node to form a single time node effective situation factor group. Based on the time series relationship, analyze the factor change characteristics of several consecutive single time node effective situation factor groups to form a reference situation factor change characteristic group. Determine the reference positioning grid mapping change of the spatial positioning grid map corresponding to the reference situation factor change feature group, and establish an association relationship with the reference situation factor change feature group to form a reference situation factor change feature group~reference positioning grid mapping change relationship library; In the process of eye tracking, the real-time situation factor group is determined in real time, and the real-time positioning grid mapping change is determined. The reference situation factor change feature group~reference positioning grid mapping change relationship library is used to perform dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change, and several reference positioning grid mapping changes for correcting the real-time grid mapping changes are selected. Based on the determined reference positioning grid mapping changes, several reference pupil center point pointing changes are formed, and the reference pupil center point pointing changes are used to optimize and correct the real-time pupil center pointing changes.

[0006] Preferably, the method steps for determining the center point of the exit pupil include: De-noising each frame of the eye dynamic video data to obtain a denoised image, graying the denoised image to obtain a grayscale image, and enhancing the contrast of the grayscale image; The pupil edge in the grayscale image is determined using an edge detection algorithm, and the pupil center is determined based on the determined pupil edge.

[0007] Preferably, the method of screening out a number of valid situation factors whose correlation with the change in the orientation of the pupil center point is greater than or equal to a preset value comprises: Parameterize the situation factor, and associate the situation factor parameter with the positioning grid pointed to by the pupil center point; Classifying the situation factor parameters whose differences are less than or equal to the preset value to form several categories of situation factor parameter sets, analyzing the relative position characteristics of the positioning grids associated with each category of the situation factor parameter set to obtain the first related parameter; A comparison relationship is established between category situation factor sets with the same situation factor category, cluster position characteristics of the positioning grids corresponding to each category situation factor set are compared, a cluster crossover ratio between the two is determined, and a second related parameter is determined based on the cluster crossover ratio; Based on the first correlation parameter and the second correlation parameter, the correlation degree between the situation factor and the change in the direction of the pupil center point is determined.

[0008] Preferably, the expression for the correlation between the situation factor and the change in the orientation of the pupil center point is: ; Among them, G is the degree of correlation, is the correlation conversion coefficient, is the straight-line distance of the ith positioning square corresponding to the category situation factor parameter set relative to the central positioning square. The method for determining the central positioning square includes calculating the average coordinates of all positioning squares corresponding to the category situation factor parameter set, and identifying the positioning square closest to the average coordinates as the central positioning square. n is the total number of all positioning squares corresponding to the category situation factor parameter set. is the weight adjustment coefficient affecting the relative position between squares, is the positioning grid coincidence judgment function. When comparing the cluster position features of the positioning grids corresponding to the category situation factor sets, if the xth positioning grid coincides, Output 1, otherwise output 0, N is the total number of positioning squares corresponding to the category situation factor set participating in the comparison, is the cluster cross-proportion impact weight adjustment coefficient, Adjust the constant for the degree of correlation.

[0009] Preferably, the method for analyzing the factor change characteristics of a plurality of consecutive single time node effective situation factor groups includes: For the time series, a time axis is established, and a situation factor parameter axis is established perpendicular to the time axis, and the effective situation factor parameters corresponding to each time node are arranged between the time axis and the situation factor parameter axis to form a number of effective situation factor mapping points; Perform linear fitting on a number of effective situation factor mapping points to form an effective situation factor mapping curve; A combination of several effective situation factor mapping curves corresponding to the effective situation factor group is identified as a reference situation factor change characteristic group.

[0010] In some embodiments disclosed in the present invention, a method for establishing an association relationship between a reference positioning grid mapping change and a reference situation factor change feature group includes: A number of associated time segments are constructed for the reference situation factor change feature group and the reference positioning grid mapping change. The first and last nodes of each associated time segment are determined by analyzing the corresponding effective situation factor group of each time node of the reference situation factor change feature group. If the difference in the change of the effective situation factor parameter in the effective situation factor group relative to the previous time node is greater than or equal to a preset value, the corresponding time node is identified as the first and last nodes of the associated time segment. According to the associated time segment, the reference situation factor change feature group and the reference positioning grid mapping change are segmented and the segmented segments are associated.

[0011] Preferably, the method for performing dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change comprises: Acquire real-time situation factor parameters, and analyze the time-varying characteristics of each real-time situation factor parameter to form a real-time situation factor change characteristic group, synchronously acquire real-time positioning grid mapping changes, and segment the real-time positioning grid mapping changes and the real-time situation factor change characteristic group according to associated time segments, and correlate the segmented segments; Extracting retrieval tags from mutually related real-time positioning grid mapping changes and real-time situation factor change feature groups, and using the extracted retrieval tags to search in the reference situation factor change feature group~reference positioning grid mapping change relationship library, and performing double feature retrieval on the reference positioning grid mapping changes and reference situation factor change feature groups retrieved initially, including first difference feature analysis between situation factor change feature groups, and second difference feature analysis between positioning grid mapping changes; Based on the first difference feature analysis and the second difference feature analysis, a reference positioning grid mapping change finally retrieved is determined.

[0012] Preferably, the method for determining the first difference characteristic between the situation factor change characteristic groups comprises: Compare the difference amounts of situation factor parameters at different time nodes between situation factor change feature groups, and mark the situation factors whose situation factor parameter difference amounts are greater than or equal to a preset value as difference situation factors, count the number of difference situation factors in real time, and if the number of difference situation factors is greater than or equal to a preset value, it is determined that the first difference feature between the situation factor change feature groups does not meet the preset standard, and the corresponding reference situation factor change feature group and reference positioning grid mapping change are eliminated; The distances between the positioning grids mapped at different time nodes are compared between the positioning grid mapping changes. If the distances between the positioning grids do not meet the preset standard, the corresponding time nodes are marked as grid difference time nodes, and the number of grid difference time nodes is counted in real time. If it is greater than or equal to the preset value, it is determined that the second difference feature does not meet the standard, and the corresponding reference situation factor change feature group and reference positioning grid mapping changes are eliminated.

[0013] The second object of the present invention is to provide an eye tracking process positioning optimization system, which is particularly characterized by comprising: The first module is used to obtain eye dynamic video data, perform image preprocessing on each frame of the eye dynamic video data, and determine the pupil center point using visual analysis technology; The second module is used to construct a spatial positioning grid map according to the eye positioning needs. Each positioning grid on the spatial positioning grid map is set with a specific spatial coordinate interval, and each direction of the pupil center point is associated with the positioning grid to construct a pupil center point direction ~ positioning grid correspondence table; The third module is used to obtain several situation factors corresponding to the eye positioning requirements, the situation factors are correlated with the change in the direction of the pupil center point, and screen out several valid situation factors whose correlation with the change in the direction of the pupil center point is greater than or equal to a preset value; The fourth module is used to combine the effective situation factors belonging to the same time node to form a single time node effective situation factor group, and analyze the factor change characteristics of several consecutive single time node effective situation factor groups based on the time series relationship to form a reference situation factor change characteristic group; The fifth module is used to determine the reference positioning grid mapping change of the spatial positioning grid map corresponding to the reference situation factor change feature group, and establish an association relationship with the reference situation factor change feature group to form a reference situation factor change feature group~reference positioning grid mapping change relationship library; The sixth module is used to determine the real-time situation factor group and the real-time positioning grid mapping change in real time during the eye tracking process, and use the reference situation factor change feature group~reference positioning grid mapping change relationship library to perform dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change, select a number of reference positioning grid mapping changes for correcting the real-time grid mapping changes, and based on the determined reference positioning grid mapping changes, form a number of reference pupil center point pointing changes, and use the reference pupil center point pointing changes to optimize and correct the real-time pupil center pointing changes.

[0014] The invention discloses an eye tracking process positioning optimization method and system, and relates to the technical field of eye recognition. Specifically, the invention discloses determining the pupil center point by image preprocessing and visual analysis technology, then constructing a spatial positioning grid diagram, and associating the orientation of the pupil center point with the positioning grid; screening effective situation factors highly correlated with the pupil center point orientation change, and analyzing the time series change characteristics thereof; establishing an association relationship library of a reference situation factor change feature group and a reference positioning grid mapping change; utilizing the relationship library to perform dual feature retrieval on real-time situation factors and real-time positioning grid mapping changes, selecting reference positioning grid mapping changes for correction, and realizing optimization correction of real-time pupil center orientation changes. The invention improves the accuracy and stability of eye tracking through the above technical scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a method step diagram of an eye tracking process positioning optimization method of the present invention. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0017] The purpose of the present invention is to provide a method and system capable of optimizing eye tracking positioning.

[0018] In order to achieve the above object, the present invention adopts the following technical solutions: A method for optimizing eye tracking positioning, see Figure 1 ,include: Step S100, acquiring eye dynamic video data, performing image preprocessing on each frame of the eye dynamic video data, and determining the pupil center point using visual analysis technology.

[0019] This step is the starting stage of the positioning optimization of the eye tracking process. The main task is to obtain the dynamic video data of the eye and preprocess the data. The purpose of preprocessing is to improve the image quality and make it more suitable for subsequent analysis and processing. Using visual analysis technology, the center of the pupil is determined from the preprocessed image, which is the basis of eye tracking, because the position change of the center of the pupil can reflect the movement trajectory of the eye.

[0020] In some embodiments disclosed in the present invention, the method steps for determining the center point of the exit pupil include: Step S101, denoising each frame of the eye dynamic video data to obtain a denoised image, graying the denoised image to obtain a grayed image, and enhancing the contrast of the grayed image.

[0021] Step 102: using an edge detection algorithm to determine the pupil edge in the grayscale image, and based on the determined pupil edge, determining the pupil center point.

[0022] Step S200, constructing a spatial positioning grid diagram according to the eye positioning requirement, each positioning grid on the spatial positioning grid diagram is set with a specific spatial coordinate interval, and each direction of the pupil center point is associated with the positioning grid, and a pupil center point direction~positioning grid correspondence table is constructed.

[0023] After the pupil center is determined, this step constructs a spatial positioning grid diagram according to the needs of eyeball positioning. Each positioning grid on this grid diagram is set with a specific spatial coordinate interval to represent the possible position of the eyeball. Then, each direction of the pupil center point is associated with the positioning grid, and a correspondence table between the pupil center point direction and the positioning grid is constructed. In this way, the position change of the eyeball in space can be inferred by the position change of the pupil center point.

[0024] Step S300, obtaining several situation factors corresponding to the eye positioning requirement, the situation factors are correlated with the change in the direction of the pupil center point, and screening out several valid situation factors whose correlation with the change in the direction of the pupil center point is greater than or equal to a preset value.

[0025] The goal of this step is to find out the situation factors that are correlated with the change in the direction of the pupil center. The situation factors here refer to some situational elements that may affect the movement of the eyeball in reality, such as the density of vehicles and pedestrians in the external environment, the complexity of the road conditions, and the angle of the next turn required in the task requirements of eye recognition. By screening out several valid situation factors whose correlation with the change in the direction of the pupil center is greater than or equal to the preset value, we can further understand the law of eyeball movement and provide a basis for subsequent eye tracking optimization.

[0026] In order to explain the corresponding content of the situation factors more clearly, the following are examples of several situation factors, but it should be noted that the corresponding content of the situation factors is not limited to the following examples. Traffic light changes in driving tasks: During driving, changes in traffic lights will directly affect the driver's eye movement. For example, when the red light is on, the driver needs to turn his eyes to the light and wait for the green light to come on. In this process, the eye movement is closely related to the change of traffic lights. Text layout in reading tasks: When performing reading tasks, the layout of the text will also affect the eye movement. For example, if the text is densely laid out and the line spacing is small, the reader's eye may need to move more frequently to track the text. On the contrary, if the text is loosely laid out and the line spacing is large, the reader's eye movement may be relatively reduced. Game situation when watching sports events: When watching sports events, changes in the game situation will directly affect the audience's eye movement. For example, when the game enters a critical moment, such as shooting, dunking, etc., the audience's eyes may be more concentrated on the key areas of the game field. The position of the target in the game task: When playing the game task, the position of the target will also affect the player's eye movement. For example, in a shooting game, when the enemy suddenly appears in a corner of the screen, the player's eyes need to move quickly to lock the target and shoot. Ingredient preparation in cooking tasks: When performing cooking tasks, the preparation process of ingredients will also affect the cook's eye movement. For example, when the cook needs to add ingredients in a specific order, their eyes may move in an orderly manner according to the placement of the ingredients.

[0027] In some embodiments disclosed in the present invention, a method for screening out a number of valid situation factors whose correlation degree with the change in the orientation of the pupil center point is greater than or equal to a preset value includes: Step S301, parameterize the situation factor, and associate the situation factor parameter with the positioning grid pointed to by the pupil center point.

[0028] First, the specific definition and possible range of variation of each situation factor need to be clarified. For example, in the driver eye tracking recognition task, situation factors may include the density of vehicles and pedestrians in the external environment, the complexity of road conditions, turning angles, etc. These factors need to be specifically defined. For example, the density of vehicles and pedestrians can be measured by the number of vehicles and pedestrians per unit area, and the complexity of road conditions can be comprehensively evaluated by indicators such as road type (such as highways, urban streets, rural roads), number of intersections, and frequency of lane changes.

[0029] Step S302, classify the situation factor parameters whose differences are less than or equal to the preset value to form several categories of situation factor parameter sets, analyze the relative position characteristics of the positioning grids associated with each category of the situation factor parameter set, and obtain the first related parameter.

[0030] Step S303, establishing a comparison relationship between category situation factor sets with the same situation factor category, comparing the cluster position characteristics of the positioning grids corresponding to the category situation factor sets, determining the cluster crossover ratio between the two, and determining the second related parameter based on the cluster crossover ratio.

[0031] Step S304: Determine the correlation degree between the situation factor and the change in the orientation of the pupil center point based on the first correlation parameter and the second correlation parameter.

[0032] In some embodiments disclosed in the present invention, the expression for calculating the correlation between the situation factor and the change in the orientation of the pupil center point is: .

[0033] Among them, G is the degree of correlation, is the correlation conversion coefficient, is the straight-line distance of the ith positioning square corresponding to the category situation factor parameter set relative to the central positioning square. The method for determining the central positioning square includes calculating the average coordinates of all positioning squares corresponding to the category situation factor parameter set, and identifying the positioning square closest to the average coordinates as the central positioning square. n is the total number of all positioning squares corresponding to the category situation factor parameter set. is the weight adjustment coefficient affecting the relative position between squares, is the positioning grid coincidence judgment function. When comparing the cluster position features of the positioning grids corresponding to the category situation factor sets, if the xth positioning grid coincides, Output 1, otherwise output 0, N is the total number of positioning squares corresponding to the category situation factor set participating in the comparison, is the cluster cross-proportion impact weight adjustment coefficient, Adjust the constant for the degree of correlation.

[0034] Step S400, combining the effective situation factors belonging to the same time node to form a single time node effective situation factor group, and analyzing the factor change characteristics of several consecutive single time node effective situation factor groups based on the time series relationship to form a reference situation factor change characteristic group.

[0035] In this step, the effective situation factors belonging to the same time node are combined to form a single time node effective situation factor group. Then, based on the time series relationship, the factor change characteristics of several consecutive single time node effective situation factor groups are analyzed to form a reference situation factor change characteristic group. The purpose of this step is to find out the law of situation factor changes over time, so as to better understand and predict eye movement, especially how the real situation elements related to eye movement affect eye movement.

[0036] In some embodiments disclosed in the present invention, a method for analyzing factor change characteristics of a plurality of consecutive single-time node effective situation factor groups includes: Step S401, for the time series, establish a time axis, and establish a situation factor parameter axis perpendicular to the time axis, and configure the effective situation factor parameters corresponding to each time node between the time axis and the situation factor parameter axis to form a number of effective situation factor mapping points.

[0037] Step S402: linearly fit a number of effective situation factor mapping points to form an effective situation factor mapping curve.

[0038] Step S403: identifying a combination of several valid situation factor mapping curves corresponding to the valid situation factor group as a reference situation factor change characteristic group.

[0039] Step S500, determining the reference positioning grid mapping change of the spatial positioning grid diagram corresponding to the reference situation factor change feature group, and establishing an association relationship with the reference situation factor change feature group to form a reference situation factor change feature group~reference positioning grid mapping change relationship library.

[0040] The purpose of this step is to determine the reference positioning grid mapping changes of the spatial positioning grid corresponding to the reference situation factor change feature group. By establishing an association relationship with the reference situation factor change feature group, a relationship library of reference situation factor change feature groups and reference positioning grid mapping changes can be formed. This relationship library can be used in the subsequent eye tracking process to optimize the eye tracking effect according to the real-time situation factors and positioning grid mapping changes, especially considering the real situation elements that affect eye movement.

[0041] In some embodiments disclosed in the present invention, a method for establishing an association relationship between a reference positioning grid mapping change and a reference situation factor change feature group includes: Step S501, constructing a number of associated time segments for reference situation factor change feature groups and reference positioning grid mapping changes, the method for determining the first and last nodes of each associated time segment includes analyzing the corresponding valid situation factor group of each time node of the reference situation factor change feature group, if there is a valid situation factor parameter in the valid situation factor group whose change difference relative to the previous time node is greater than or equal to a preset value, then the corresponding time node is identified as the first and last nodes of the associated time segment.

[0042] Step S502 , segment-cutting the reference situation factor change feature group and the reference positioning grid mapping change according to the associated time segment, and associating the segment-cutting segments.

[0043] Step S600, during the eye tracking process, a real-time situation factor group is determined in real time, and a real-time positioning grid mapping change is determined, and a reference situation factor change feature group ~ reference positioning grid mapping change relationship library is used to perform dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change, and a number of reference positioning grid mapping changes for correcting the real-time grid mapping changes are selected, and based on the determined reference positioning grid mapping changes, a number of reference pupil center point pointing changes are formed, and the reference pupil center point pointing changes are used to optimize and correct the real-time pupil center pointing changes.

[0044] This is the last step of the positioning optimization of the eye tracking process. In this step, the real-time situation factor group and the real-time positioning grid mapping change are determined in real time, and the reference situation factor change feature group and the reference positioning grid mapping change relationship library established previously are used for dual feature retrieval. Through the retrieval, several reference positioning grid mapping changes for correcting the real-time grid mapping changes can be selected. Then, based on these selected reference positioning grid mapping changes, several reference pupil center point pointing changes are formed, and these reference changes are used to optimize and correct the real-time pupil center pointing changes. In this way, a more accurate and stable eye tracking effect can be achieved, especially when considering the situation elements that may affect the eye movement in reality.

[0045] In some embodiments disclosed in the present invention, the method for performing dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change includes: Step S601, obtain the real-time situation factor parameters, and analyze the temporal change characteristics of each real-time situation factor parameter to form a real-time situation factor change feature group, synchronously obtain the real-time positioning grid mapping changes, and segment the real-time positioning grid mapping changes and the real-time situation factor change feature group according to the associated time segments, and correlate the segmented segments.

[0046] Step S602, extracting retrieval tags for mutually related real-time positioning grid mapping changes and real-time situation factor change feature groups, and using the extracted retrieval tags to search in the reference situation factor change feature group~reference positioning grid mapping change relationship library, and performing double feature retrieval on the reference positioning grid mapping changes and reference situation factor change feature groups retrieved initially, including a first difference feature analysis between the situation factor change feature groups, and a second difference feature analysis between the positioning grid mapping changes.

[0047] In some embodiments disclosed in the present invention, a method for determining a first difference feature between situation factor change feature groups includes: Step S6021, compare the difference in situation factor parameters at different time nodes between situation factor change feature groups, and mark the situation factors whose situation factor parameter difference is greater than or equal to the preset value as difference situation factors, and count the number of difference situation factors in real time. If the number of difference situation factors is greater than or equal to the preset value, it is determined that the first difference feature between the situation factor change feature groups does not meet the preset standard, and the corresponding reference situation factor change feature group and reference positioning grid mapping changes are eliminated.

[0048] Step S6022, compare the distances between the positioning grids mapped at different time nodes between the positioning grid mapping changes. If the distances between the positioning grids do not meet the preset standard, mark the corresponding time nodes as grid difference time nodes, and count the number of grid difference time nodes in real time. If it is greater than or equal to the preset value, it is determined that the second difference feature does not meet the standard, and the corresponding reference situation factor change feature group and reference positioning grid mapping changes are eliminated.

[0049] Step S603: Determine the mapping change of the reference positioning grid finally retrieved based on the first difference feature analysis and the second difference feature analysis.

[0050] In some embodiments disclosed in the present invention, there is also disclosed an eye tracking process positioning optimization system, comprising: The first module is used to obtain eye dynamic video data, perform image preprocessing on each frame of the eye dynamic video data, and determine the pupil center point using visual analysis technology; The second module is used to construct a spatial positioning grid map according to the eye positioning needs. Each positioning grid on the spatial positioning grid map is set with a specific spatial coordinate interval, and each direction of the pupil center point is associated with the positioning grid to construct a pupil center point direction ~ positioning grid correspondence table; The third module is used to obtain several situation factors corresponding to the eye positioning requirements, the situation factors are correlated with the change in the direction of the pupil center point, and screen out several valid situation factors whose correlation with the change in the direction of the pupil center point is greater than or equal to a preset value; The fourth module is used to combine the effective situation factors belonging to the same time node to form a single time node effective situation factor group, and analyze the factor change characteristics of several consecutive single time node effective situation factor groups based on the time series relationship to form a reference situation factor change characteristic group; The fifth module is used to determine the reference positioning grid mapping change of the spatial positioning grid map corresponding to the reference situation factor change feature group, and establish an association relationship with the reference situation factor change feature group to form a reference situation factor change feature group~reference positioning grid mapping change relationship library; The sixth module is used to determine the real-time situation factor group and the real-time positioning grid mapping change in real time during the eye tracking process, and use the reference situation factor change feature group~reference positioning grid mapping change relationship library to perform dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change, select a number of reference positioning grid mapping changes for correcting the real-time grid mapping changes, and based on the determined reference positioning grid mapping changes, form a number of reference pupil center point pointing changes, and use the reference pupil center point pointing changes to optimize and correct the real-time pupil center pointing changes.

[0051] The invention discloses an eye tracking process positioning optimization method and system, and relates to the technical field of eye recognition. Specifically, the invention discloses determining the pupil center point by image preprocessing and visual analysis technology, then constructing a spatial positioning grid diagram, and associating the orientation of the pupil center point with the positioning grid; screening effective situation factors highly correlated with the pupil center point orientation change, and analyzing the time series change characteristics thereof; establishing an association relationship library of a reference situation factor change feature group and a reference positioning grid mapping change; utilizing the relationship library to perform dual feature retrieval on real-time situation factors and real-time positioning grid mapping changes, selecting reference positioning grid mapping changes for correction, and realizing optimization correction of real-time pupil center orientation changes. The invention improves the accuracy and stability of eye tracking through the above technical scheme.

[0052] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A positioning optimization method for an eye tracking process, characterized in that: include: Acquire eye dynamic video data, perform image preprocessing on each frame of the eye dynamic video data, and use visual analysis technology to determine the pupil center point; According to the eyeball positioning requirements, a spatial positioning grid is constructed. Each positioning grid on the spatial positioning grid is set with a specific spatial coordinate interval, and each direction of the pupil center is associated with the positioning grid to construct a pupil center direction ~ positioning grid correspondence table; Acquire several situation factors corresponding to the eyeball positioning requirement, the situation factors are correlated with the change in the direction of the pupil center point, and select several valid situation factors whose correlation with the change in the direction of the pupil center point is greater than or equal to a preset value; Combine the effective situation factors belonging to the same time node to form a single time node effective situation factor group. Based on the time series relationship, analyze the factor change characteristics of several consecutive single time node effective situation factor groups to form a reference situation factor change characteristic group. Determine the reference positioning grid mapping change of the spatial positioning grid map corresponding to the reference situation factor change feature group, and establish an association relationship with the reference situation factor change feature group to form a reference situation factor change feature group~reference positioning grid mapping change relationship library; In the process of eye tracking, the real-time situation factor group is determined in real time, and the real-time positioning grid mapping change is determined. The reference situation factor change feature group~reference positioning grid mapping change relationship library is used to perform dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change, and several reference positioning grid mapping changes for correcting the real-time grid mapping changes are selected. Based on the determined reference positioning grid mapping changes, several reference pupil center point pointing changes are formed, and the reference pupil center point pointing changes are used to optimize and correct the real-time pupil center pointing changes.

2. The eye tracking process positioning optimization method according to claim 1, characterized in that: The method steps for determining the center point of the exit pupil include: De-noising each frame of the eye dynamic video data to obtain a denoised image, graying the denoised image to obtain a grayscale image, and enhancing the contrast of the grayscale image; The pupil edge in the grayscale image is determined using an edge detection algorithm, and the pupil center is determined based on the determined pupil edge.

3. The eye tracking process positioning optimization method according to claim 1, characterized in that: The method of screening out a number of valid situation factors whose correlation with the change in the direction of the pupil center point is greater than or equal to a preset value comprises: Parameterize the situation factor, and associate the situation factor parameter with the positioning grid pointed to by the pupil center point; Classifying the situation factor parameters whose differences are less than or equal to the preset value to form several categories of situation factor parameter sets, analyzing the relative position characteristics of the positioning grids associated with each category of the situation factor parameter set to obtain the first related parameter; A comparison relationship is established between category situation factor sets with the same situation factor category, cluster position characteristics of the positioning grids corresponding to each category situation factor set are compared, a cluster crossover ratio between the two is determined, and a second related parameter is determined based on the cluster crossover ratio; Based on the first correlation parameter and the second correlation parameter, the correlation degree between the situation factor and the change in the direction of the pupil center point is determined.

4. The eye tracking process positioning optimization method according to claim 3, characterized in that: The expression for calculating the correlation between the situation factor and the change in the orientation of the pupil center point is: ； Among them, G is the degree of correlation, is the correlation conversion coefficient, is the straight-line distance of the ith positioning square corresponding to the category situation factor parameter set relative to the central positioning square. The method for determining the central positioning square includes calculating the average coordinates of all positioning squares corresponding to the category situation factor parameter set, and identifying the positioning square closest to the average coordinates as the central positioning square. n is the total number of all positioning squares corresponding to the category situation factor parameter set. is the weight adjustment coefficient affecting the relative position between squares, is the positioning grid coincidence judgment function. When comparing the cluster position features of the positioning grids corresponding to the category situation factor sets, if the xth positioning grid coincides, Output 1, otherwise output 0, N is the total number of positioning squares corresponding to the category situation factor set participating in the comparison, is the cluster cross-proportion impact weight adjustment coefficient, Adjust the constant for the degree of correlation.

5. The eye tracking process positioning optimization method according to claim 1, characterized in that: The method for analyzing the factor change characteristics of a plurality of consecutive single time node effective situation factor groups includes: For the time series, a time axis is established, and a situation factor parameter axis is established perpendicular to the time axis, and the effective situation factor parameters corresponding to each time node are arranged between the time axis and the situation factor parameter axis to form a number of effective situation factor mapping points; Perform linear fitting on a number of effective situation factor mapping points to form an effective situation factor mapping curve; A combination of several effective situation factor mapping curves corresponding to the effective situation factor group is identified as a reference situation factor change characteristic group.

6. The eye tracking process positioning optimization method according to claim 1, characterized in that: The method for establishing the association relationship between the reference positioning grid mapping change and the reference situation factor change feature group includes: A number of associated time segments are constructed for the reference situation factor change feature group and the reference positioning grid mapping change. The first and last nodes of each associated time segment are determined by analyzing the corresponding effective situation factor group of each time node of the reference situation factor change feature group. If the difference in the change of the effective situation factor parameter in the effective situation factor group relative to the previous time node is greater than or equal to a preset value, the corresponding time node is identified as the first and last nodes of the associated time segment. According to the associated time segment, the reference situation factor change feature group and the reference positioning grid mapping change are segmented and the segmented segments are associated.

7. The eye tracking process positioning optimization method according to claim 6, characterized in that: The method for performing dual feature retrieval by the real-time situation factor and the real-time positioning grid mapping change includes: Acquire real-time situation factor parameters, and analyze the time-varying characteristics of each real-time situation factor parameter to form a real-time situation factor change characteristic group, synchronously acquire real-time positioning grid mapping changes, and segment the real-time positioning grid mapping changes and the real-time situation factor change characteristic group according to associated time segments, and correlate the segmented segments; Extracting retrieval tags from mutually related real-time positioning grid mapping changes and real-time situation factor change feature groups, and using the extracted retrieval tags to search in the reference situation factor change feature group~reference positioning grid mapping change relationship library, and performing double feature retrieval on the reference positioning grid mapping changes and reference situation factor change feature groups retrieved initially, including first difference feature analysis between situation factor change feature groups, and second difference feature analysis between positioning grid mapping changes; Based on the first difference feature analysis and the second difference feature analysis, a reference positioning grid mapping change finally retrieved is determined.

8. The eye tracking process positioning optimization method according to claim 7, characterized in that: The method for determining the first difference feature between the situation factor change feature groups includes: Compare the difference amounts of situation factor parameters at different time nodes between situation factor change feature groups, and mark the situation factors whose situation factor parameter difference amounts are greater than or equal to a preset value as difference situation factors, count the number of difference situation factors in real time, and if the number of difference situation factors is greater than or equal to a preset value, it is determined that the first difference feature between the situation factor change feature groups does not meet the preset standard, and the corresponding reference situation factor change feature group and reference positioning grid mapping change are eliminated; The distances between the positioning grids mapped at different time nodes are compared between the positioning grid mapping changes. If the distances between the positioning grids do not meet the preset standard, the corresponding time nodes are marked as grid difference time nodes, and the number of grid difference time nodes is counted in real time. If it is greater than or equal to the preset value, it is determined that the second difference feature does not meet the standard, and the corresponding reference situation factor change feature group and reference positioning grid mapping changes are eliminated.

9. An eye tracking process positioning optimization system, characterized in that: The eye tracking process positioning optimization method according to any one of claims 1 to 8 comprises: The first module is used to obtain eye dynamic video data, perform image preprocessing on each frame of the eye dynamic video data, and determine the pupil center point using visual analysis technology; The second module is used to construct a spatial positioning grid map according to the eye positioning needs. Each positioning grid on the spatial positioning grid map is set with a specific spatial coordinate interval, and each direction of the pupil center point is associated with the positioning grid to construct a pupil center point direction ~ positioning grid correspondence table; The third module is used to obtain several situation factors corresponding to the eye positioning requirements, the situation factors are correlated with the change in the direction of the pupil center point, and screen out several valid situation factors whose correlation with the change in the direction of the pupil center point is greater than or equal to a preset value; The fourth module is used to combine the effective situation factors belonging to the same time node to form a single time node effective situation factor group, and analyze the factor change characteristics of several consecutive single time node effective situation factor groups based on the time series relationship to form a reference situation factor change characteristic group; The fifth module is used to determine the reference positioning grid mapping change of the spatial positioning grid map corresponding to the reference situation factor change feature group, and establish an association relationship with the reference situation factor change feature group to form a reference situation factor change feature group~reference positioning grid mapping change relationship library; The sixth module is used to determine the real-time situation factor group and the real-time positioning grid mapping change in real time during the eye tracking process, and use the reference situation factor change feature group~reference positioning grid mapping change relationship library to perform dual feature retrieval on the real-time situation factor and the real-time positioning grid mapping change, select a number of reference positioning grid mapping changes for correcting the real-time grid mapping changes, and based on the determined reference positioning grid mapping changes, form a number of reference pupil center point pointing changes, and use the reference pupil center point pointing changes to optimize and correct the real-time pupil center pointing changes.

Images