阅读说明：本技术 一种基于眼动技术的谎言评判方法 (Lie evaluation method based on eye movement technology ) 是由 武杰 任怡洁 刘乐 于 2021-09-10 设计创作，主要内容包括：本发明提供一种基于眼动技术的谎言评判方法,其特征在于,包括以下步骤：步骤S1,采集受试者的眼动数据；步骤S2,对眼动数据进行预处理；步骤S3,进行均值化处理；步骤S4,行归一化处理；步骤S5,将眼动数据评价指标进行Pearson相关性分析,筛选出相关性最大的归一化后平均瞳孔直径和归一化后平均注视时长作为影响指标；步骤S6,建立多元回归模型；步骤S7,采集待测者的眼动数据,计算出待测者的归一化后平均瞳孔直径和归一化后平均注视时长并输入到多元回归模型；步骤S8,输出待测谎言评价值。本发明利用眼动技术构建多元回归模型,并采集待测者的眼动数据进行谎言评判,具有操作简单,测量效度高,客观性大等优点。(The invention provides a lie evaluation method based on an eye movement technology, which is characterized by comprising the following steps of: step S1, collecting eye movement data of the subject; step S2, preprocessing the eye movement data; step S3, performing averaging processing; step S4, row normalization processing; step S5, performing Pearson correlation analysis on the eye movement data evaluation index, and screening out the normalized average pupil diameter and the normalized average fixation time length with the maximum correlation as influence indexes; step S6, establishing a multiple regression model; step S7, collecting eye movement data of the person to be measured, calculating the normalized average pupil diameter and the normalized average fixation time length of the person to be measured and inputting the normalized average pupil diameter and the normalized average fixation time length into a multiple regression model; and step S8, outputting the evaluation value of the lie to be detected. The invention constructs the multiple regression model by using the eye movement technology, collects the eye movement data of the person to be detected for lie judgment, and has the advantages of simple operation, high measurement efficiency, high objectivity and the like.)

1. A lie judging method based on an eye movement technology is characterized by comprising the following steps:

step S1, collecting eye movement data of the subject within a certain time;

step S2, preprocessing the eye movement data to obtain preprocessed data;

step S3, averaging the preprocessed data to obtain the average pupil diameter of each subjectAnd average gaze duration

Step S4, the mean pupil diameter for each subjectAnd average gaze durationNormalization is carried out to obtain the normalized average pupil diameter x of each subject₁And normalized average gaze duration x₂；

Step S5, the evaluation indexes of the eye movement data are subjected to Pearson correlation analysis, and the normalized average pupil diameter x with the maximum correlation is screened out₁And the normalized average gaze duration x₂As an impact indicator;

step S6, according to the x₁X is the same as₂And testing the lie evaluation value to establish a multiple regression model;

step S7, collecting eye movement data of the person to be measured, and calculating the normalized average pupil diameter x of the person to be measured₁And normalized average gaze duration x₂Inputting the value into the multiple regression model to obtain a value y;

and step S8, outputting the evaluation value of the lie to be detected according to the value y.

2. The lie evaluation method based on eye movement technology according to claim 1, wherein:

wherein, step S6 includes the following steps:

step S6-1, establishing a sample database comprising the normalized mean pupil diameter x for each subject₁Normalized average post-fixation time length x₂And the test lie evaluation value;

step S6-2, constructing a multiple regression equation model, wherein the formula is as follows:

in the formula, b₀As random error, b₁、b₂、b₃、b₄、b₅Is a parameter to be estimated;

step S6-3, substituting the data of the sample database into the multiple regression equation model to determine the parameter b₀、b₁、b₂、b₃、b₄、b₅；

Step S6-4, according to the determined regression model parameter b₀、b₁、b₂、b₃、b₄、b₅Determining the normalized mean pupil diameter x₁And normalized average gaze duration x₂And determining the influence degree of lie evaluation and determining a value interval of the value of the y.

3. The lie evaluation method based on eye movement technology according to claim 2, characterized in that:

the y value interval is divided into a real speech interval, an undetermined interval and a lie speech interval according to the size;

the real-word interval is from 0 to 0.35,

the undeterminable interval is from 0.35 to 0.65,

the lie interval is from 0.65 to 1.

4. The lie evaluation method based on eye movement technology according to claim 1, wherein:

wherein the normalized average pupil diameter x₁The specific expression of (A) is as follows:

x₁(average pupil diameter of a certain subject-average pupil diameter min of all subjects)/(average pupil diameter max of all subjects-average pupil diameter min of all subjects),

the normalized average gaze duration x₂The specific expression of (A) is as follows:

x₂(average gaze duration for a subject-average gaze duration minimum for all subjects)/(average gaze duration maximum for all subjects-average gaze duration for all subjects —)Long minimum).

5. The lie evaluation method based on eye movement technology according to claim 1, wherein:

wherein the eye movement data evaluation index comprises the normalized average pupil diameter x₁Normalized average post-fixation time length x₂Normalized average eye jump speed and number of blinks.

6. The lie evaluation method based on eye movement technology according to claim 1, wherein:

wherein the preprocessing comprises denoising and processing missing values.

Technical Field

The invention relates to a lie judgment method based on an eye movement technology.

Background

The eye movement technology is to extract eye movement data of real-time eye states, such as fixation time, pupil diameter and the like, on the basis of taking the recorded eye movement locus as a basic target, so as to analyze the true activity state of the inner heart of an individual. The eye movement technology is evolved in various methods such as an observation method, a posterior image method, a mechanical recording method and an optical recording method, the eye movement data of eyeballs are collected in real time by mainly utilizing an eye movement instrument, the eye movement instrument is continuously developed towards the direction that evaluation indexes are more accurate and sampling rate is higher along with the development of science and technology, and the application of the eye movement technology in psychology and related disciplines is greatly promoted. The study of ocular kinesis psychology has become a useful paradigm for contemporary psychological studies.

The lie detection technology is a science which integrates multiple disciplines such as psychology, biomedicine, investigation and interrogations, data analysis technology, computer knowledge and the like and detects the true intention and state of the inner heart of an individual. The appearance of the eye movement instrument brings a new discrimination means for lie detection, and the eye movement data can reflect the processing mode of visual information. Has important significance for revealing the psychological mechanism of cognitive processing. Currently, there are two main methods for judging lie, one of which is to analyze the emotion and tone of an individual during the narration by a psychological expert or related researchers to evaluate whether the individual lies. The method is simple to operate, but has high subjectivity and low reliability, and needs professional assistance. Another method is that a professional collects body data of an individual during telling by means of a professional lie detection device, statistical analysis is carried out, and whether the individual lies or not is systematically evaluated. This multi-pass evaluation technique has high accuracy, but is complicated in operation. Both methods lack the acquisition and utilization of eye movement data.

Disclosure of Invention

In order to solve the problems, the invention provides a lie judgment method based on an eye movement technology, which adopts the following technical scheme:

the invention provides a lie evaluation method based on an eye movement technology, which is characterized by comprising the following steps of: step S1, collecting eye movement data of the subject within a certain time; step S2, preprocessing the eye movement data to obtain preprocessed data; step S3, averaging the preprocessed data to obtain the average pupil diameter of each subjectAnd average gaze durationStep S4, average pupil diameter for each subjectAnd average gaze durationNormalization is carried out to obtain the normalized average pupil diameter x of each subject₁And normalized average gaze duration x₂(ii) a Step S5, the evaluation indexes of the eye movement data are subjected to Pearson correlation analysis, and the normalized average pupil diameter x with the maximum correlation is screened out₁And normalized average gaze duration x₂As an impact indicator; step S6, according to x₁、x₂And testing the lie evaluation value to establish a multiple regression model; step S7, collecting eye movement data of the person to be measured, and calculating the normalized average pupil diameter x of the person to be measured₁And normalized average gaze duration x₂Inputting the value into a multiple regression model to obtain a numerical value y; and step S8, outputting the evaluation value of the lie to be detected according to the value y.

The lie evaluation method based on the eye movement technology provided by the invention can also have the technical characteristics that the step S6 comprises the following steps: step S6-1, establishing a sample database including normalized mean pupil diameter x for each subject₁Normalized average post-fixation time duration x₂And testing lie evaluation values; step S6-2, constructing a multiple regression equation model, wherein the formula is as follows:

in the formula, b₀As random error, b₁、b₂、b₃、b₄、b₅Is a parameter to be estimated; step S6-3, substituting the data of the sample database into the multiple regression equation model to determine the parameter b₀、b₁、b₂、b₃、b₄、b₅(ii) a Step S6-4, according to the confirmationFixed regression model parameter b₀、b₁、b₂、b₃、b₄、b₅Determining the normalized mean pupil diameter x₁And normalized average gaze duration x₂And determining the influence degree of lie evaluation and determining a value interval of the value of the y.

The lie evaluation method based on the eye movement technology provided by the invention can also have the technical characteristics that the value interval of the value y is divided into a real-speech interval, an interval which cannot be judged and a lie interval according to the size; the real speech interval is from 0 to 0.35, the untrue speech interval is from 0.35 to 0.65, and the lie speech interval is from 0.65 to 1.

The lie evaluation method based on the eye movement technology provided by the invention can also have the technical characteristics that the average pupil diameter x after normalization₁The specific expression of (A) is as follows:

x₁(average pupil diameter of a certain subject-average pupil diameter min of all subjects)/(average pupil diameter max of all subjects-average pupil diameter min of all subjects),

normalized mean gaze duration x₂The specific expression of (A) is as follows:

x₂(average gaze duration for a certain subject-average gaze duration minimum for all subjects)/(average gaze duration maximum for all subjects-average gaze duration minimum for all subjects).

The lie evaluation method based on the eye movement technology provided by the invention can also have the technical characteristics that the evaluation index of the eye movement data comprises the normalized average pupil diameter x₁Normalized average post-fixation time duration x₂Normalized average eye jump speed and number of blinks.

The lie evaluation method based on the eye movement technology provided by the invention can also have the technical characteristics that the preprocessing comprises denoising and processing the missing value.

Action and Effect of the invention

Lie evaluation based on eye movement technologyThe method comprises the steps of firstly, carrying out Pearson correlation analysis on the evaluation index of the eye movement data, and carrying out average pupil diameter analysisAnd average gaze durationAnd constructing a multiple regression equation model by using the two evaluation indexes with large correlation, and inputting the eye movement data of the person to be tested into the multiple regression equation model to obtain the evaluation value of the lie to be tested. Compared with a plurality of judging technologies such as a professional lie detection device, the measurement effectiveness is higher than that of the plurality of judging technologies, and the operation is simple.

In addition, compared with the lying analysis of the emotion and tone during individual telling, the eye movement technology is high in objectivity and reliable.

Drawings

Fig. 1 is a flowchart of a lie evaluation method based on eye movement technology in an embodiment of the present invention;

FIG. 2 is a diagram illustrating the results of pretreatment in an embodiment of the present invention;

FIG. 3 is a flowchart of the method for building a multiple regression model according to an embodiment of the present invention.

Detailed Description

In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the following embodiment and the accompanying drawings are used to specifically describe the lie judgment method based on the eye movement technology.

< example >

Fig. 1 is a flowchart of a lie evaluation method based on an eye movement technique according to an embodiment of the present invention.

As shown in fig. 1, a lie evaluation method based on an eye movement technology in this embodiment includes the following steps:

and step S1, acquiring eye movement data by using the real-time eye movement instrument, and acquiring the eye movement data of the subject within a period of time.

First, subjects meeting the requirements are recruited. After the subjects were enrolled, the subjects were informed of the experimental precautions and prepared prior to the experiment, such as checking for occlusion of the eyes, wearing eye movement equipment and calibration.

After calibration is complete, acquisition of subject eye movement data may be performed. The subject can sit in a comfortable chair in a clean and bright environment looking straight at a white wall around one meter away, presenting the subject with a dialog presenting a guide, which may be, for example, "next we will ask you twenty questions, ask you answer exactly, what do you have? ".

If the subject has no question and is ready, the subject can enter a question-answering link, and the question-answering link is carried out twice in total, and data of the subject under the situation of all real words and all lies are collected respectively. The whole process is recorded with an eye movement device.

And step S2, preprocessing the eye movement data to obtain preprocessed data.

FIG. 2 is a diagram illustrating a result of the pre-processing according to an embodiment of the present invention.

As shown in fig. 2, after preprocessing the acquired eye movement data such as the average gaze duration, the pupil diameter, the eye jump speed, and the number of blinks, including processing missing values, data averaging, and the like, there is no outlier. Meanwhile, the difference of the eye movement data evaluation indexes in the situations of the actual speech and the lie speech is preliminarily proved.

Step S3, averaging the preprocessed data to obtain the average pupil diameter of each subjectAnd average gaze duration

Step S4, average pupil diameter for each subjectAnd average gaze durationNormalization is carried out to obtain the normalized average pupil diameter x of each subject₁And normalized average gaze duration x₂。

Before data analysis, the data is usually normalized, and the data analysis is performed by using the normalized data. Different evaluation indexes often have different dimensions and dimension units, which affect the result of data analysis, and in order to eliminate the dimension influence among the indexes, data normalization processing is required to solve the comparability among the data indexes. After the raw data are subjected to data normalization processing, all indexes are in the same order of magnitude, and the method is suitable for comprehensive comparison and evaluation.

Normalized mean pupil diameter x₁The specific expression of (A) is as follows:

x₁(average pupil diameter of a certain subject-average pupil diameter min of all subjects)/(average pupil diameter max of all subjects-average pupil diameter min of all subjects),

normalized mean gaze duration x₂The specific expression of (A) is as follows:

x₂(average gaze duration for a subject-average gaze duration minimum for all subjects)/(average gaze duration maximum for all subjects-average gaze duration minimum for all subjects)

Step S5, the evaluation indexes of the eye movement data are subjected to Pearson correlation analysis, and the normalized average pupil diameter x with the maximum correlation is screened out₁And normalized average gaze duration x₂As an influence indicator.

The eye movement data evaluation index comprises normalized average pupil diameter x₁Normalized average post-fixation time duration x₂Normalized average eye jump speed and number of blinks.

Pearson correlation analysis was performed on the evaluation index of the eye movement data to obtain table 1. Such as a watch1, the two indexes with the maximum correlation are respectively the normalized average pupil diameter x₁And normalized average gaze duration x₂These two indices are used as the influence indices.

TABLE 1 Pearson correlation analysis of evaluation index of eye movement data

Step S6, according to x₁、x₂And testing the lie evaluation value to construct a multiple regression model.

FIG. 3 is a flowchart of building a multiple regression model according to an embodiment of the present invention.

As shown in fig. 3, step S6 includes the following steps:

step S6-1, a sample database is established, as shown in Table 2. As shown by Table 2, the sample database included the normalized mean pupil diameter x for each subject₁Normalized average post-fixation time duration x₂And test lie evaluation values (0 is true, 1 is lie).

TABLE 2 sample database

Step S6-2, constructing a multiple regression equation model, wherein the formula is as follows:

in the formula, b₀As random error, b₁、b₂、b₃、b₄、b₅Is the parameter to be estimated.

Step S6-3, establishing a multiple regression model by using MATLAB according to the first 18 data of the sample database, wherein the specific expression of y is as follows:

wherein, b₀＝-0.0782,b₁＝-0.6442,b₂＝0.5885,b₃＝-1.267,b₄＝2.216,b₅＝-0.828。

And (3) regression testing, namely importing the residual 2 pieces of data as model verification data into the model for validity testing, and obtaining the lie judgment scores of the verifier as follows:

y₁＝-0.0782-0.6442×0.66572+0.5885×0.07002-1.267×0.6657×0.0700+2.216×0.6657-0.828×0.0700＝0.0725

y₂＝-0.0782-0.6442×0.55772+0.5885×0.04612-1.267×0.5577×0.0461+2.216×0.5577-0.828×0.0461＝0.8878

this score is very close to the lie evaluation score in the tester database and is considered valid for the multiple linear regression model. Meanwhile, the value interval of the y value in real speech is determined to be 0, 0.35, the value interval of the y value in lie speech is 0.65, 1, and whether the value interval of the y value in lie speech is 0.35,0.65 cannot be determined.

Step S7, when a new person to be tested lies, wearing an eye tracker and collecting eye movement data of the person to be tested, calculating the normalized average pupil diameter x of the person to be tested₁And normalized average gaze duration x₂To obtain x₁＝0.5671，x₂When 0.1748, the above-determined model equation is substituted, the following is obtained:

y＝-0.0782-0.6442×0.56712+0.5885×0.17482-1.267×0.5671×0.1748+2.216×0.5671-0.828×0.1748＝0.7190。

step S8, outputting the evaluation value of lie to be detected as 1 (lie) according to the fact that the y value is in the lie interval [0.65, 1 ].

Examples effects and effects

The lie judging method based on the eye movement technology is provided according to the embodiment. First, the present embodiment analyzes the Pearson correlation of the evaluation index of the eye movement data byNormalized mean pupil diameter x₁Normalized average post-fixation time duration x₂And constructing a multiple regression equation model by using the two evaluation indexes with large correlation, and inputting the eye movement data of the person to be tested into the multiple regression equation model to obtain the evaluation value of the lie to be tested. Compared with a plurality of judging technologies such as a professional lie detection device, the measurement effectiveness is higher than that of the plurality of judging technologies, and the operation is simple.

In addition, compared with the lying analysis of the emotion and tone during individual telling, the eye movement technology is high in objectivity and reliable.

Finally, this embodiment is through using the eye movement appearance to gather eye movement data, compares in electro-oculography, and it is simpler to equip adjustment and operation use, and data analysis is comparatively audio-visual simultaneously, is difficult for receiving the influence of experimenter's physiological difference.

The above-described embodiments are merely illustrative of specific embodiments of the present invention, and the present invention is not limited to the description of the above-described embodiments.