JP2007301087A - Visual line direction detection method or apparatus for vehicle driver - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a highly reliable visual line vector detection method or apparatus performing correction of the measured result of a visual line direction (visual line vector) with an individual difference taken into consideration. <P>SOLUTION: The method of obtaining the visual line direction of a vehicle driver comprises the steps of: obtaining the measured value of the visual line angle in a vertical direction within prescribed time of the driver; obtaining the median of the measured value; obtaining a difference amount between the median of the vehicle driver and the median of a standard driver; applying the correction of the difference amount to the median of the visual line angle in the vertical direction of the vehicle driver; and determining the visual line vector of the vehicle driver on the basis of the corrected median. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a method and apparatus for detecting a human eye vector (or eye direction). More specifically, the present invention relates to a technique for obtaining a more correct line-of-sight vector by correcting the line-of-sight vector of a car driver or the like.

2. Description of the Related Art Conventionally, a technique using a human eye vector as an auxiliary means for a car driver has been developed. The technique for obtaining (or measuring) a gaze vector of a main person that has been performed so far has been able to obtain a gaze vector for each of the left and right eyes. For example, Patent Literature 1 and Patent Literature 2 disclose methods for detecting a human eye vector. According to these methods, both of the left eye eye vector and the right eye eye vector are used. It was possible to obtain the gaze vector.
JP 2005-253778 A, Gaze detection method and apparatus Japanese Patent Application No. 2006-29428, eye vector detection method and apparatus

  These two line-of-sight vectors are considered to coincide on the target object. Therefore, if this knowledge (facts) is taken into consideration, the obtained line-of-sight vector can be checked. As long as the face is in an upright state, the left and right eyes are in a horizontal positional relationship with each other. From this, the above check is effective for the horizontal component of the line-of-sight vector, but is not very effective for the vertical direction (vertical component). Further, since an effective check method has not been found for the vertical direction (vertical component), even if there is an error in the vertical component of the line-of-sight vector, it has been left alone.

  Therefore, an experiment was conducted to examine the presence or absence of individual differences in the vertical direction of the gaze vector. 3 and 4 show the experimental results of five insured persons (A to E). FIG. 3 shows the relationship between the vertical angle (horizontal axis) and the frequency (vertical axis) when five people are looking forward. FIG. 4 shows the vertical angle (horizontal axis) and frequency (vertical axis) when the same five people (A to E) are gazing at the same point (vertical angle = 0 point). ). FIG. 5 shows values obtained by calculating the median (median) in the cases of FIG. 3 and FIG. Here, the median is defined as the measurement value closest to the average value obtained by calculating the average value of the measured values of the angle of the injured person in the vertical direction.

  As described above (FIGS. 3 to 5), the conventional gaze vector detection method or apparatus has a large deviation due to individual differences in the vertical angle, and therefore the conventional gaze vector detection method or There is a possibility that a large error is included in the apparatus, and there is a problem that it is not reliable. That is, as can be understood from FIGS. 3 and 4, it has been found that the median value in the vertical line-of-sight direction differs greatly depending on the individual. Therefore, the median value of each individual was matched with the average value of the median values of standard drivers. 6 and 7 are graphs showing results when the average value of the median value of standard drivers is set to zero and the median value of the experimental results (FIGS. 3 and 4) is made to coincide with zero. From this result, it was considered that the deviation between individuals would be significantly reduced if the median was appropriately corrected. Here, the median value of the standard driver may be an average value with respect to the median value of the vertical line-of-sight angles of at least three normal drivers, and the median value may be obtained by excluding the nonconforming value.

  As described above, the conventional gaze vector detection method or apparatus has a large deviation due to individual differences in the vertical angle, and thus the conventional gaze vector detection method or apparatus that does not consider individual differences includes a large error. There was a problem that it was possible and lacked reliability. An object of the present invention is to provide a highly reliable line-of-sight vector detection method or apparatus by correcting the measurement result of the line-of-sight direction (line-of-sight vector) in consideration of individual differences based on the above findings.

The present invention adopts the following configuration as means for solving the above problems. That is,
The method invention according to claim 1 is a method for obtaining a gaze direction of a vehicle driver, a step of obtaining a measured value of a gaze angle in a vertical direction of the driver within a predetermined time, and a step of obtaining a median value of the measured values. Determining the difference between the median value of the vehicle driver and the median value of the standard driver, correcting the difference value to the median value of the vertical gaze angle of the vehicle driver, Determining a gaze vector of the vehicle driver based on the calculated median value.

  According to a second aspect of the present invention, there is provided a device for determining a gaze direction of a vehicle driver, a measuring device for determining a gaze angle in a vertical direction of the driver within a predetermined time, and a median value of the measured values. A difference amount between the median value and the median value of the standard driver is obtained, and the vehicle driver is corrected based on the corrected median value by correcting the difference value with respect to the median value of the vertical gaze angle of the vehicle driver. And an arithmetic unit for determining the line-of-sight vector.

  The method and apparatus according to claim 3 is the invention according to any one of claims 1 and 2, wherein the median value of the driver's measurement value is obtained by excluding the nonconformity value, and the gaze angle in the vertical direction It is characterized by a median value of.

  The method and apparatus invention according to claim 4 is characterized in that the median value of the standard driver is an average value of median values of the viewing angles in the vertical direction of a plurality of vehicle drivers.

  As described above, according to the configuration of the present invention, since the vertical deviation of the line-of-sight vector due to individual differences is corrected, an effect that a correct line-of-sight vector can be detected (or measured) as compared with the conventional case can be obtained. Further, in claims 3 and 4, there is an effect that a more accurate line-of-sight vector can be obtained.

  FIG. 2 shows the configuration of an apparatus according to an embodiment of the present invention. In FIG. 2, the vehicle driver 10 views an object 12 such as an outside landscape through a windshield 11. Further, a camera 14 for detecting the direction of the line of sight of the vehicle driver 10 is disposed above the front of the vehicle, and photographing data of the camera 14 is connected to a computer 16 by a data wiring 15. Further, a monitor 17, a storage device 18 and input means 19 a and 19 b are connected to the computer 16. The camera 14 captures the eyes of the vehicle driver, and the captured image data of the eyes is transmitted to the computer 16. The computer 16 and the storage device 18 include a program for executing the procedure of FIG. 1 in addition to a program for analyzing the gaze direction (or gaze vector) of the vehicle driver 10 from the image data and a program for obtaining the gaze angle in the vertical direction. Etc. are recorded.

  FIG. 1 shows a flow chart of a method for carrying out the present invention. In FIG. 1, in step S10, the line-of-sight vector of the vehicle driving vehicle is measured a plurality of times over a predetermined time (for example, 1 minute to several minutes). By decomposing in the y direction (vertical direction), the median value (y1) of the y direction components of the line-of-sight vector is obtained. Note that when the median value (y1) is obtained, data (nonconformity value) greatly deviating from the median value may be ignored.

  In step S11, the average value of the median value (Y0) of the y-direction components of the standard driver's line-of-sight vector is read from the recording memory. Next, a deviation (E) between the average value of the read median value (Y0) and the median value (y1) obtained in step S10 is obtained. However, E = y1-Y0. Here, the median value (Y0) of the y-direction component of the standard driver's line-of-sight vector is the median value (Y1) of the y-direction component of the line-of-sight vector for a certain time (for example, several minutes) for a plurality (n) of vehicle drivers. ~ Yn) may be determined as an average value. Note that (n people) is preferably at least 5 or more. This is because the average is more stable as the number of people increases.

In step S12, the current gaze vector of the vehicle driver is measured (detected).
In step S13, the detected line-of-sight vector is decomposed into an x component and a y component. In step S14, the deviation (E) obtained in step S11 is added to the y component obtained in the previous step to correct the y component. . The corrected y component value is y *. In step S15, a corrected line-of-sight vector is obtained by combining the previously obtained x component and y * component. Thereafter, when the line-of-sight vector of the vehicle driver is continuously measured, the process returns to step S12 and the above operations (S12 to S15) are repeated. When it is interrupted, the interrupt (or RETURN) is executed.

  As described above, according to the present embodiment, since an appropriate correction (correction) is performed on the vertical component of the line-of-sight vector, a more accurate line-of-sight vector is determined (measured) than in the past. The effect that it can be obtained. This effect is also reflected in various uses and purposes using the line-of-sight vector.

  The embodiments and examples of the present invention have been described in detail with reference to the drawings. However, the specific configuration is not limited to the examples, and there are design changes and the like without departing from the gist of the present invention. Are also included in the present invention.

It is a flowchart figure which shows the whole procedure of embodiment. The specific structural example of embodiment is shown. An experimental result when an insured person (AE) was looking forward is shown. An experimental result when an insured person (AE) gazes ahead is shown. Indicates the median value of the insured. The result of having corrected the result of FIG. 3 is shown. The result of having corrected the result of FIG. 4 is shown.

Explanation of symbols

10 Vehicle driver
11 Windshield
12 Target Object 14 Camera 15 Signal Wiring 16 Computer 17 Monitor 18 Magnetic Desk

Claims (4)

In the method for obtaining the gaze direction of the vehicle driver, a step of obtaining a measured value of the gaze angle in the vertical direction of the driver within a predetermined time, a step of obtaining a median value of the measured value, and a median value of the vehicle driver A difference amount between the vehicle driver and the standard value of the standard driver, a step of correcting the difference amount to a median value of the vertical gaze angle of the vehicle driver, and the vehicle based on the corrected median value A method for detecting a gaze direction of a vehicle driver, comprising a step of determining a gaze vector of the driver. In an apparatus for determining the line-of-sight direction of a vehicle driver, a measuring device for determining the vertical line-of-sight angle within a predetermined time of the driver, a median value of the measured values, a median value obtained and a median value of a standard driver An arithmetic unit that calculates a difference amount with respect to the median value of the gaze angle in the vertical direction of the vehicle driver, and determines the gaze vector of the vehicle driver based on the corrected median value; An apparatus for detecting a gaze direction of a vehicle driver, comprising: The vehicle driver's gaze direction detection according to claim 1 or 2, wherein the median value of the driver's measured value is a median value of the gaze angle in the vertical direction obtained by excluding non-conforming values. Method or apparatus. 4. The vehicle driver's line-of-sight direction according to claim 1, wherein the median value of the standard driver is an average value with respect to a median value of a plurality of vehicle drivers' vertical line-of-sight angles. Detection method or apparatus.

Images