JPH10275212A - Dozing state detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a device always stably detecting a dozing state by changing the extracting condition of a point extracting means according to a density value on a pixel string in the vertical direction of a face and setting the point extracting condition to each pixel string to any optical environmental situation. SOLUTION: Both of and eye position detecting means and an eye opening/ closing detecting means detect the density of pixels along the pixel string in the vertical direction of the face, change the extracting condition of the point extracting means according to the density value on the pixel string and set the point extracting condition to each pixel string. At this device a density detecting means CL2 detects the density of the pixel string in the vertical direction of an inputted face image. A point extracting condition changing means CL3 changes the extracting condition of a point extracting means CL4 by a density value at the pixel value. The means CL4 extracts a point by the rising of density at the pixel string and its changing state. The the awakening degree is judged by the change of an eye opening/closing state.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dozing state detecting apparatus, and more particularly, to a dozing state detecting apparatus for detecting a dozing state of a vehicle driver, a boat operator, an operator of a plant or the like.

[0002]

2. Description of the Related Art As a conventional device for detecting a dozing state of a driver or the like, there is an apparatus described in Japanese Patent Application No. 5-346094. This dozing state detection device detects the position of the eye by performing binarization processing of grayscale image data taken by a camera installed in front of the driver and performing density projection, labeling processing, etc. on the binarized image. In addition, a dozing state can be detected from a change in the state of the open / closed eyes.

[0003] As a dozing state detection apparatus for a grayscale image, an apparatus described in Japanese Patent Application No. 8-2077782 has been proposed by the present applicant.
This dozing state detection device detects the density of pixels along a pixel row in the vertical direction of the face, determines one pixel for each local increase in density in the pixel row, and sets it as an extraction point.
Extraction points adjacent to each other in the pixel column direction of adjacent pixel columns are connected to detect the position of the eye from a group of curves extending in the horizontal direction of the face, and then the same processing as in the detection of the position of the eye in a predetermined area including the eye Is performed, the open / closed state of the eyes is determined, and the dozing state can be detected from a change in the open / closed state.

[0004]

However, in such a conventional dozing state detecting device, as shown in FIG. 20, when the light hits unbalanced, the contrast on the grayscale image is weak. Therefore, it is difficult to accurately set the binarization threshold, and there has been a problem that the eye position and the open / closed eye detection accuracy are reduced.

Further, even in a dozing state detection apparatus for a grayscale image, in an image in which the contrast is extremely weak, the eyes cannot be reliably recognized as continuous data of the extracted points due to the fixed point extraction conditions. However, there is also a problem that the eye position and the open / closed eye detection accuracy are reduced.

The present invention has been made in view of such a conventional problem, and can cope with various light environment conditions, and can always detect a dozing state stably at all times. The purpose is to provide.

[0007]

According to the present invention, an image input means CL1 for inputting a face image and a face image input from the image input means CL1 are provided as shown in FIG. A pixel row density detecting means CL2 for detecting the density of the vertical pixel row, a point extracting condition changing means CL3 for changing an extraction condition of the following point extracting means CL4 based on a density value in the pixel row, Point extraction means CL4 for extracting points based on the increase in density and the change state thereof, and extraction of a curve group for successively extracting a curve group in the horizontal direction of the face from points adjacent to each other in the pixel column direction. Means CL5, eye position detecting means CL6 for selecting an eye from a group of curves, and extraction of the following point extracting means CL8 based on a density value in a vertical direction within a predetermined area including the eyes. A point extraction condition changing unit CL7 for changing a matter; a point extraction unit CL8 for extracting a point from a density increase in a vertical direction in a predetermined area including an eye and a change state thereof; Point continuity judgment means CL for judging point continuity
9, an eye opening detecting means CL10 for obtaining an eye opening value from within the continuous data, an arousal degree determining means CL11 for determining an arousal degree from a change in the open / closed state of the eye, and an awakening degree is decreasing. Alarm means CL1 for giving an alarm when it is determined that
2 is provided.

According to the present invention, in both means of eye position detection and open / closed eye detection, the density of pixels is detected along the vertical pixel row of the face, and point extraction is performed based on the density value on the pixel row. By changing the extraction condition of the means and setting the point extraction condition for each pixel column, the eye can always be stably captured as continuous data of the extraction point, so that it corresponds to any light environment state This can improve the detection accuracy of the dozing state.

[0009]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a doze detection apparatus according to the present invention. Here, in the case of a vehicle driver, the change in the light environment state is considered to be large depending on the height of the sun and the direction of the vehicle. Therefore, in the present embodiment, a case where the present invention is applied to a vehicle will be described.

FIG. 1 is a block diagram showing a configuration of a dozing state detection device according to the present invention, and FIG. 2 is a block diagram showing one embodiment. First, the configuration will be described with reference to FIGS. T as image input means CL1 installed on the instrument and for photographing the driver's face from the front
There is a V-camera 21, and an input image of the TV camera 21 is composed of 512 pixels in the horizontal (X) direction and 480 pixels in the vertical (Y) direction in the present embodiment. An input image captured by the TV camera 21 is stored in the image memory 23 as digital input image data via the AD converter 22. The image memory 23 includes a pixel line density detecting unit CL2 for detecting the density of the vertical pixel line of the face based on the input image data stored therein, and a point extraction condition changing unit based on the pixel line density value. CL3, a point extracting means CL4 for extracting points based on the density increase in the pixel row and its change state, and a curve for connecting points adjacent in the direction of adjacent pixel rows to extract a group of curves in the horizontal direction of the face Image data operation circuit 24 including group extraction means CL5
And an eye position detection circuit 25 of an eye position detection means CL6 for detecting an eye position by selecting an eye from the curve group;
Point extraction condition changing means CL7 based on the density value in the vertical direction in the predetermined area including the eye, and point extraction means CL8 for extracting points from the increase in the density in the vertical direction and the change state in the predetermined area including the eye. Continuity determining means CL9 for determining the continuity of points in the horizontal direction of the face, and eye opening detecting means CL for detecting an eye opening value from the continuous data
An open / closed eye detection circuit 26 comprising 10 is connected. There is a wakefulness determination circuit 27 of the wakefulness determination means CL11 that determines the wakefulness from the detection result of the open / closed eye sent from the open / closed eye detection circuit, and eliminates the dozing state when it is determined that the wakefulness has decreased. An alarm is generated from the alarm device 28 of the alarm means CL12.

Next, the flow of operation in the above configuration will be described with reference to the flowchart of FIG. First, in step S301, the face portion is photographed by the TV camera 21, and in step S302, the input image for one frame is converted into a digital signal by the A / D converter 22 and stored in the image memory 23.

Next, in step S303, it is checked whether or not an eye existence area has been set. The eye existence area indicates a predetermined area including the eye, and is an area in which the processing of detecting the opening of the eye is performed. This predetermined area is also used as an eye tracking area. When the eye existence region is not set, the width in the horizontal direction (X direction) and the width in the vertical direction (Y direction) of the region where the eye exists are set in steps S304 and S305. The details of the eye position detection will be described later with reference to the flowchart of FIG. 4 and the explanatory diagrams shown in FIGS. If it is determined in step S303 that the eye existence area has been set, step S306 is performed.
Is used to detect the degree of eye opening. Details of this processing are shown in FIG.
4 and the explanatory diagrams shown in FIGS. 15 to 17.

Thereafter, in step S307, it is checked whether or not the eye is correctly tracked based on the eye opening value detected in the previous step S306. If it is determined that the eye is correctly tracked, the process proceeds to step S308,
It is determined whether the eye is open or closed based on the eye opening value. afterwards,
The process proceeds to step S309 to track the eyes, that is, to change the region where the eyes are present.

Thereafter, in step S310, the degree of arousal of the driver is determined from the open / closed eye pattern detected in the previous step S309. Details of the determination of the arousal level will be described later with reference to an explanatory diagram shown in FIG.

If it is determined in step S307 that the eye is not correctly tracked, the flow shifts to step S311 to clear the eye opening value and the eye existence area, and returns to step S301 to return to the next frame. Move on to processing. The details of the eye tracking will be described later with reference to an explanatory diagram shown in FIG.

Next, details of the eye position detection will be described. The flow of the eye position detection process will be described with reference to the flowchart of FIG. First, in step S401, as shown in FIG. 5, point extraction processing is performed on data on 480 lines in the Y-axis direction, and after one line is completed, processing is shifted to the processing of the next adjacent line. It is determined whether or not point extraction has been completed for all lines in a predetermined direction.

If it is determined in step S401 that point extraction has not been performed on all lines, the process proceeds to step S402. In this step S402, an arithmetic average of the density values of one line in a predetermined direction is performed. This processing is intended to eliminate small variations in the change in density value at the time of capturing image data, and to capture a global change in density value. FIG. 6A shows a processing result of the arithmetic averaging operation of the line data of Xa in FIG.

In step S403 of FIG.
A differential operation is performed on the arithmetic average value that is the operation result of 402. FIG. 6B shows the processing result.

In step S404 in FIG. 4, FIG.
, The brightest density value of one line in the predetermined direction indicated by Nmax is stored, and the reference value for the differential value, which is the point extraction condition of each line, is changed and set according to the value.

In step S405 of FIG.
Point extraction is performed based on the differential value that is the calculation result of 403. The method of extracting the points is to extract the points (p1 to p10) where the differential value changes from negative to positive as shown in FIG. 6B, and extract the points where the graph becomes convex downward in FIG. 6A. . Next, it is determined whether or not the minimum differential value, which is the maximum point of the change in the density value up to the point, is equal to or less than the determination reference value shown in FIG. Then, the Y coordinate values (a3, a4) are extracted for the point (q3, q4) having.

Next, with reference to FIGS. 7 and 8, a description will be given of a case in which direct light from the right side of the vehicle causes imbalance in light hitting. As shown in FIG. 7, when the left half of the face becomes a shadow due to the direct light from the right side of the vehicle and the contrast becomes weak, the Xb line in FIG. The point corresponding to the eye indicated by p3 cannot be extracted if the same criterion value of the differential value as in FIG. 6B is used, and the eye cannot be recognized as continuous data of the extracted points. In such a case, step S4 in FIG.
It is possible to determine whether the contrast on the line is strong or weak based on the brightest density value of the corresponding line indicated by Nmax in FIG. 8A in the process of 04. Therefore, when Nmax is small, the contrast is weak. And the differential determination reference value can be changed as shown in FIG. Therefore, the points of the eyes and eyebrows in the shadow,
It can be extracted as a2 and a3 in FIG.

Next, a case in which direct light from the front side of the vehicle causes imbalance in light hitting will be described with reference to FIGS. 9 and 10. FIG. As shown in FIG. 9, when the upper half of the face is shadowed by the direct light from the front side of the vehicle and the contrast is weakened, the line Xc in FIG.
The point of the part corresponding to the eye indicated by p3 in FIG.
If the same differential reference value as (b) is used, it cannot be extracted,
The eyes cannot be recognized as continuous data of the extraction points. When the density values are unbalanced in the vertical direction as described above, the lightest density value of the corresponding line indicated by Nmax in FIG. 10A in the process of step S404 in FIG. It cannot be determined whether the contrast of a certain part is strong or weak. As described above, when there is an imbalance in how the light hits due to the direct light from the front side, as shown in FIG. 10B, the fact that continuous data of the extraction point corresponding to the eye is not detected is used as feedback information. , The reference value of the differential value can be changed. Therefore, points of the eyes and eyebrows in the shadow can be extracted as a2 and a3 in FIG.

As shown in FIG. 9, when the upper half of the face is shadowed and the wall has an eye, the driver often uses a sun visor. The criterion value for the differential value can also be changed.

As another case where the contrast of the eyes becomes weak, there is a case where the driver wears glasses. Therefore, the reference value for the differential value can be changed according to the input signal indicating whether or not the spectacles are worn.

After completion of one line, step S406 in FIG.
Then, the processing is switched to the processing of the next line.

If it is determined in step S401 in FIG. 4 that point extraction for all lines has been completed, points as shown in FIG. 11 are extracted. That is, two points A1 and A2 are extracted on the X1 line in FIG. 11, and four points A1, A2, A3 and A4 are extracted on the X2 line.

Thereafter, the flow shifts to step S407, where the Y coordinate values of the extraction points (A1, A2, A3,...) Of the adjacent lines are compared. It stores the data group number, continuous start line number, and continuous data number. The specific processing content will be described with reference to FIG.

Line 1 has Y coordinate values of 192 and 229
There are two extraction points. Y coordinate value of line 1 is 1
Point 92 is because there is no line on the left
Since there is no continuous data at this stage, the group number of the continuous data is “1”. For the same reason, there is no continuous data at this stage for the point having the Y coordinate value of 229. Therefore, the group number of the continuous data is set to the next “2”.

Next, in line 2 on the right side, there are two extraction points whose Y coordinate values are 191 and 224. Line 2
Is a point within 10 of the Y coordinate value 192 of the line 1 on the left, so that the group number of the continuous data is “1”. At this time, the number of continuous data is two. When the same determination is made at the point of the Y coordinate value 224 of the line 2, the group number of the continuous data is “2” and the number of continuous data is 2.

At the point of the Y coordinate value 360 of the next line 3, there is no point within 360 and 10 in the line 2 on the left, so the group number of continuous data is "3" and the number of continuous data is 1 Becomes

The continuous start line number in step S407 refers to a line number having a point where the number of continuous data is determined to be one.

In step S407, the continuity of the points on each line is determined until the processing is completed for all the lines, and the process proceeds to step S408.

In step S408, the average value of the Y coordinate values of points having the same continuous data group number is calculated as
Stores the average value of consecutive points. This value can be used as a representative Y coordinate value of the group. Further, a continuous end line is obtained from the continuous start line and the number of continuous data, and an average value of the continuous start line and the continuous end line is stored. This value can be used as the representative X coordinate value of the group.

In step S409, each continuous group data obtained in this way is used to calculate the length of each continuous group,
By determining based on (X, Y) coordinate values, the position of the eye can be specified.

Here, a specific method of detecting the position of the eye will be described with reference to FIG. First, considering the features of the eye, it can be defined as having a horizontally long and upwardly convex bow shape, and if narrowing down continuous data based on this definition, the eye is said to be long horizontally. Based on the condition, the number of continuous points continues for 5 points or more, and the condition of the bow shape allows the difference between the Y coordinate values of the continuous start point and the continuous end point to be narrowed down to small continuous data. When narrowing down the continuous data based on this determination, the groups G1 to G as shown in FIG.
6 is extracted.

Next, considering the positions of the representative coordinate values of X and Y of each group described above, as shown in FIG.
From the degree of approach in the X coordinate direction, ZONE: L, ZON
It can be classified into E: C and ZONE: R. This is because the left eye and the left eyebrow do not greatly separate in the X coordinate direction, and the right eye and the right eyebrow do not greatly separate in the X coordinate direction. Also, continuous data due to the shadow under the nose,
Mouth continuous data is located near the center.

As described above, by further classifying the data according to the degree of approach in the X coordinate direction and narrowing down the data, the position of the eye can be easily detected. The data included in ZONE: L includes the left eye and the left eyebrow.
E: If it is determined that the continuous data included in R is the right eye and the right eyebrow, the positions of the eyes are G3 and G4, and the coordinate values thereof can be specified.

Next, the details of the eye opening detection will be described. The flow of the eye opening detection process will be described with reference to the flowchart of FIG. First, in step S501, FIG.
As shown in (a), point extraction processing is performed on data on a line in the Y-axis direction, and after the completion of one line, processing is shifted to the next adjacent line, and processing is performed on all lines in a predetermined direction. It is determined whether the point extraction has been completed.

If it is determined in step S501 that point extraction has not been performed on all lines, the process proceeds to step S502. In this step S502,
Arithmetic averaging of the density values of one line in a predetermined direction is performed. This processing is intended to eliminate small variations in the change in density value at the time of capturing image data, and to capture a global change in density value. FIG.
9A shows the processing result of the arithmetic averaging operation of the line data of X in FIG.

In step S503 in FIG. 14, a differential operation is performed on the arithmetic mean value that is the operation result of step S502. The result of this processing is shown in FIG.

In step S504 in FIG. 14, the point extraction condition in step S505 is changed based on the brightest density value of one vertical line, as in the case of detecting the position of the eye.

In step S505 in FIG. 14, point extraction is performed using the differential value that is the result of the operation in step S503. The point is extracted by extracting a point p at which the differential value changes from negative to positive, that is, a point at which the graph becomes convex to the left in FIG. 15B. Next, FIG.
As shown in (c), it is determined whether or not the change in the density value before and after the point is equal to or less than the determination reference value, and a point p having a change in the density value that satisfies the above condition is extracted.

Next, a case where the eye is in shadow and the contrast is weak will be described with reference to FIG. In the detection of the open / closed eye, similarly to the detection of the position of the eye, when the eye is in the shadow, the brightest density value of one line in the vertical direction and Nmax in FIG. 16B are obtained in step S504 of FIG. Accordingly, it can be determined that the contrast of the eye is in a weak state, and the determination reference value of the differential value can be changed as shown in FIG. Therefore, even in a poor light environment where the contrast is weak, the eyes can be reliably recognized as continuous data. In addition, in the case of the open / closed eye detection, since the target area is small, it is possible to determine the light environment state in which the eye is placed also based on the average density of the entire area where the eye is present.

In the detection of the open / closed eyes, it is naturally effective to switch the reference value for the differential value depending on whether or not the sun visor is used and whether or not the spectacles are worn, similarly to the detection of the eye position.

If it is determined in step S506 in FIG. 14 that the extraction point p exists, the process proceeds to step S507.
To q, which is the maximum positive / negative value of the differential value before and after the p point.
An interval H between the point and the r point in the Y coordinate direction is obtained. Step S5
In step 07, when the extraction points are continuously present in the vertical line, the intervals are compared and the maximum value H is updated.

In step S508, the number of consecutive extraction points is counted up. In step S509,
It is determined whether or not the number of consecutive extraction points exceeds a predetermined value, and the processing of the next line is continued until the number exceeds the predetermined value. At this time, the continuity of the extraction points is interrupted, and step S
If it is determined in step 506 that there is no extraction point, the process proceeds to step S512. In step S512, the maximum value of the q and r intervals and the counter being updated are cleared, and the processing of the next line is performed.

In step S509, if the number of consecutive extraction points exceeds a predetermined value, the flow shifts to step S510 to store the maximum value of the interval between q and r. Step S5
At 10, the memory of the maximum value in the continuous data is updated,
Updating between other continuous data exceeding a predetermined value is also performed.

In section D of FIG. 17, there is no extraction point that satisfies the condition of step S505, and in section L, there is an extraction point.

When the processing for all the lines is completed in this way, the flow shifts to step S513, and the maximum value of the q and r intervals is output as the eye opening value.

Next, step S in the flowchart of FIG.
The details of an eye tracking in 307, S309, and S311 and a recovery method when a tracking error occurs are shown. Step S306
As shown in FIG. 17, the eye opening value obtained in
If x is a large value determined to be an eye open, the X coordinate of the point, the median of Hmax is defined as the Y coordinate, and Hma
When x is a small value determined to be an eye closed, an eye existence area is set using the median of the number of continuous data as an X coordinate and the median of Hmax as a Y coordinate. Immediately after the start of the system based on this flowchart, that is, in the first frame, since the eye existence area is not set, the eye existence area is set in steps S304 and S305. At this time, the center coordinates of the eye and the center coordinates of the region where the eye exists are coincident as shown in FIG.

After a series of processes is completed, the process proceeds to the process of the second frame, and proceeds to step S303. Since the present region of the eye is already set here, the process proceeds to step S306 to detect the opening degree of the eye. The process moves to step S307. At this time, if the eyes are correctly captured,
The result is as shown in FIG. Since the eye existence region in FIG. 18B is at the position set in the first frame, the eye position is the eye position of the image data captured in the second frame. Due to movement or the like, the center point of the eye is shifted with respect to the region where the eye exists. However, as long as the eye does not touch the region where the eye exists, the eye opening degree detection described above can be performed.

By changing the reference point of the eye existence area in step S309 to the center coordinates of the eye in FIG. 18B thus captured, it is possible to correspond to the movement of the driver's face. it can.

FIGS. 18C and 18D show the third frame,
FIG. 9 shows a positional relationship between an eye position and an eye window in face image data captured in the fourth frame.

The presence of the eye tracking logic in steps S307, S309, and S311 allows step S30.
4, since the position detection of the eyes from the entire face in S305 can be omitted, the speed of the eye opening detection can be increased.

In step S307, the determination as to whether or not the eye is correctly tracked is made based on the eye opening value. In short, once the photographer has been identified, the eye opening value only changes in the range from when the eye is open to when the eye is closed, so if a value outside this range is output, eye tracking mistakes will occur. The output value of the eye opening and the area where the eye is present are cleared in step S311 to determine that
At 303, the process is again started from the whole face.

Next, the details of the method for determining the open / closed eye in step S308 will be described. As described above, when the photographer is specified, the output value of the degree of eye opening changes between the open state and the closed state. Therefore, the reference value for determining whether the eyes are open or closed, that is, the threshold for determining whether the eyes are open or closed, is within the range. Here, as a result of examining how it is appropriate to set the threshold, a person who is dozing is not in a deep sleep state. It may be desirable to have the median as the threshold.

Finally, referring to FIG.
The details of the method for determining the arousal level at 0 will be described. In the open / closed eye pattern output in step S308 of FIG. 3, as shown in FIG. 19, the arousal level determination section (for example, about 1 minute)
The awakening degree can be determined based on the eye-closed integrated value output to.

The arousal level obtained in this manner makes it possible to detect a dozing state, and to properly activate an alarm device to eliminate a dozing state and prevent a dozing accident or the like.

[0059]

As described in detail above, according to the present invention, in both means for detecting the position of the eye and detecting the open / closed eye,
The density of pixels is detected along the vertical pixel row of the face, the extraction condition of the point extracting means is changed according to the density value on the pixel row, and the point extraction condition is set for each pixel row. By doing so, the eyes can always be stably captured as continuous data of the extraction points, so that the position of the eyes and the degree of opening of the eyes can be accurately detected from the data of the continuous extraction points, and can be used in any light environment. And the detection accuracy of the dozing state can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a dozing state detection device according to the present invention.

FIG. 2 is a block diagram showing a configuration of an embodiment.

FIG. 3 is a flowchart showing an overall operation of the embodiment.

FIG. 4 is a flowchart illustrating an operation of detecting an eye position.

FIG. 5 is an explanatory diagram related to eye position detection.

FIG. 6 is an explanatory diagram related to eye position detection.

FIG. 7 is an explanatory diagram relating to eye position detection.

FIG. 8 is an explanatory diagram related to eye position detection.

FIG. 9 is an explanatory diagram related to eye position detection.

FIG. 10 is an explanatory diagram relating to eye position detection.

FIG. 11 is an explanatory diagram related to eye position detection.

FIG. 12 is an explanatory diagram related to eye position detection.

FIG. 13 is an explanatory diagram related to eye position detection.

FIG. 14 is a flowchart illustrating an operation of detecting an eye opening.

FIG. 15 is a diagram illustrating a method of outputting an eye opening value.

FIG. 16 is an explanatory diagram related to a method of outputting an eye opening value.

FIG. 17 is an explanatory diagram regarding a method of outputting an eye opening value.

FIG. 18 is an explanatory diagram regarding an eye tracking method.

FIG. 19 is an explanatory diagram related to a method for determining a degree of arousal.

FIG. 20 is an explanatory diagram relating to a conventional technique.

[Description of Signs] 21 TV camera 22 A / D converter 23 Image memory 24 Image data calculation open circuit 25 Eye position detection circuit 26 Open / closed eye detection circuit 27 Arousal level determination circuit 28 Alarm device

Claims (6)

[Claims] 1. A dozing state detection apparatus for processing a face image data to detect a dozing state, wherein a density of pixels is detected along a pixel row in a vertical direction of the face, and a density value on the pixel row is detected. A point extracting means for specifying a pixel of a one-way peak of the density value as an extraction point, provided that the minimum differential value of the density change before the peak in the direction exceeds a predetermined value; A point extraction condition changing means for changing a level of a predetermined value, which is a point extraction condition of the point extraction means, from the information; and the extraction points adjacent to each other in the pixel column direction of an adjacent pixel column are continuously extended in the lateral direction of the face. An eye position detecting means for detecting the position of the eye by extracting a curve group; detecting a density of pixels along a vertical pixel row in a predetermined area including the eye; and a density value on the pixel row. Density before and after the one-way peak A point extracting means for specifying a pixel having a unidirectional peak of the density value as an extraction point on condition that the minimum and maximum differential values of the change exceed a predetermined value; and A point extraction condition changing means for changing a level of a predetermined value, which is a point extraction condition of the extraction means; and determining the continuity of the face in the horizontal direction based on the presence or absence of the extraction points for each column, and obtaining data of the continuous extraction points. An eye opening detecting means for detecting the degree of opening of the eye, and determining the open / closed state of the eye from the openness value of the eye, and determining the arousal level from a change in the open / closed state. apparatus. 2. The dozing state detection device according to claim 1, wherein the point extraction condition changing unit determines a level of a predetermined value which is a point extraction condition of the point extraction unit based on a brightest density value on the pixel row. A dozing state detection device characterized by changing 3. The dozing state detection device according to claim 1, wherein the point extraction condition changing means changes a level of a predetermined value, which is a point extraction condition of the point extraction means, according to an average density value of a predetermined area. A dozing state detection device characterized by the above-mentioned. 4. A dozing state detection device for processing a face image data to detect a dozing state, wherein a density of pixels is detected along a pixel row in a vertical direction of the face, and one of density values on the pixel row is detected. A point extracting means for identifying a pixel of a one-way peak of the density value as an extraction point on condition that the minimum differential value of the density change before the peak in the direction exceeds a predetermined value; and a pixel of an adjacent pixel row. An eye position detecting means for detecting the position of the eye by continuously extracting the extraction points adjacent in the column direction and extracting a curve group extending in the horizontal direction of the face; and a vertical pixel in a predetermined area including the eye. The density of the pixels is detected along the column, and on the condition that the minimum and maximum differential values of the density change before and after the one-way peak of the density value on the pixel row exceed a predetermined value, A point extracting means for specifying a peak pixel to be an extraction point; A point extraction condition changing means for changing a level of a predetermined value, which is a point extraction condition of the both point extraction means, depending on whether a sun visor is used or not; Eye opening degree detecting means for detecting the degree of opening of the eye from the data of the continuous extraction points, and determining the open / closed state of the eye from the value of the degree of eye opening, the degree of arousal from the change of the open / closed state A dozing state detection device, characterized in that: 5. A dozing state detection device for processing a face image data to detect a dozing state, wherein a density of pixels is detected along a pixel row in a vertical direction of the face, and one of density values on the pixel row is detected. A point extracting means for identifying a pixel of a unidirectional peak of the density value as an extraction point on condition that the minimum differential value of the density change before the peak in the direction exceeds a predetermined value; Point extraction condition changing means for changing the level of a predetermined value, which is a point extraction condition of the point extraction means, depending on the presence / absence of the point extraction means; An eye position detecting means for detecting the position of the eye by extracting a group of extending curves; detecting a density of pixels along a vertical pixel row in a predetermined area including the eye; Before and after the unidirectional peak of the value On the condition that the minimum and maximum differential values of the density change exceed a predetermined value, a point extracting means for specifying a pixel of a unidirectional peak of the density value as an extraction point, and whether or not the driver wears glasses. A point extraction condition changing means for changing a level of a predetermined value which is a point extraction condition of the point extraction means; and determining the continuity of the face in the horizontal direction depending on the presence or absence of the extraction point for each column, Eye opening detecting means for detecting the degree of opening of the eye from the data of the eye, and determining the open / closed state of the eye from the value of the degree of opening of the eye, and determining the degree of awakening from the change in the open / closed state State detection device. 6. A dozing state detection apparatus for processing a face image data to detect a dozing state, wherein a density of pixels is detected along a pixel row in a vertical direction of the face, and one of density values on the pixel row is detected. A point extracting means for identifying a pixel of a one-way peak of the density value as an extraction point on condition that the minimum differential value of the density change before the peak in the direction exceeds a predetermined value; and a pixel of an adjacent pixel row. Eye position detecting means for detecting the position of the eye by continuously extracting the extraction points close to the column direction and extracting a curve group extending in the lateral direction of the face, equivalent to the eye by the eye position detecting means A point extraction condition changing means for changing a level of a predetermined value, which is a point extraction condition of the point extraction means, when it is determined that there is no curve group data to be executed; To detect pixel density On the condition that the minimum and maximum differential values of the density change before and after the one-way peak of the density value on the pixel row exceed a predetermined value, the pixel of the one-way peak of the density value is specified and the extraction point is specified. Eye extraction means for determining the continuity of the face in the horizontal direction based on the presence or absence of the extraction point for each column, and detecting the opening degree of the eye from the data of the continuous extraction points; A point extraction condition changing unit that changes a level of a predetermined value, which is a point extraction condition of the point extraction unit, when the eye opening degree detection unit determines that there is no continuous data corresponding to the eye; A dozing state detection device, wherein an open / closed state of an eye is determined from an opening degree value of the eye, and an awakening degree is determined from a change in the open / closed state.