JPH08153288A - Doze judging device - Google Patents

Doze judging device

Info

Publication number
JPH08153288A
JPH08153288A JP29679394A JP29679394A JPH08153288A JP H08153288 A JPH08153288 A JP H08153288A JP 29679394 A JP29679394 A JP 29679394A JP 29679394 A JP29679394 A JP 29679394A JP H08153288 A JPH08153288 A JP H08153288A
Authority
JP
Japan
Prior art keywords
blink
drowsiness
eye
time
subject
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP29679394A
Other languages
Japanese (ja)
Inventor
Makoto Nishida
誠 西田
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Toyota Motor Corp
Original Assignee
Toyota Motor Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Toyota Motor Corp filed Critical Toyota Motor Corp
Priority to JP29679394A priority Critical patent/JPH08153288A/en
Publication of JPH08153288A publication Critical patent/JPH08153288A/en
Pending legal-status Critical Current

Links

Landscapes

  • Auxiliary Drives, Propulsion Controls, And Safety Devices (AREA)
  • Emergency Alarm Devices (AREA)

Abstract

PURPOSE: To correctly judge a doze by weighting a distribution with wink feature quantity as a parameter. CONSTITUTION: A wink detecting means M1 measures the vertical eye width of a person to be examined, compares it with a prescribed threshold value and detects a wink. A wink state detecting means M2 detects the vertical eye width before winking, vertical eye width at the time of winking when the eyes are closed and a eye closing time at the time of winking concerning the detected wink. A sleepiness estimating means M3 executes prescribed weighting as against the distribution with vertical eye width before winking, vertical eye width at the time of winking and the eye closing time at the time of winking as the parameter concerning the detected wink within a prescribed time so as to obtain sleepiness estimating quantity. A doze judging means M4 compares sleepiness estimating quantity with the prescribed threshold value so as to judge the doze of the person to be examined.

Description

【発明の詳細な説明】Detailed Description of the Invention

【０００１】[0001]

【産業上の利用分野】本発明は被験者の瞬きを検出する
瞬き検出装置、また検出された瞬きから被験者の居眠り
を判定する居眠り判定装置に関する。BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a blink detection device for detecting a subject's blink, and a doze determination device for determining a subject's drowsiness from the detected blink.

【０００２】[0002]

【従来の技術】従来より運転者の顔を撮像した顔画像か
ら運転者の瞬きを検出し、この瞬きの状態から運転者の
居眠り等を検出する装置が提案されている。例えば、特
開平６−３２１５４号公報には、運転者の目の縦幅を検
出し、この目の縦幅の最大値及び最小値から閾値を設定
して瞬きを検出し、瞬きの時間変化から閉眼状態が所定
時間以上継続したとき運転者が居眠り状態にあると判定
している。2. Description of the Related Art Conventionally, there has been proposed a device for detecting the blink of the driver from a face image obtained by capturing the face of the driver and detecting the drowsiness of the driver or the like from the state of the blink. For example, in Japanese Unexamined Patent Publication No. 6-32154, the vertical width of a driver's eyes is detected, a threshold is set from the maximum value and the minimum value of the vertical width of the driver's eyes to detect blinks, and the temporal change of the blinks is detected. It is determined that the driver is in a dozing state when the eyes closed state continues for a predetermined time or longer.

【０００３】[0003]

【発明が解決しようとする課題】運転者の目の開閉状態
は一様ではなく、眠気がさすと、閉眼時間が長い瞬きが
発生するだけでなく、閉眼時間が比較的短く、完全に眼
を閉じない不完全な瞬きが連続して発生する場合があ
り、従来装置ではこのような不完全な瞬きの連続から居
眠りを検出することができないという問題があった。When the driver's eyes are not opened and closed uniformly, not only do the eyes blink for a long time, but the eyes are closed for a relatively short time. Incomplete blinks that do not close may occur consecutively, and the conventional device has a problem that it is not possible to detect drowsiness from such a sequence of incomplete blinks.

【０００４】また、瞬き以外にも、運転者があくびをし
た時、笑った時、又は太陽光が目に入った時等において
は、運転者は細目になり、従来装置ではこの細目を瞬き
とみなし誤検出してしまうという問題があった。本発明
は上記の点に鑑みなされたもので、瞬きの特徴量をパラ
メータとする分布に重み付けをして眠気推定量を求める
ことにより、居眠りを正確に判定する居眠り判定装置を
提供することを目的とする。In addition to blinking, when the driver yawns, laughs, or gets into the eyes of the sun, the driver becomes finer. There was a problem of false detection. The present invention has been made in view of the above points, and it is an object of the present invention to provide a drowsiness determination device that accurately determines drowsiness by determining a drowsiness estimation amount by weighting a distribution with a blink feature amount as a parameter. And

【０００５】また、居眠り度合いが低いときの長時間の
閉眼を瞬きとして採用しないことにより、瞬きの誤検出
を防止し、居眠りを正確に判定する居眠り判定装置を提
供することを目的とする。また、被験者の目の照度に応
じて瞬き検出の閾値を変更することにより、瞬きの誤検
出を防止し、居眠りを正確に判定する居眠り判定装置を
提供することを目的とする。Another object of the present invention is to provide a drowsiness determination device which prevents erroneous detection of a blink and accurately determines drowsiness by not adopting eyes closed for a long time as a blink when the degree of drowsiness is low. Another object of the present invention is to provide a drowsiness determination device that prevents erroneous blink detection and accurately determines drowsiness by changing the threshold for blink detection according to the illuminance of the subject's eyes.

【０００６】[0006]

【課題を解決するための手段】請求項１に記載の発明
は、図１（Ａ）に示す如く、被験者の目を撮像した画像
から目縦幅を測定して所定の閾値と比較することで瞬き
を検出し、瞬きの時間変化から被験者の居眠り判定を行
う居眠り判定装置において、瞬き検出手段Ｍ１で検出し
た瞬きについて、瞬き前の目縦幅と、瞬き時の閉眼時目
縦幅と、瞬きの閉眼時間とを検出する瞬き状態検出手段
Ｍ２と、所定時間内で検出した瞬きについて上記瞬き前
の目縦幅と瞬き時目縦幅と閉眼時間とをパラメータとす
る分布に対して所定の重み付けを行い眠気推定量を得る
眠気推定手段Ｍ３と、上記眠気推定量を所定の閾値と比
較して被験者の居眠り判定を行う居眠り判定手段Ｍ４と
を有する。According to a first aspect of the present invention, as shown in FIG. 1A, the eye vertical width is measured from an image of a subject's eye and compared with a predetermined threshold value. In a drowsiness determination device that detects a blink and determines the subject's drowsiness from the time change of the blink, regarding the blink detected by the blink detection means M1, the eye vertical width before the blink, the eye vertical length when the eye is closed, and the blink. Blinking state detection means M2 for detecting the eye-closing time of the eye, and a predetermined weighting for the distribution of the eye-width before the blink, the eye-width at the time of blinking, and the eye-closing time as parameters for the blink detected within the predetermined time. And drowsiness estimation means M4 for performing drowsiness determination of the subject by comparing the estimated drowsiness with a predetermined threshold value.

【０００７】請求項２に記載の発明は、図１（Ｂ）に示
す如く、被験者の目の目縦幅を測定して所定の閾値と比
較することで瞬きを検出し、判定手段Ｍ５で瞬きの時間
変化から被験者の居眠り判定を行う居眠り判定装置にお
いて、瞬き検出手段Ｍ１で検出した瞬きの平均閉眼時間
から居眠り度合いが所定レベル未満の状態において瞬き
の閉眼時間が所定時間を越えるとき、検出した瞬きを誤
検出として除外する瞬き除外手段Ｍ６を有する。According to the second aspect of the invention, as shown in FIG. 1B, the eyebrow width of the subject is measured and compared with a predetermined threshold value to detect blinking, and the determining means M5 blinks. In the drowsiness determination device for determining the subject's drowsiness from the time change, the blinking eye-closing time is detected when the drowsiness level is less than a predetermined level from the average eye-closing time of the blink detected by the blink detection means M1. It has a blink excluding means M6 for excluding the blink as a false detection.

【０００８】請求項３に記載の発明は、図１（Ｃ）に示
す如く、被験者の目の目縦幅を測定して所定の閾値と比
較することで瞬きを検出し、判定手段Ｍ５で瞬きの時間
変化から被験者の居眠り判定を行う居眠り判定装置にお
いて、被験者の目の位置の照度を検出する照度検出手段
Ｍ７と、検出された照度に応じて上記瞬き検出用の閾値
を変更する閾値変更手段Ｍ８とを有する。According to the third aspect of the invention, as shown in FIG. 1C, the eyebrow width of the subject is measured and compared with a predetermined threshold value to detect blinking, and the determining means M5 blinks. In the drowsiness determination device that determines the drowsiness of the subject from the change with time, an illuminance detection unit M7 that detects the illuminance at the eye position of the subject, and a threshold change unit that changes the threshold for blink detection according to the detected illuminance. M8 and.

【０００９】[0009]

【作用】請求項１に記載の発明においては、瞬きの特徴
量をパラメータとする分布に所定の重み付けを行って眠
気推定量を得るため、眠気がさしたときの瞬きと眠気が
ないときの瞬きとで眠気推定量に与える影響が異ならせ
ることができ、眠気がさしたときには眠気推定量が大き
くなり、正確な居眠り判定が可能となる。According to the invention of claim 1, since the drowsiness estimation amount is obtained by performing predetermined weighting on the distribution using the blink feature amount as a parameter, the blinks when drowsiness occurs and the blinks when there is no drowsiness. The influences on the drowsiness estimation amount can be made different by, and when the drowsiness is caused, the drowsiness estimation amount becomes large, so that it becomes possible to accurately determine the drowsiness.

【００１０】請求項２に記載の発明においては、居眠り
度合いが低いときの長時間の閉眼はあくびや笑いによる
ものとして、瞬きから除外することにより、あくびや笑
いを長時間の瞬き、つまり居眠りと誤判定とすることを
防止し、居眠りを正確に判定することができる。請求項
３に記載の発明においては、被験者の目の照度に応じて
瞬き検出の閾値を変更することにより、目に太陽が入り
細目になる状態では上記閾値を下げて長時間の細目状態
を長時間の瞬き、つまり居眠りと誤判定することを防止
し、居眠りを正確に判定することができる。According to the second aspect of the present invention, the long-time closing of the eyes when the degree of drowsiness is low is caused by yawning or laughing, and by excluding it from the blink, the yawning or laughing is blinked for a long time, that is, the drowsiness. It is possible to prevent erroneous determination and accurately determine dozing. In the invention according to claim 3, by changing the threshold value for blink detection according to the illuminance of the eye of the subject, the threshold value is lowered in the state where the sun is in the eyes and the eyes are fine, and the long-term fine state is extended. It is possible to prevent erroneous determination of blinking of time, that is, dozing, and accurately determine dozing.

【００１１】[0011]

【実施例】図２は本発明装置の一実施例の構成図を示
す。同図中、ＣＣＤカメラ１０は運転者に対向して例え
ばステアリングホイール又はダッシュボードに固定され
ており、運転者の顔を撮像する。ここで得られた例えば
モノクロームの顔画像は２値化されて瞬き検出部１１に
供給される。また、照度検出手段Ｍ７としての照度セン
サ１２は運転者に対向して例えばＣＣＤカメラ１０の近
傍に固定されており、運転者の目の位置の照度Ｌを検出
して瞬き検出部１１に供給する。FIG. 2 is a block diagram of an embodiment of the device of the present invention. In the figure, a CCD camera 10 is fixed to a steering wheel or a dashboard so as to face the driver, and images the driver's face. The monochrome face image obtained here is binarized and supplied to the blink detection unit 11. Further, the illuminance sensor 12 as the illuminance detecting means M7 is fixed, for example, in the vicinity of the CCD camera 10 facing the driver, detects the illuminance L at the driver's eye position, and supplies the illuminance L to the blink detection unit 11. .

【００１２】瞬き検出部１１は顔画像内の目を認識し
て、下瞼から上瞼までの目縦幅ＷＶの画素数を計測す
る。次にこの目縦幅ＷＶを閾値Ｗｈthと比較して瞬きを
検出し、検出した瞬きについて瞬きによる閉眼時間Ｔ
ｅ、瞬き前の目縦幅Ｗｂ、瞬き時の目縦幅Ｗｃ夫々を瞬
き特徴量即ち瞬きデータとして計測する。更に、瞬き検
出部１１は過去に検出した瞬きの平均閉眼時間Ｔeav が
閾値Ｔeth1未満で居眠りの度合いが所定レベル未満の状
態で、瞬きの閉眼時間Ｔｅが閾値Ｔeth2を越えて長時間
であれば、その瞬きして検出された閉眼時間があくび又
は笑いによる細目状態とみなし瞬きから除外する。ま
た、照度Ｌが大なる程、瞬き判定用の閾値Ｗｈthを小さ
く変更する。The blink detection unit 11 recognizes the eyes in the face image and measures the number of pixels of the eye vertical width WV from the lower eyelid to the upper eyelid. Next, the eye vertical width WV is compared with a threshold value Whth to detect a blink, and the eye closing time T due to the blink is detected for the detected blink.
e, the eye vertical width Wb before blinking, and the eye vertical width Wc during blinking are measured as blink characteristic amounts, that is, blink data. Furthermore, the blink detection unit 11 detects that the average eye closing time Teav of the blinks detected in the past is less than the threshold value Teth 1 and the degree of drowsiness is less than a predetermined level, and the blink eye closing time Te exceeds the threshold value Teth 2 for a long time. For example, the eye-closing time detected by the blink is regarded as a fine state due to yawning or laughing and is excluded from the blink. Further, the larger the illuminance L is, the smaller the threshold value Whth for blink determination is changed.

【００１３】瞬き検出部１１で検出された所定時間内の
瞬きの瞬きデータは居眠り検出部１３に供給される。瞬
き検出部１１はこの瞬きデータから閉眼時間Ｔｅ、瞬き
前の目縦幅Ｗｂ、瞬き時の目縦幅Ｗｃをパラメータとす
る３次元分布のヒストグラムを作成し、これに重み付け
を行って眠気推定量Ｙを算出する。そして眠気推定量Ｙ
が所定の閾値を越えたとき警報器１４を駆動して警報を
発する。The blink data of the blink within the predetermined time detected by the blink detection unit 11 is supplied to the doze detection unit 13. The blink detection unit 11 creates a three-dimensional distribution histogram with the eye closing time Te, the eye vertical width Wb before blinking, and the eye vertical width Wc during blinking as parameters from this blink data, and weights the histogram to estimate the drowsiness. Calculate Y. And sleepiness estimation amount Y
Exceeds a predetermined threshold, the alarm device 14 is driven to issue an alarm.

【００１４】図３は居眠り判定部１３が実行する居眠り
判定処理のフローチャートを示す。同図中、ステップＳ
１０では瞬き検出部１１で所定期間（例えば３０秒間）
内で最新の瞬きデータを読み込む。瞬きデータは１回の
瞬きについて閉眼時間Ｔｅ、瞬き前目縦幅Ｗｂ、瞬き時
目縦幅Ｗｃからなる。次のステップＳ１２では所定期間
内の瞬きデータから３次元分布ヒストグラムＨｌ（Ｔ
ｅ，Ｗｂ，Ｗｃ）を作成する。FIG. 3 shows a flow chart of the drowsiness determination process executed by the drowsiness determination unit 13. In the figure, step S
In 10, the blink detection unit 11 has a predetermined period (for example, 30 seconds).
Read the latest blink data in. The blink data includes an eye closing time Te, a blink front eye vertical width Wb, and a blink eye vertical width Wc for one blink. In the next step S12, the three-dimensional distribution histogram Hl (T
e, Wb, Wc).

【００１５】ところで、眠気がないときの完全な瞬きは
図４（Ａ）に示す如く、瞬き前の目縦幅Ｗｂの値が大き
く、かつ瞬き時の目縦幅Ｗｃの値が小さいのに対し、眠
気がさしたときの不完全な瞬きは図４（Ｂ）に示す如
く、瞬き前の目縦幅Ｗｂの値が大きく、かつ瞬き時の目
縦幅Ｗｃの値が大きくなる。即ち、眠気があるときの瞬
き時の目縦幅は、眠気がないときに比べて大きい。By the way, in complete blinking without drowsiness, as shown in FIG. 4 (A), the eye vertical width Wb before blinking is large and the eye vertical width Wc during blinking is small. As shown in FIG. 4 (B), the incomplete blink when the person is drowsy has a large eye vertical width Wb before the blink and a large eye vertical width Wc during the blink. That is, the vertical width of the eyes when the person is sleepy is larger than that when the person is not sleepy.

【００１６】上記の３次元ヒストグラムを視認できる形
式の２次元分布ヒストグラムＨｌ1（Ｔｅ，Ｔｂ）、Ｈ
ｌ2 （Ｔｅ，Ｗｃ）として図５（Ａ），（Ｂ）に示す。
図５（Ａ）のヒストグラムＨｌ1 （Ｔｅ，Ｗｂ）におい
て、実線Ｉａで示すピークは眠気がない正常時の瞬きで
あり、瞬き前の目縦幅Ｗｂが大きく、かつ閉眼時間Ｔｅ
が０．５秒程度と短かい。実線Ｉｂで示すピークは眠気
がさしたときの瞬きであり、瞬き前の目縦幅Ｗｂが正常
時の約１／２程度と小さく、かつ閉眼時間Ｔｅが０．８
秒程度と長い。実線Ｉｃで示すピークは眠気がさしたと
きの不完全な瞬きであり、瞬き前の目縦幅Ｗｂ及び閉眼
時間Ｔｅは正常時の瞬きと同程度である。A two-dimensional distribution histogram Hl 1 (Te, Tb), H in a format in which the above three-dimensional histogram can be visually recognized.
This is shown in FIGS. 5A and 5B as l 2 (Te, Wc).
In the histogram Hl 1 (Te, Wb) of FIG. 5A, the peak indicated by the solid line Ia is a blink in a normal state without drowsiness, the eye vertical width Wb before the blink is large, and the eye closing time Te is Te.
Is as short as 0.5 seconds. The peak indicated by the solid line Ib is the blink when drowsiness occurs, the eye vertical width Wb before blinking is as small as about 1/2 of the normal state, and the eye closing time Te is 0.8.
Seconds and long. The peak indicated by the solid line Ic is an incomplete blink when drowsiness occurs, and the eye vertical width Wb before the blink and the eye closing time Te are similar to those in the normal blink.

【００１７】図５（Ｂ）のヒストグラムＨｌ2 （Ｔｅ，
Ｗｃ）において、実線IIａで示すピークは眠気がない正
常時の瞬きであり、瞬き時の目縦幅Ｗｃが小さく、かつ
閉眼時間Ｔｅが０．５秒程度と短かい。実線IIｂで示す
ピークは眠気がさしたときの瞬きであり、瞬き時の目縦
幅Ｗｃは正常時と同程度であり、かつ閉眼時間Ｔｅが
０．８秒程度と長い。実線IIｃで示すピークは眠気がさ
したときの不完全な瞬きであり、瞬き時の目縦幅Ｗｃは
正常時に比べて大きく、かつ閉眼時間Ｔｅは正常時の瞬
きと同程度である。The histogram Hl 2 (Te,
In Wc), the peak indicated by the solid line IIa is a blink in a normal state without drowsiness, the eye vertical width Wc at the time of blinking is small, and the eye closing time Te is short at about 0.5 seconds. The peak shown by the solid line IIb is the blink when the person is drowsy, the eye vertical width Wc at the time of blinking is about the same as in the normal state, and the eye closing time Te is about 0.8 seconds, which is long. The peak indicated by the solid line IIc is an incomplete blink when drowsiness occurs, the eye vertical width Wc at the time of blinking is larger than that in the normal state, and the eye closing time Te is about the same as the blink in the normal state.

【００１８】図３に戻って説明するに、ステップＳ１２
で３次元分布ヒストグラムＨｌ（Ｔｅ，Ｗｂ，Ｗｃ）を
作成した後、ステップＳ１４で眠気推定量Ｙを（１）式
のように計算する。Returning to FIG. 3, step S12 will be described.
After the three-dimensional distribution histogram Hl (Te, Wb, Wc) is created in step S14, the sleepiness estimation amount Y is calculated as in equation (1) in step S14.

【００１９】[0019]

【数１】 [Equation 1]

【００２０】ここで、Ｚ（Ｔｅ，Ｗｂ，Ｗｃ）は図５
（Ａ），（Ｂ）における実線Ｉｂ，IIｂのピーク及び実
線Ｉｃ，IIｃのピーク、つまり眠気がさしたときの瞬き
量Ｈｌを増大し、かつ実線Ｉａ，IIａのピーク、つまり
眠気がないときの瞬き量Ｈｌを減少させる重み付け量で
ある。この重み付け量ＺとヒストグラムＨｌとの積和で
ある眠気推定量Ｙは眠気がさしたときの瞬きの回数が増
加することによってのみ値が増大する。Here, Z (Te, Wb, Wc) is shown in FIG.
The peaks of the solid lines Ib and IIb and the peaks of the solid lines Ic and IIc in (A) and (B), that is, the blink amount Hl when drowsiness is increased, and the peaks of the solid lines Ia and IIa, that is, when there is no drowsiness, This is a weighting amount that reduces the blink amount Hl. The drowsiness estimation amount Y, which is the sum of the weighting amount Z and the histogram Hl, increases only when the number of blinks when drowsiness increases.

【００２１】ステップＳ１６では眠気推定量Ｙを所定の
閾値Ｙthと比較し、Ｙ＞Ｙthのときは眠気推定量Ｙが大
きく居眠り運転のおそれがあるとしてステップＳ１８に
進み、居眠り判定を行う。そして次のステップＳ２０で
警報器１４に警報を発せさせステップＳ２２に進む。ま
たステップＳ１６でＹ≦Ｙthの場合は眠気推定量Ｙが小
さく居眠り運転のおそれがないため、そのままステップ
Ｓ２２に進む。In step S16, the sleepiness estimation amount Y is compared with a predetermined threshold value Yth. If Y> Yth, it is determined that the sleepiness estimation amount Y is large and there is a possibility that the driver may fall asleep. Then, in the next step S20, the alarm device 14 is caused to issue an alarm, and the process proceeds to step S22. If Y ≦ Yth in step S16, the estimated amount of drowsiness Y is small and there is no danger of drowsy driving, so the process directly proceeds to step S22.

【００２２】ステップＳ２２ではイグニッションスイッ
チがオフであるかどうかを判別し、イグニッションスイ
ッチオンの場合はステップＳ１０に進んでステップＳ１
０〜Ｓ２２を繰り返し、イグニッションスイッチがオフ
でエンジンが停止すると処理を終了する。上記のステッ
プＳ１０〜Ｓ１４が眠気推定手段Ｍ３に対応し、ステッ
プＳ１６〜Ｓ２０が居眠り判定手段Ｍ４に対応する。In step S22, it is determined whether or not the ignition switch is off. If the ignition switch is on, the process proceeds to step S10 and step S1.
0 to S22 are repeated, and the process ends when the ignition switch is off and the engine stops. Steps S10 to S14 described above correspond to the drowsiness estimation means M3, and steps S16 to S20 correspond to the drowsiness determination means M4.

【００２３】図６は瞬き検出部１１が実行する瞬き除外
手段Ｍ６としてのノイズ除去処理のフローチャートを示
す。この処理は瞬きの閉眼時間Ｔｅ、瞬き前の目縦幅Ｗ
ｂ、瞬き時の目縦幅Ｗｃを検出する毎に実行される。同
図中、ステップＳ３０では前回の３０秒間に検出された
全ての瞬きの閉眼時間Ｔｅの平均値である平均閉眼時間
Ｔeav が所定の閾値Ｔeth1（Ｔeth1は例えば０．８秒）
未満かどうかを判別し、Ｔeav ＜Ｔeth1のときステップ
Ｓ３２で検出した瞬き（Ｔｅ，Ｗｂ，Ｗｃ）を確認す
る。FIG. 6 shows a flow chart of the noise removing process as the blink excluding means M6 executed by the blink detecting section 11. This processing is the eye closing time Te of the blink, the eye length W before the blink.
b, it is executed each time the eye vertical width Wc at the time of blinking is detected. In the figure, in step S30, the average eye-closing time Teav, which is the average value of all the eye-closing times Te of all blinks detected in the last 30 seconds, is a predetermined threshold value Teth 1 (Teth 1 is 0.8 seconds, for example).
If Teav <Teth 1 , the blink (Te, Wb, Wc) detected in step S32 is confirmed.

【００２４】次のステップＳ３４で確定した瞬きの閉眼
時間Ｔｅが閾値Ｔeth2（Ｔeth2は例えば２秒）を越える
かどうかを判別する。Ｔｅ＞Ｔeth2の場合、つまり平均
閉眼時間Ｔeav がＴeth1未満で短く居眠りの度合いが低
いにも拘らず１回の瞬きの閉眼時間Ｔｅが閾値Ｔeth2を
越えて長い場合は、この瞬きがあくび又は笑いを誤検出
したものとみなし、ステップＳ３６に進んで一旦確認し
た瞬きを除去して処理を終了する。In the next step S34, it is determined whether or not the blink eye closing time Te exceeds the threshold value Teth 2 (Teth 2 is, for example, 2 seconds). In the case of Te> Teth 2 , that is, when the average eye closing time Teav is less than Teth 1 and the degree of dozing is low, but the eye closing time Te of one blink is longer than the threshold value Teth 2 , this blinking occurs. Alternatively, it is considered that laughter has been erroneously detected, and the process proceeds to step S36 to remove the blink once confirmed and end the process.

【００２５】ステップＳ３０でＴeav ≧Ｔeth1の場合
は、あくび又は笑いを判定できないため、そのまま処理
を終了し、ステップＳ３４でＴｅ≦Ｔeth2の場合は検出
した瞬きがあくび又は笑いの誤検出ではないため、その
まま処理を終了する。これによって、あくび又は笑い等
によって目縦幅が小さくなった状態を瞬きとして誤検出
することを防止できる。If Teav ≧ Teth 1 in step S30, yawning or laughing cannot be determined, so the process is ended. If Te ≦ Teth 2 in step S34, the detected blink is not a false detection of yawning or laughing. Therefore, the processing is ended as it is. As a result, it is possible to prevent erroneous detection of a state in which the vertical width of the eyes is reduced due to yawning or laughing as a blink.

【００２６】図７は瞬き検出部１１が実行する閾値変更
手段Ｍ８としてのノイズ除去を含む瞬き検出処理のフロ
ーチャートを示す。同図中、ステップＳ４０では照度セ
ンサ１２で検出した照度Ｌを読み込む。次のステップＳ
４２で定数ｋを照度Ｌで割算して目縦幅の閾値Ｗｈthを
求める。照度Ｌと閾値Ｗｈthとは図８に示す如き関係で
ある。FIG. 7 shows a flow chart of a blink detection process which is executed by the blink detection unit 11 and includes noise removal as the threshold value changing means M8. In the figure, in step S40, the illuminance L detected by the illuminance sensor 12 is read. Next step S
At 42, the constant k is divided by the illuminance L to obtain the threshold value Whth of the eye vertical width. The illuminance L and the threshold Whth have a relationship as shown in FIG.

【００２７】この後、ステップＳ４４で、顔画像におけ
る目縦幅Ｗｈを計測し、次のステップＳ４６で目縦幅Ｗ
ｈが閾値Ｗｈth未満か否かを判別する。Ｗｈ＜Ｗｈthの
場合は瞬きと判定し、瞬きの始まりかどうかを瞬きフラ
グの値から判別する。瞬きフラグは開眼時に０で閉眼時
に１となるフラグである。瞬きフラグが０の場合は現時
点で瞬きが始まったとしてステップＳ５０に進み、瞬き
フラグを１にセットすると共に、現在時刻を瞬き開始点
として記憶し、ステップＳ４０に進む。ステップＳ４８
で瞬きフラグが１であって瞬きが継続している場合はそ
のままステップＳ４０に進む。Thereafter, in step S44, the eye vertical width Wh in the face image is measured, and in the next step S46, the eye vertical width W is calculated.
It is determined whether or not h is less than the threshold Whth. When Wh <Whth, it is determined to be a blink, and whether or not the blink has started is determined from the value of the blink flag. The blink flag is a flag that is 0 when the eyes are open and 1 when the eyes are closed. When the blink flag is 0, it is determined that the blink has started at the present time, and the process proceeds to step S50. The blink flag is set to 1, the current time is stored as the blink start point, and the process proceeds to step S40. Step S48
If the blink flag is 1 and the blink continues, the process directly proceeds to step S40.

【００２８】ステップＳ４６でＷｈ≧Ｗｈthの場合は瞬
きではないと判定してステップＳ５２に進み、瞬きフラ
グが１かどうかにより瞬きの終りか否かを判別する。瞬
きフラグが１の場合は現時点で瞬きが終わったときとし
てステップＳ５４に進み、瞬きフラグを０にリセットす
ると共に、現在時刻を瞬き終了点として記憶し、ステッ
プＳ５６で１回分の瞬き検出を完了してこの処理を終了
する。ステップＳ５２で瞬きフラグが０であって瞬きで
ない状態が継続している場合はステップＳ４０に進む。If Wh ≧ Whth in step S46, it is determined that it is not a blink, and the process proceeds to step S52, in which it is determined whether or not the blink is over depending on whether the blink flag is 1 or not. If the blinking flag is 1, it is determined that the blinking has ended at the present time, the process proceeds to step S54, the blinking flag is reset to 0, the current time is stored as the blinking end point, and one blink detection is completed in step S56. The lever processing ends. If the blink flag is 0 in step S52 and the non-blinking state continues, the process proceeds to step S40.

【００２９】なお、この１回分の瞬き検出が完了した
後、瞬き検出部１１は瞬き開始点と瞬き終了点とから閉
眼時間Ｔｅを算出し、この閉眼時間Ｔｅと、別途検出し
た瞬き前の目縦幅Ｗｂ及び瞬き時の目縦幅Ｗｃと共に記
憶する。ここで、運転者の目の位置の位置の照度Ｌが図
９（Ｂ）の実線に示す如く、暗から明に上昇すると、瞬
き検出の閾値Ｗｈthは図９（Ａ）に破線に示す如く低下
する。これによって、運転者の目に光が入って運転者が
目を細めた場合にも、細目時の目縦幅に応じて閾値Ｗｈ
thが低下し、従来では、細目状態を閉眼状態として瞬き
検出ができなかったのに対し、瞬きを検出することが可
能となる。After the completion of this blink detection, the blink detection unit 11 calculates the eye closing time Te from the blink starting point and the blink ending point, and this eye closing time Te and the separately detected eye blink before and after It is stored together with the vertical width Wb and the eye vertical width Wc when blinking. Here, when the illuminance L at the position of the driver's eyes rises from dark to bright as shown by the solid line in FIG. 9 (B), the threshold Whth for blink detection decreases as shown by the broken line in FIG. 9 (A). To do. With this, even when the driver's eyes shine and the driver narrows his eyes, the threshold value Wh is set according to the vertical width of the eyes when the eyes are narrow.
As a result, th has decreased, and in the related art, blink detection cannot be performed with the fine eye state as the eye-closed state, whereas it is possible to detect blink.

【００３０】なお、上記の実施例では被験者の目を撮像
した画像から目縦幅を測定しているが、この他に光セン
サや眼球近傍の皮膚の電圧変化検出等により目縦幅を測
定しても良く、上記実施例に限定されない。In the above embodiment, the eye vertical width is measured from the image of the subject's eyes. In addition to this, the eye vertical width is measured by an optical sensor or voltage change detection of skin near the eyeball. However, the present invention is not limited to the above embodiment.

【００３１】[0031]

【発明の効果】上述の如く、請求項１に記載の発明によ
れば、瞬きの特徴量をパラメータとする分布に所定の重
み付けを行って眠気推定量を得るため、眠気がさしたと
きの瞬きと眠気がないときの瞬きとで眠気推定量に与え
る影響が異ならせることができ、眠気がさしたときには
眠気推定量が大きくなり、正確な居眠り判定が可能とな
る。As described above, according to the first aspect of the present invention, the drowsiness estimation amount is obtained by performing a predetermined weighting on the distribution having the blink feature amount as a parameter. The influence on the estimated amount of drowsiness can be made different by the blinking when there is no drowsiness, and when the drowsiness occurs, the estimated amount of drowsiness becomes large, and accurate drowsiness determination can be performed.

【００３２】また、請求項２に記載の発明によれば、居
眠り度合いが低いときの長時間の閉眼はあくびや笑いに
よるものとして、瞬きから除外することにより、あくび
や笑いを長時間の瞬き、つまり居眠りと誤判定とするこ
とを防止し、居眠りを正確に判定することができる。ま
た、請求項３に記載の発明によれば、被験者の目の照度
に応じて瞬き検出の閾値を変更することにより、目に太
陽が入り細目になる状態では上記閾値を下げて長時間の
細目状態を長時間の瞬き、つまり居眠りと誤判定するこ
とを防止し、居眠りを正確に判定することができ、実用
上きわめて有用である。According to the second aspect of the present invention, the long-term closing of the eyes when the degree of dozing is low is caused by yawning or laughing, and by excluding it from the blink, the yawning or laughing is blinked for a long time. That is, it is possible to prevent the erroneous determination of dozing and to accurately determine the dozing. Further, according to the invention described in claim 3, by changing the threshold value of blink detection according to the illuminance of the eye of the subject, the threshold value is lowered in the state where the sun is in the eyes and the eyes are fine, and the fine eyes for a long time are selected. It is possible to prevent the erroneous determination of the state as blinking for a long time, that is, dozing, and to accurately determine the dozing, which is extremely useful in practice.

【図面の簡単な説明】[Brief description of drawings]

【図１】本発明の原理図である。FIG. 1 is a principle diagram of the present invention.

【図２】本発明装置の構成図である。FIG. 2 is a configuration diagram of the device of the present invention.

【図３】居眠り判定処理のフローチャートである。FIG. 3 is a flowchart of a drowsiness determination process.

【図４】瞬きを説明するための図である。FIG. 4 is a diagram for explaining blinking.

【図５】瞬きを説明するための図である。FIG. 5 is a diagram for explaining blinking.

【図６】ノイズ除去処理のフローチャートである。FIG. 6 is a flowchart of noise removal processing.

【図７】ノイズ除去を含む瞬き検出処理のフローチャー
トである。FIG. 7 is a flowchart of blink detection processing including noise removal.

【図８】照度と閾値の関係を示す図である。FIG. 8 is a diagram showing a relationship between illuminance and a threshold value.

【図９】図７の瞬き検出処理を説明するための図であ
る。9 is a diagram for explaining the blink detection process in FIG. 7. FIG.

【符号の説明】 １０ ＣＣＤカメラ １１ 瞬き検出部 １２ 照度センサ １３ 居眠り判定部 １４ 警報器 Ｍ１ 瞬き検出手段 Ｍ２ 瞬き状態検出手段 Ｍ３ 眠気推定手段 Ｍ４ 居眠り判定手段 Ｍ６ 瞬き除外手段 Ｍ７ 照度検出手段 Ｍ８ 閾値変更手段[Explanation of Codes] 10 CCD camera 11 Blink detection unit 12 Illuminance sensor 13 Drowsiness determination unit 14 Alarm device M1 Blink detection unit M2 Blink state detection unit M3 Drowsiness estimation unit M4 Drowsiness determination unit M6 Blink exclusion unit M7 Illuminance detection unit M8 Threshold change means

Claims (3)

【特許請求の範囲】[Claims] 【請求項１】 被験者の目の目縦幅を測定して所定の閾
値と比較することで瞬きを検出し、瞬きの時間変化から
被験者の居眠り判定を行う居眠り判定装置において、 検出した瞬きについて、瞬き前の目縦幅と、瞬き時の閉
眼時目縦幅と、瞬きの閉眼時間とを検出する瞬き状態検
出手段と、 所定時間内で検出した瞬きについて上記瞬き前の目縦幅
と瞬き時目縦幅と閉眼時間とをパラメータとする分布に
対して所定の重み付けを行い眠気推定量を得る眠気推定
手段と、 上記眠気推定量を所定の閾値と比較して被験者の居眠り
判定を行う居眠り判定手段とを有することを特徴とする
居眠り判定装置。1. A drowsiness determination device for detecting a blink by measuring the eye vertical length of a subject and comparing it with a predetermined threshold, and determining a subject's drowsiness from the time change of the blink, in the detected blink, Blink state detection means that detects the eye height before blinking, eye length when the eyes are closed during blinking, and eye closing time during blinking, and blinks detected within a predetermined time Drowsiness estimation means that obtains a drowsiness estimation amount by performing predetermined weighting on the distribution having eye vertical width and eye-closing time as parameters, and drowsiness determination that determines the subject's drowsiness by comparing the drowsiness estimation amount with a predetermined threshold value A drowsiness determination device comprising: 【請求項２】 被験者の目の目縦幅を測定して所定の閾
値と比較することで瞬きを検出し、瞬きの時間変化から
被験者の居眠り判定を行う居眠り判定装置において、 検出した瞬きの平均閉眼時間から居眠り度合いが所定レ
ベル未満の状態において瞬きの閉眼時間が所定時間を越
えるとき、検出した瞬きを誤検出として除外する瞬き除
外手段を有することを特徴とする居眠り検出装置。2. An average of the detected blinks in a drowsiness determination device that detects a blink by measuring the eye vertical length of the subject's eye and comparing it with a predetermined threshold value and determines the subject's drowsiness from the time change of the blink. A drowsiness detecting device having a blink excluding means for excluding a detected blink as an erroneous detection when the eye closing time of the blink exceeds a predetermined time in a state where the degree of drowsiness is less than a predetermined level from the eye closing time. 【請求項３】 被験者の目の目縦幅を測定して所定の閾
値と比較することで瞬きを検出し、瞬きの時間変化から
被験者の居眠り判定を行う居眠り判定装置において、 被験者の目の位置の照度を検出する照度検出手段と、 検出された照度に応じて上記瞬き検出用の閾値を変更す
る閾値変更手段とを有することを特徴とする瞬き検出装
置。3. A drowsiness determination device for detecting a blink by measuring the vertical length of the eye of the subject and comparing it with a predetermined threshold, and determining the drowsiness of the subject from the time change of the blink, the position of the eyes of the subject. The illuminance detection device for detecting the illuminance of the illuminance, and the threshold change device for changing the threshold for the blink detection according to the detected illuminance.
JP29679394A 1994-11-30 1994-11-30 Doze judging device Pending JPH08153288A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP29679394A JPH08153288A (en) 1994-11-30 1994-11-30 Doze judging device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP29679394A JPH08153288A (en) 1994-11-30 1994-11-30 Doze judging device

Publications (1)

Publication Number Publication Date
JPH08153288A true JPH08153288A (en) 1996-06-11

Family

ID=17838212

Family Applications (1)

Application Number Title Priority Date Filing Date
JP29679394A Pending JPH08153288A (en) 1994-11-30 1994-11-30 Doze judging device

Country Status (1)

Country Link
JP (1) JPH08153288A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2009276848A (en) * 2008-05-12 2009-11-26 Toyota Motor Corp Driving state estimating device and driving state estimating method
JP2010191853A (en) * 2009-02-20 2010-09-02 Nissan Motor Co Ltd Arousal level estimation device
US7868771B2 (en) 2008-03-18 2011-01-11 Hyundai Motor Company Doze-off warning apparatus for vehicle
JP2011167398A (en) * 2010-02-19 2011-09-01 Toyota Motor Corp Biological state determining apparatus
US10045727B2 (en) 2015-03-09 2018-08-14 Fujitsu Limited Arousal level determination device and computer-readable recording medium
JP2020010865A (en) * 2018-07-19 2020-01-23 本田技研工業株式会社 Driver state determining device and driver state determination method

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7868771B2 (en) 2008-03-18 2011-01-11 Hyundai Motor Company Doze-off warning apparatus for vehicle
JP2009276848A (en) * 2008-05-12 2009-11-26 Toyota Motor Corp Driving state estimating device and driving state estimating method
JP2010191853A (en) * 2009-02-20 2010-09-02 Nissan Motor Co Ltd Arousal level estimation device
JP2011167398A (en) * 2010-02-19 2011-09-01 Toyota Motor Corp Biological state determining apparatus
US10045727B2 (en) 2015-03-09 2018-08-14 Fujitsu Limited Arousal level determination device and computer-readable recording medium
JP2020010865A (en) * 2018-07-19 2020-01-23 本田技研工業株式会社 Driver state determining device and driver state determination method

Similar Documents

Publication Publication Date Title
KR101604232B1 (en) Dozing-off detection method and device
US7830266B2 (en) Drowsy state determination device and method
US20200334477A1 (en) State estimation apparatus, state estimation method, and state estimation program
US7362885B2 (en) Object tracking and eye state identification method
JP5737400B2 (en) Red eye detector
WO2011125166A1 (en) Biological body state assessment device
JP4677963B2 (en) Dozing detection device, dozing detection method
JP4776581B2 (en) Sleep state detection method, sleep state detection device, sleep state detection system, and computer program
CN112183502B (en) Method for determining driving state of driver, computer storage medium, and electronic device
JP5139470B2 (en) Sleepiness level estimation device and sleepiness level estimation method
JP2019087170A (en) Eyelid opening determination device and sleepiness detector
JP4989249B2 (en) Eye detection device, dozing detection device, and method of eye detection device
JP3681657B2 (en) Arousal level estimation device and arousal level estimation method
JP3556439B2 (en) Face image processing device
JP3286887B2 (en) Face image processing apparatus and face image processing method
JPH08153288A (en) Doze judging device
JP5292671B2 (en) Awakening degree estimation apparatus, system and method
JP2001307076A (en) Eye state detector and nap detector
JP2021037216A (en) Eye closing determination device
JP2007034436A (en) Arousal estimation device and method
JP3036319B2 (en) Driver status monitoring device
JP5050794B2 (en) Sleepiness detection device, sleepiness detection method
JP7240910B2 (en) Passenger observation device
JP5167027B2 (en) Blink state detection device
EP3920525B1 (en) Video processing device, video processing method and video processing program