JP2008027242A - Part determination device of object, and sex determination device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To determine a predetermined part of an object without being affected by ambient light. <P>SOLUTION: The difference (120) between the pixel value of the image of an object taken while being illuminated with light of the near-infrared wavelength region, when projecting the light of the near-infrared wavelength region to the object, and the pixel value of the image of the object taken while being not irradiated with light is calculated for eliminating the influence of the ambient light. Based on the calculated difference, the respective parts of the object, such as a driver's skin, hair and eyeballs, are determined, and the facial region is detected (130). The occupancy of the pixel having the pixel value of mustaches or the like in a part of the facial region between a nose and a mouth is obtained (134 and 136), and based on the result, it is determined whether the object is male or female. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an object region discriminating device and a gender determining device, and in particular, distinguishes a region such as skin, hair, eyeball, etc. of an object such as a driver's head without being affected by ambient light such as sunlight. The present invention relates to a part determination apparatus and a gender determination apparatus for an object that can be determined separately.

A technique for discriminating a person by performing image processing such as pattern recognition for identifying a face image from an image of a person is generally known. Non-Patent Document 1 discloses a gender determination technique as an application of a person determination technique. In this technique, feature amounts are extracted from face images, and the gender and age are estimated using the feature amounts.
Image Lab 2003. pages 35-38 “Automatic Gender Dating System Using Face Images”

  However, the non-patent document 1 does not consider a person who is camouflaged, and the gender determination of the person who camouflaged may be erroneously determined. For example, it is possible to easily capture a face image that is recognized as a gender that is different from the actual sex by disguise such as disguise. This is thought to be because it is insufficient to determine each part of the object.

  The present invention has been made to solve the above-described problem, and an object thereof is to provide an object part discriminating apparatus and a gender determining apparatus capable of easily and reliably discriminating a predetermined part of an object.

  In order to achieve the above object, the present invention provides illumination means for irradiating light in a near-infrared wavelength region, light having a predetermined first light amount and light having a second light amount less than the first light amount. Control means for controlling the amount of light so that each is irradiated, photographing means for photographing an object in synchronization with the light quantity control of the control means, and an image of the object photographed in an illumination state by the light of the first light quantity An arithmetic means for calculating a difference between a pixel value and a pixel value of an image of an object photographed in an illumination state with the second light amount, and based on the difference calculated by the arithmetic means, the skin of the object Detection means for detecting the area and the face area of the object, setting means for setting a part of the face area detected by the detection means as a search area, and within the search area set by the setting means, The skin area and a pixel value different from the pixel value of the skin area Calculation means for calculating the ratio of the specific region consisting of pixels having, based on the ratio calculated by the calculation means is configured to include a discriminating means for discriminating a predetermined portion of the object.

  In the present invention, the difference between the pixel value of the image of the object photographed in the illumination state with the first light amount and the pixel value of the image of the object photographed in the illumination state with the second light amount less than the first light amount. Based on the above, each part of the object is determined. For example, using the pixel value of the image of the object taken in the state where no light is irradiated from the illumination means corresponding to the pixel value of the image of the object taken in the state where the object is exposed to ambient light, the near infrared Based on the difference between the pixel value of the image of the object imaged in a state illuminated with light in the wavelength range and the pixel value of the image of the object imaged in a state where no light is emitted from the illumination means, Since the part is discriminated, each part of the object can be discriminated without being influenced by disturbance light. A skin area and a face area are detected from each part of the object. Since there are many hair root parts such as wrinkles having gender differences in the face region, the skin region and a specific region made up of pixels having pixel values different from the pixel values of the skin region using a part of the face region as a search region, Is determined, and a predetermined part of the object is determined from the ratio. As described above, since the predetermined part of the target object is determined based on the ratio of the specific region having a pixel value different from the pixel value of the skin region obtained from a part of the search region in the face region, the target necessary for gender determination A predetermined part of an object can be determined.

Moreover, the illumination means which irradiates the light of the 1st wavelength range and the light of the 2nd wavelength range different from the 1st wavelength range,
Control means for controlling the amount of light so that the illumination means is irradiated with each of a predetermined first amount of light and a second amount of light less than the first amount of light, and in synchronization with the light amount control of the control means Photographed with photographing means for photographing an object, pixel values of an image of an object photographed in the illumination state with the first light amount in the first wavelength region, and illumination state with the first light amount in the second wavelength region Calculating means for calculating a difference between the pixel value of the image of the target object, and detecting means for detecting a skin area of the target object and a face area of the target object based on the difference calculated by the calculating means; A setting unit that sets a part of the face area detected by the detection unit as a search area, and a pixel value different from the pixel value of the skin area in the search area set by the setting unit A calculation method for calculating a ratio with a specific area including pixels having If, based on the ratio calculated by the calculation means, and determination means for determining a predetermined portion of the object may be configured to include.

  In this case, the control unit controls the light amount of the illumination unit so that the first light amount light in the first wavelength region and the first light amount light in the second wavelength region are alternately irradiated. be able to.

  Further, an illumination unit that irradiates light in a first wavelength region and light in a second wavelength region different from the first wavelength region, and the illumination unit has a predetermined first light amount and less than the first light amount. Control means for controlling the amount of light so that each of the second amount of light is emitted, photographing means for photographing an object in synchronization with the light amount control of the control means, and light of the first amount of light in the first wavelength range A first difference between a pixel value of the image of the object photographed in the illumination state by the light source and a pixel value of the image of the object photographed in the illumination state by the second amount of light in the first wavelength range; The pixel value of the image of the object imaged in the illumination state with the first light amount light in the wavelength range and the pixel value of the image of the object imaged in the illumination state with the second light amount light in the second wavelength region A calculation means for calculating a difference between the first difference and a second difference, and a difference calculated by the calculation means Based on the detection means for detecting the skin area of the object and the face area of the object, a setting means for setting a part of the face area detected by the detection means as a search area, and the setting means Based on the ratio calculated by the calculation means for calculating a ratio between the skin area and a specific area composed of pixels having a pixel value different from the pixel value of the skin area for the set search area And determining means for determining a predetermined part of the object.

  In this case, the control means irradiates the first light amount of light so that the first light amount light of the first wavelength region and the first light amount of the second wavelength region are sequentially irradiated. As described above, it is possible to control the amount of light of the illumination unit so that the second amount of light is irradiated after the amount of light of the illumination unit is controlled.

  An imaging means comprising a first filter that transmits light in a first wavelength range, and a second filter that transmits light in a second wavelength range different from the first wavelength range, and the first filter A control unit that switches and controls the imaging unit so that the image is captured by the transmitted light of the filter or the transmitted light of the second filter, and the pixel value of the image of the object captured by the transmitted light of the first filter And calculating means for calculating the difference between the pixel value of the image of the object photographed by the transmitted light of the second filter, and the skin area of the object and the object based on the difference calculated by the calculating means Detecting means for detecting a face area of an object, setting means for setting a part of the face area detected by the detecting means as a search area, and the skin area for the search area set by the setting means Different from the pixel value of the skin area Comprising: a calculating means for calculating a ratio with a specific area composed of pixels having prime values; and a determining means for determining a predetermined part of the object based on the ratio calculated by the calculating means. Can do.

  In addition, an illuminating unit that irradiates light in a wavelength range including the first wavelength range and the second wavelength range, a predetermined first light amount of light, and a second light amount less than the first light amount. A control means for controlling the amount of light so that each of the light is irradiated, a first filter that transmits light in the first wavelength range, and a second that transmits light in a second wavelength range different from the first wavelength range. An image capturing means for capturing an image of the object in synchronism with the light amount control of the control means, and an object imaged by the transmitted light of the first filter in the irradiation state with the light of the first light amount from the illumination means. The first difference between the pixel value of the image of the image and the pixel value of the image of the object imaged by the transmitted light of the first filter in the irradiation state with the second light amount from the illumination unit, and the first value from the illumination unit Photographed with the light transmitted through the second filter in the irradiation state with one light amount Calculating a second difference between the pixel value of the image of the target object and the pixel value of the image of the target object captured by the transmitted light of the second filter in the irradiation state with the second light amount from the illumination unit; And a calculating means for calculating the difference between the first difference and the second difference, and a detecting means for detecting a skin area of the object and a face area of the object based on the difference calculated by the calculating means. And setting means for setting a part of the face area detected by the detecting means as a search area, and the skin area and the pixel value of the skin area are different for the search area set by the setting means Comprising: a calculating means for calculating a ratio with a specific area composed of pixels having pixel values; and a determining means for determining a predetermined part of the object based on the ratio calculated by the calculating means. Can do.

  In this case, the setting means can obtain the central axis of the face area detected by the detecting means and set a part around the obtained central axis as a search area.

  In the present invention, the first wavelength region and the second wavelength region are preferably set to wavelengths in the near infrared region of 800 nm to 1100 nm.

  Moreover, it is preferable that the first wavelength region is a wavelength region between 800 nm and 970 nm, and the second wavelength region is a wavelength region between 970 nm and 1100 nm.

  The control means sets the state in which light from the illuminating means is irradiated to the state in which light of the first light quantity is emitted, and the state in which no light is emitted from the illuminating means as the second less than the first light quantity. It is preferable to execute the on / off control of the off control in a state where a light amount of light is irradiated as the light amount control.

  By including the part determination device and including a determination unit that determines the gender of the person based on the determination result of the determination unit, the gender determination of the target can be easily and reliably performed.

  As described above, according to the present invention, the predetermined part of the object is determined based on the ratio of the specific area having a pixel value different from the pixel value of the skin area obtained from the partial search area in the face area. An excellent effect is obtained that it is possible to determine a predetermined part of an object necessary for gender determination.

  Hereinafter, an embodiment in which the present invention is applied to a vehicle discriminating device that discriminates each part of a driver's head will be described in detail with reference to the drawings. In the first embodiment, as shown in FIG. 1, an illuminating device composed of a light source 11 that is an illuminating means composed of an LED or the like that irradiates light in the near-infrared wavelength λ0 region, and an illumination device A camera 14 composed of a CCD camera or the like for photographing the target object or the like is provided.

  The light source 11 and the camera 14 are connected to a light source / camera control 16 that controls opening and closing of a shutter of the camera in synchronization with on / off control of the light source 11. Further, the camera 14 includes a pixel value storage device 18 configured by a memory that stores a pixel value such as an image density value of each pixel of an image captured by the camera, and a pixel calculated according to a processing routine described below. It is connected to the arithmetic / determination device 20 that determines what each part of the object is based on the difference in values and the reflectance data stored in the database 22. The arithmetic / judgment device 20 can also be connected to the light source / camera control 16. Note that the light source / camera control device 16 and the calculation / determination device 20 can each be configured by a computer, but may be configured by one computer.

  In the following description, a case where the light source is on / off controlled will be described, but the present invention is not limited to this. Instead of the on / off control, the light amount of the light source may be controlled. In this case, what is necessary is just to control a light source so that light may be irradiated with several different light quantity. For example, the light amount may be controlled so that the light source is irradiated with a predetermined first light amount and a second light amount less than the first light amount. The first light amount may be a light amount that allows the reflected light to be detected by the camera, and the second light amount may be a light amount that is difficult to detect the reflected light by the camera. Further, it is sufficient that the difference is obtained when the reflected light is detected by the camera, that is, there is a predetermined light amount difference between the first light amount and the second light amount. A typical example of controlling so that the first amount of light is irradiated is ON control, and a typical example of controlling so that the second amount of light is irradiated is OFF control.

  As the CCD, XC-E150 (trade name, manufactured by Sony Corporation) having sensitivity in the near infrared region can be used. As shown in FIG. 2, the CCD has sensitivity in a wavelength range of 400 nm to 1000 nm.

  In the present embodiment, a case will be described as an example where an object is used as a head of a driver of a vehicle and each part of the driver's skin and hair is detected and used for gender determination. FIG. 3 shows the spectral reflectance of the skin and the spectral reflectance of the hair. In FIG. 3, reference numeral 33 denotes the spectral reflectance of the human skin, and reference numeral 34 denotes the spectral reflectance of the human hair. When the spectral reflectance of the skin and the spectral reflectance of the hair are compared, in the near-infrared wavelength region exceeding the visible wavelength region, the spectral reflectance at approximately 1100 nm or less is larger than that of the hair. For this reason, it is possible to extract a skin region based on a threshold value of a set value determined corresponding to the skin in image processing.

  Therefore, the driver's head region is extracted as an object by image processing such as template matching, and the skin part in the head region is detected by extracting pixels that exceed the threshold for the pixel value of the captured image. Is possible.

  The wavelength (near infrared wavelength λ0) is preferably 880 nm or 970 nm, but it is not necessary to limit to this. However, the following conditions must be satisfied.

  The first is to use light in the near infrared region. The reason for this is that when used while driving in a vehicle, the driver's face is illuminated with invisible light in order not to distract the driver, but using ultraviolet light will harm the eyes. This is because there is a fear and it is not desirable.

  Secondly, light having a wavelength of 1100 nm or less is used. The reason for this is that an imaging device made of a silicon semiconductor that is excellent in terms of cost, reliability, and responsiveness has sensitivity only to wavelengths up to 1107 nm as illustrated in FIG.

  Third, the reflectance at the wavelength λ0 is different for each skin and hair. In addition, when an illumination device that irradiates light having a wavelength in the vicinity of the visible region is used, it is necessary to consider that light may leak into the visible region and that the sensitivity of the image sensor is almost not 1000 nm or more.

  In the present embodiment, the database 22 stores in advance the spectral reflectance distribution data shown in FIG.

  Next, processing for determining each part of the head executed by the calculation / determination device of the present embodiment will be described.

  First, as shown in FIG. 6, the light source / camera control device 16 performs control so as to repeatedly blink the light source 11 (the light source is turned on and in the state A, and the lighted light source 11 is turned off and in the state C). Therefore, the light source 11 takes two states, that is, only the light of wavelength λ0 is irradiated (state A) and the light source 11 is off (state C).

  In addition, in each of the state A and the state C, the object is photographed by opening and closing the shutter of the camera in synchronization with the on / off control of the light source 11 so that the head of the driver as the object is photographed by the camera. Thereby, an image of the object is taken in each of the state A and the state C.

  FIG. 4 is a flowchart showing processing in the calculation / determination device 20. In step 100 and step 102, it is determined whether the state is C or A. If it is determined in step 100 that the state is C, the pixel value Vn of the image captured by the camera in state C is calculated in step 110 and stored in the pixel value storage device 18. If the state A is determined in step 102, the pixel value V (λ 0) of each pixel of the image captured by the camera in the state A is calculated in step 112 and stored in the pixel value storage device 18. In the next step 120, a difference (V (λ0) −Vn) obtained by subtracting the pixel value of each corresponding pixel of the state C image from the pixel value of each pixel in the state A is calculated and stored for each pixel. Thus, by subtracting the pixel value in the state C from the pixel value in the state A, it is possible to obtain a pixel value (V (λ0) −Vn) based on the spectral intensity of only the near-infrared wavelength. The influence can be removed.

  By performing the above processing, the pixel value of each pixel is calculated on the basis of the state where the object is not illuminated with the light source.

  In the next step 130, a face area is detected. As described above, the background is affected by obtaining two images, an image obtained by irradiating the face with light in the near-infrared wavelength region and a non-irradiated image, and calculating a difference image from the two images. An image in which only the face area has a bright pixel value can be acquired. Thereafter, an appropriate threshold is set according to the brightness of the image, and a face area is extracted. This face area may be a contour, an edge, or an outermost contour. Further, the head region of the driver may be extracted as an object by image processing such as template matching.

  In this face area detection process, the skin part in the head area can be detected. That is, the skin part in the head region can be detected by extracting pixels that exceed the threshold value for the pixel value of the captured image.

  Since the brightness of the light source is known in advance, the magnitude relationship between the skin and hair pixel values in the state A can be stored in the database. By comparing the difference between the database and each pixel (V (λ0) −Vn), it can be determined whether each pixel belongs to hair or skin. Therefore, it is possible to determine whether each pixel belongs to the hair and skin from the spectral reflectance distribution and difference (V (λ0) −Vn) stored in the database for each pixel. Then, it is possible to output identifiable information by giving the determined attribute to each pixel.

  By the way, when the face is irradiated with light in the near-infrared wavelength region, the reflectance differs greatly between hair such as hair and skin. For this reason, in an image taken by irradiating the face with light in the near-infrared wavelength region, the region corresponding to the hair root where no hair appears on the surface has a pixel value different from that of the skin region. This is because light in the near-infrared wavelength region has a characteristic of entering and reflecting inside the skin. Therefore, even if it is a site | part confirmed as skin with visible light, when the area | region containing a hair root exists under skin, it can discriminate | determine by the reflectance of the light of a near-infrared wavelength range. In particular, as a region including the hair root in the face region, there is a unique region that depends on gender. For example, many adult men have wrinkles around the mouth. This indicates that there is a high possibility that the hair roots are concentrated around the mouth. In the present embodiment, a predetermined part where the hair root concentrates in the face region is determined.

  In step 132, the center axis of the face area is detected. The face area is often the target shape with respect to the vertical line. Therefore, the horizontal center of the detected face area can be obtained and the approximate straight line can be used as the center axis of the face area. Note that the derivation of the central axis of the face region is not limited to this, and other calculation methods may be used.

  In the next step 134, an area between the nose area and the mouth area in the face area is set as a search area based on the central axis of the face area and the face area. There is a high probability that the nose region and the mouth region in the face region will be in a predetermined arrangement on the central axis of the face region. In the present embodiment, the standard positional relationship between the nose region and the mouth region on the central axis of the face region is stored in the database 22 in advance. Therefore, a standard area between the nose area and the mouth area can be determined from the positional relationship of the database and the central axis of the face area, and this is set as the search area. The setting of the search area is not limited to this method, and the search area may be obtained from an image specifying a nose area or a mouth area in the face area.

  Next, in step 136, a specific area composed of pixels having a pixel value different from the pixel value of the skin area is obtained in the search area set in step 134, and the ratio (occupancy) of the specific area to the search area Ask for. The pixel value different from the pixel value of the skin region may be the pixel value of the hair stored in the database 22, or the pixel value predetermined as the region including the hair root under the skin is stored in the database 22 and used. May be. Thereby, the ratio of the region where the hair root concentrates under the skin between the nose region and the mouth region (the heel region) can be obtained.

  In the next step 138, the ratio of the region where the hair roots concentrate in the region between the nose region and the mouth region is output as a determination result. When the determination result is equal to or greater than a predetermined occupation ratio, the region where the hair roots concentrate in the face region may be determined as a predetermined portion predicted to be a wrinkle region, and the result may be output. In addition, when the obtained occupancy is equal to or higher than a predetermined occupancy estimated to be a male stored in the database 22 in advance, a determination result that the target person is a male can be output. Further, when the obtained occupancy rate is less than the predetermined occupancy rate, a determination result that the target person is a woman can be output.

  As described above, the presence / absence and amount of the wrinkle region can be confirmed by irradiating the face with light in the near-infrared wavelength region to obtain the ratio of the region where the hair root concentrates in the region between the nose region and the mouth region. In addition, since the hair root under the skin, which is difficult to confirm at visible wavelengths, can be obtained by irradiation with light in the near-infrared wavelength region, it is possible to determine the part existing on the face and easily determine sex. .

  Next, a second embodiment will be described. In this embodiment, each part of an object is determined by alternately irradiating light in two different near-infrared wavelength (λ1, λ2) regions. In addition, since this embodiment is a structure substantially the same as the said embodiment, the same code | symbol is attached | subjected to the same part and detailed description is abbreviate | omitted.

  As shown in FIG. 5, in the present embodiment, the first light source 10 that is an illuminating unit configured by an LED that irradiates light in the first wavelength λ1 region, the LED that irradiates light in the second wavelength λ2 region, and the like. And a camera 14 constituted by a CCD camera or the like for photographing an object or the like illuminated by the illumination device.

  In the present embodiment, two different cases will be described by taking as an example a case in which each part such as the driver's skin or hair is detected by using light in the near-infrared wavelength region as the head of the driver of the vehicle. As shown in FIG. 3, the spectral reflectance of skin and the spectral reflectance of hair are compared as follows when the spectral reflectances at wavelengths of 870 nm and 970 nm are compared.

-Skin: reflectance at 870 nm> reflectance at 970 nm-Hair: reflectance at 870 nm <reflectance at 970 nm

  Accordingly, the head region of the driver is extracted as an object by image processing such as template matching, the pixel values of the captured images are compared at the respective wavelengths of 870 nm and 970 nm, and the sign of the comparison result is determined. It becomes possible to discriminate each part of skin and hair in the region.

  In the present embodiment, the wavelength need not be limited to 870 nm and 970 nm, but the following conditions must be satisfied.

  The first condition and the second condition are the same as in the above embodiment. The third condition is that the reflectance at the two wavelengths λ1 and λ2 differs between skin and hair. In addition, when an illumination device that irradiates light having a wavelength in the vicinity of the visible region is used, it is necessary to consider that light may leak into the visible region and that the sensitivity of the image sensor is almost not 1000 nm or more.

  Note that in this embodiment, two illumination devices are used: a first light source that emits light of 970 nm or more (light with a wavelength λ1) and a second light source that emits light of less than 970 nm (light with a wavelength λ2). .

  Next, processing for determining each part of the head executed by the calculation / determination device of the present embodiment will be described. In the present embodiment, a first light source 10 that is an illumination unit configured to irradiate light having a first wavelength λ1 region, and an illumination unit configured to include an LED unit that irradiates light having a second wavelength λ2 region. Without causing the lighting device configured by the second light source 12 to generate the state C, only the first light source is turned on (state A) and only the second light source is turned on (state B), thereby causing the object Are determined. The first light source and the second light source open or close the shutter of the camera in synchronization with the two light sources alternately emitted as shown in FIG. 6, or the shutter of the camera by the control means 16 as shown in FIG. Two light sources are caused to emit light at a timing synchronized with each other.

  FIG. 8 is a flowchart showing processing in the calculation / determination device 20. In step 102 and step 104, it is determined whether the state is A or B.

  That is, the object is photographed in a state in which the on / off control of the first light source and the second light source and the opening / closing of the shutter of the camera are controlled at the timing shown in FIG. 6, and in step 102 in FIG. Whether or not shooting is performed in a state where the object is illuminated (state A) and whether the second light source is turned on in step 104 to illuminate the object (state B). Judge whether or not.

  In the case of the state A, in step 112, the pixel value V (λ1) of the image obtained by photographing the object with the light of the first wavelength λ1 from the first light source is calculated and stored. In the case of the state B, in step 114, the pixel value V (λ2) of the image obtained by photographing the object with the light of the second wavelength λ2 from the second light source is calculated and stored.

  In the next step 122, the difference (V) between the pixel value V (λ1) of each pixel of the image captured by the camera in the state A and the pixel value V (λ2) of each pixel of the image captured by the camera in the state B. (Λ1) −V (λ2)) is calculated and stored in the pixel value storage device 18 for each pixel. Thus, by subtracting the pixel value in state B from the pixel value in state A, a pixel value based on the spectral intensity between the first wavelength and the second wavelength can be obtained, and the influence of disturbance light is removed. be able to. By performing the above processing, the pixel value of each pixel is calculated on the basis of the relative state when the object is illuminated with light sources in two different wavelength ranges while removing the influence of disturbance light.

  Since the brightness of the light source is known in advance, the magnitude relationship among the pixel values of the skin, hair, and eyes in the state A and the state B can be stored in the database. By comparing the database with the difference between each pixel (V (λ1) −V (λ2)), it can be determined whether each pixel belongs to hair and skin.

  In step 131, the distribution of spectral reflectance stored in the database for each pixel and the difference (V (λ1) −V (λ2)) are compared to determine which part of the hair or skin each pixel belongs to. For example, if the sign of the difference value (V (λ1) −V (λ2)) is a negative sign, it is determined that it belongs to the face region (at least skin), and the difference value (V (λ1) −V (λ2)) If the sign of is positive, it can be determined that it belongs to the hair, and the determined attribute is given to each pixel, and an area including pixels belonging to the face area (at least skin) is detected as a face area. .

  As described above, the face area is detected by acquiring two images obtained by irradiating the face with light of two different near-infrared wavelength regions and calculating a difference image from the two images, thereby obtaining a background image. It is possible to acquire an image in which only the face area has a bright pixel value without being affected by the above. Thereafter, an appropriate threshold is set according to the brightness of the image, and a face area is extracted. This face area may be a contour, an edge, or an outermost contour. Further, the head region of the driver may be extracted as an object by image processing such as template matching.

  Accordingly, in step 122, the difference in spectral reflectance at different wavelengths of the object is calculated using the pixel values V (λ1) and V (λ2) at the wavelengths λ1 and λ2 of the object. Each attribute of hair and face (at least skin) can be given to each pixel by comparing the difference value and the spectral reflectance data of each part of the known object stored in the database. Thus, it can be determined whether each part of the object is hair or face (at least skin).

  Next, a third embodiment will be described. In the present embodiment, three states of the first light source only (state A), only the second light source only on (state B), and two light sources both off (state C) are generated, and each part of the object is determined. Judgment. In addition, since this embodiment is a structure substantially the same as the said embodiment, the same code | symbol is attached | subjected to the same part and detailed description is abbreviate | omitted.

  First, as shown in FIG. 9, the light source / camera control device 16 turns on and turns on the first light source 10 and the second light source 12 alternately (the first light source is on and the second light source is off). A, and the state where the first light source is off and the second light source is on (state B), and after alternately turning on, the first light source 10 and the second light source 12 are turned off and turned off (state C). To control. Accordingly, the two light sources are irradiated with only light of the first wavelength λ1 (state A), irradiated only with light of the second wavelength λ2 (state B), and both the first light source and the second light source are off (state C). Take.

  Further, the camera shutter is synchronized with the on / off control of the first light source 10 and the second light source 12 so that the head of the driver as the object is photographed by the camera in each of the state A, the state B, and the state C. Open and close to shoot the object. Thereby, an image of the object is captured in each of the state A, the state B, and the state C.

  FIG. 10 is a flowchart showing processing in the calculation / determination device 20 in this embodiment. In each of Step 100 to Step 104, it is determined whether the state is C, A, or B. If it is determined in step 100 that the state is C, the pixel value Vn of the image captured by the camera in state C is calculated in step 110 and stored in the pixel value storage device 18.

  If it is determined in step 102 that the state is A, the pixel value V (λ1) of each pixel of the image captured by the camera in state A is calculated in step 112 and stored in the pixel value storage device 18; If it is determined in step 104 that the state is B, the pixel value V (λ2) of each pixel of the image captured by the camera in state B is calculated in step 114 and stored in the pixel value storage device 18.

  In the next step 120, in order to obtain the pixel value of the difference image a for the images in each of the states A and C, the pixel value of each corresponding pixel in the state C image is obtained from the pixel value of each pixel in the state A. The subtracted difference (V (λ1) −Vn) is calculated and stored for each pixel. Next, in step 124, in order to obtain the pixel value of the difference image b for the image in each of the states B and C, the pixel value of each corresponding pixel in the state C image is determined from the pixel value of each pixel in the state B. The difference (V (λ2) −Vn) obtained by subtracting is calculated and stored for each pixel. Thus, by subtracting the pixel value in the state C from the pixel values in the state A and the state B, the pixel values (V (λ1) −Vn, V (λ2) − based on the spectral intensity of only the near infrared wavelength are obtained. Vn) can be obtained. By performing the above processing, the pixel value of each pixel is calculated on the basis of the state where the object is not illuminated with the light source.

  In the next step 126, the pixel value V (λ1) of each pixel of the difference image a obtained in step 120 and the pixel value V (λ1) obtained in step 120 in order to obtain the pixel value of the difference image for the image in each of the states A and B. The difference (V (λ1) −V (λ2)) from the pixel value V (λ2) of each pixel of the obtained difference image b is calculated and stored in the pixel value storage device 18 for each pixel. Thus, by subtracting the pixel value in state B from the pixel value in state A, a pixel value based on the spectral intensity between the first wavelength and the second wavelength can be obtained, and the influence of disturbance light is removed. be able to. By performing the above processing, the pixel value of each pixel is calculated on the basis of the state where the object is not illuminated with the light source. By performing the above processing, the pixel value of each pixel is calculated on the basis of the relative state when the object is illuminated with light sources in two different wavelength ranges while removing the influence of disturbance light.

  Then, in the same manner as described above, the process after step 131 determines whether each pixel belongs to the part of the hair or skin with the sign of the difference value (V (λ1) −V (λ2)), and detects it as a face region. After that, it is determined whether each part of the object is hair or face (at least skin).

  Next, a fourth embodiment will be described. In the present embodiment, normal light including light in the near-infrared wavelength region is divided into light in the first wavelength λ1 region and light in the second wavelength λ2 region on the camera side without using a light source. It is a thing. In addition, since this embodiment is a structure substantially the same as the said embodiment, the same code | symbol is attached | subjected to the same part and detailed description is abbreviate | omitted.

  In this embodiment, as shown in FIG. 11, a camera 14 provided with a bandpass filter 14A (BPF_A) that passes light in the first wavelength λ1 region, and a bandpass that passes light in the second wavelength λ2 region. A camera 15 including a filter 15A (BPF_B) is provided. In the present embodiment, the light source / camera control 16 does not directly control the light source. However, the camera shutter 14 is controlled to be opened and closed in synchronization with a predetermined frequency instead of being synchronized with the on / off control of the light source, and the camera 14, 15 is switched. That is, in the light source / camera control 16, shutter switching and camera switching correspond to light source on / off control.

  Next, a processing routine for determining each part of the object according to the present embodiment will be described with reference to FIG. The shutter of the camera is opened and closed in synchronization with a predetermined frequency by the light source / camera control means, and the camera on the shutter opening side is switched as the target camera, and the image of the object by the camera 14 and the image of the object by the camera 15 are taken. Is done.

  In step 103, it is determined whether or not the target camera is the first camera 14. In step 105, it is determined whether or not the target camera is the second camera 15. In the case of the first camera 14, the first wavelength is determined in step 113. The pixel value of the image (A image) obtained by the light passing through the band pass filter 14A is calculated and stored, and in the case of the second camera 15, it passes through the band pass filter 15A of the second wavelength in step 115. The pixel value of the image (B image) obtained by light is calculated and stored.

  In the next step 123, the pixel value V (λ1) of each pixel of the image (A image) obtained by the light that has passed through the first wavelength bandpass filter 14A and the light that has passed through the second wavelength bandpass filter 15A. The difference (V (λ1) −V (λ2)) from the pixel value V (λ2) of each pixel of the image (B image) obtained by the above is calculated and stored in the pixel value storage device 18 for each pixel. In this way, by subtracting the pixel value in the B image from the pixel value in the A image, a pixel value based on the spectral intensity between the first wavelength and the second wavelength can be obtained, and the influence of disturbance light can be removed. Can do. By performing the above processing, the pixel value of each pixel is calculated on the basis of the relative state when the object is illuminated with light sources in two different wavelength ranges while removing the influence of disturbance light.

  Then, in the same manner as described above, the process after step 131 determines whether each pixel belongs to the part of the hair or skin with the sign of the difference value (V (λ1) −V (λ2)), and detects it as a face region. After that, it is determined whether each part of the object is hair or face (at least skin).

  In this embodiment, an example using two cameras 14 and 15 having different optical axes has been described, but the present invention is not limited to this. For example, as shown in FIG. 13, the same optical axis may be branched into a plurality and a filter may be inserted at the branch destination. FIG. 13 shows an example in which a three-branch prism 13 is inserted on the optical axis L with respect to the object. Incident light from the object is branched into three optical axes by the three-branching prism 13. In the three-branch prism, the first branch optical axis is a bandpass filter 14A (BPF_A) that passes light in the first wavelength λ1 region, and the second branch optical axis is a bandpass filter 15A that passes light in the second wavelength λ2 region. (BPF_B) and the third branch optical axis are guided toward a normal visible wavelength range filter (RGB in FIG. 13).

  Next, a fifth embodiment will be described. In the present embodiment, a single light source is used, and the camera side divides and shoots light in the first wavelength λ1 region and light in the second wavelength λ2 region. In this embodiment, as shown in FIG. 14, a light source 11 is provided in the fourth embodiment. As the light source 11 of the present embodiment, a single light source that irradiates light in a wavelength region including the first wavelength λ1 and the second wavelength λ2 is used. In addition, since this embodiment is a structure substantially the same as the said embodiment, the same code | symbol is attached | subjected to the same part and detailed description is abbreviate | omitted.

  Next, a processing routine for determining each part of the object of the present embodiment will be described with reference to FIG. The light source 11 is turned on / off by the light source / camera control means, the shutter of the camera is opened and closed in synchronization with the on / off of the light source, and the camera on the shutter open side is switched as the target camera. An image of the object, an image of the object by the camera 15, and an image of the object when the light source is off are taken.

  In step 101, it is determined whether the light source is turned on. If the light source is turned on, it is determined whether the target is the first camera 14 (step 103) or the second camera 15 (step 105). In the case of 14, the pixel value (step 113) of the image (A image) obtained by the light passing through the first wavelength bandpass filter 14A, in the case of the second camera 15, the second wavelength bandpass filter 15A is passed. The pixel value (step 115) of the image (B image) obtained by the processed light is calculated and stored. If the light source is off (Yes in step 101), the pixel value of the image obtained by the light that has passed through the bandpass filter 14A of the first wavelength when the light source is off in step 111 is calculated and stored, and the second wavelength of the light source is off. The pixel value of the image obtained by the light passing through the band pass filter 15A is calculated and stored.

  In the next step 127, the object is illuminated with the pixel value V (λ1) of each pixel of the image (a image) obtained by the light passing through the bandpass filter 14A of the first wavelength and the light source obtained in step 125. The difference (V (λ1) −V (λ2)) from the pixel value V (λ2) of each pixel of the image (b image) obtained by the light that has passed through the bandpass filter 15A of the second wavelength on the basis of the state that is not And is stored in the pixel value storage device 18 for each pixel. Thus, by subtracting the pixel value of the b image from the pixel value of the a image, a pixel value based on the spectral intensity between the first wavelength and the second wavelength can be obtained, and the influence of disturbance light can be removed. Can do. By performing the above processing, the pixel value of each pixel is calculated on the basis of the relative state when the object is illuminated with light sources in two different wavelength ranges while removing the influence of disturbance light.

  Then, in the same manner as described above, the process after step 131 determines whether each pixel belongs to the part of the hair or skin with the sign of the difference value (V (λ1) −V (λ2)), and detects it as a face region. After that, it is determined whether each part of the object is hair or face (at least skin).

  As described above, according to each of the above-described embodiments, it is possible to accurately determine each part of a driver's hair and skin, particularly a man-specific wrinkle, without being affected by ambient light. Therefore, it is particularly effective for gender determination.

It is a block diagram which shows the 1st Embodiment of this invention. It is a diagram which shows the spectral sensitivity characteristic of CCD of the camera used for embodiment of this invention. It is a diagram which shows the spectral reflectance characteristic of human skin and hair. It is a flowchart which shows the routine which discriminate | determines each site | part of the target object of 1st Embodiment. It is a block diagram which shows the 2nd Embodiment of this invention. It is a diagram which shows the example of the on-off state of the light source of 2nd Embodiment, and the opening / closing timing of the shutter of a camera. It is a diagram which shows the other example of the on-off state of the light source of 2nd Embodiment, and the opening / closing timing of the shutter of a camera. It is a flowchart which shows the routine which discriminate | determines each site | part of the target object of 2nd Embodiment. It is a diagram which shows the example of the on-off state of the light source of 3rd Embodiment, and the opening / closing timing of the shutter of a camera. It is a flowchart which shows the routine which discriminate | determines each site | part of the target object of 4th Embodiment. It is a block diagram which shows the 4th Embodiment of this invention. It is a flowchart which shows the routine which discriminate | determines each site | part of the target object of 4th Embodiment. It is an optical path figure of the prism which branches an optical axis. It is a block diagram which shows 5th Embodiment. It is a flowchart which shows the routine which discriminate | determines each site | part of the target object of 5th Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 1st light source 11 Light source 12 2nd light source 14 Camera 15 Camera 16 Light source and camera control apparatus 18 Memory | storage device

Claims (12)

Illumination means for irradiating light in the near-infrared wavelength region;
Control means for controlling the amount of light so that the illumination means is irradiated with each of a predetermined first amount of light and a second amount of light less than the first amount of light;
Photographing means for photographing the object in synchronization with the light amount control of the control means;
An arithmetic means for calculating a difference between a pixel value of an image of an object photographed in an illumination state with the first light amount of light and a pixel value of an object image photographed in an illumination state of the second light amount of light;
Detecting means for detecting a skin area of the object and a face area of the object based on the difference calculated by the calculating means;
Setting means for setting a part of the face area detected by the detection means as a search area;
For the search area set by the setting means, a calculation means for calculating a ratio between the skin area and a specific area composed of pixels having pixel values different from the pixel values of the skin area;
Discrimination means for discriminating a predetermined portion of the object based on the ratio calculated by the calculation means;
Discriminating device for an object including Illuminating means for irradiating light in a first wavelength range and light in a second wavelength range different from the first wavelength range;
Control means for controlling the amount of light so that the illumination means is irradiated with each of a predetermined first amount of light and a second amount of light less than the first amount of light;
Photographing means for photographing the object in synchronization with the light amount control of the control means;
The pixel value of the image of the object imaged in the illumination state with the first light amount light in the first wavelength range and the pixel value of the image of the object imaged in the illumination state with the first light amount light in the second wavelength region Computing means for computing the difference between
Detecting means for detecting a skin area of the object and a face area of the object based on the difference calculated by the calculating means;
Setting means for setting a part of the face area detected by the detection means as a search area;
For the search area set by the setting means, a calculation means for calculating a ratio between the skin area and a specific area composed of pixels having pixel values different from the pixel values of the skin area;
Discrimination means for discriminating a predetermined portion of the object based on the ratio calculated by the calculation means;
Discriminating device for an object including   The control means controls the light quantity of the illuminating means so that the first light quantity light in the first wavelength range and the first light quantity light in the second wavelength range are alternately irradiated. Item 2. The object region discriminating apparatus according to Item 2. Illuminating means for irradiating light in a first wavelength range and light in a second wavelength range different from the first wavelength range;
Control means for controlling the amount of light so that the illumination means is irradiated with each of a predetermined first amount of light and a second amount of light less than the first amount of light;
Photographing means for photographing the object in synchronization with the light amount control of the control means;
The pixel value of the image of the object imaged in the illumination state with the first light amount light in the first wavelength range and the pixel value of the image of the object imaged in the illumination state with the second light amount light in the first wavelength region The pixel value of the image of the object imaged in the illumination state with the first light amount in the second wavelength range and the illumination state in the illumination state with the second light amount in the second wavelength region Calculating means for calculating a second difference between the pixel value of the image of the target object and calculating a difference between the first difference and the second difference;
Detecting means for detecting a skin area of the object and a face area of the object based on the difference calculated by the calculating means;
Setting means for setting a part of the face area detected by the detection means as a search area;
For the search area set by the setting means, a calculation means for calculating a ratio between the skin area and a specific area composed of pixels having pixel values different from the pixel values of the skin area;
Discrimination means for discriminating a predetermined portion of the object based on the ratio calculated by the calculation means;
Discriminating device for an object including   The controller emits the first light amount so that the first light amount light in the first wavelength region and the first light amount light in the second wavelength region are sequentially irradiated. 5. The apparatus for discriminating a part of an object according to claim 4, wherein the light amount of the illumination unit is controlled so that the second amount of light is irradiated after the light amount of the illumination unit is controlled. An imaging means comprising a first filter that transmits light in a first wavelength range, and a second filter that transmits light in a second wavelength range different from the first wavelength range;
Control means for switching and controlling the photographing means so as to be photographed by the transmitted light of the first filter or photographed by the transmitted light of the second filter;
A computing means for computing a difference between a pixel value of the image of the object photographed by the transmitted light of the first filter and a pixel value of the image of the object photographed by the transmitted light of the second filter;
Detecting means for detecting a skin area of the object and a face area of the object based on the difference calculated by the calculating means;
Setting means for setting a part of the face area detected by the detection means as a search area;
For the search area set by the setting means, a calculation means for calculating a ratio between the skin area and a specific area composed of pixels having pixel values different from the pixel values of the skin area;
Discrimination means for discriminating a predetermined portion of the object based on the ratio calculated by the calculation means;
Discriminating device for an object including Illumination means for irradiating light in a wavelength range including the first wavelength range and the second wavelength range;
Control means for controlling the amount of light so that the illumination means is irradiated with each of a predetermined first amount of light and a second amount of light less than the first amount of light;
A first filter that transmits light in the first wavelength range; and a second filter that transmits light in a second wavelength range different from the first wavelength range; and synchronized with the light amount control of the control means. Photographing means for photographing the object,
The pixel value of the image of the object photographed with the transmitted light of the first filter in the irradiation state with the first light amount from the illumination unit, and the first filter in the irradiation state with the second light amount from the illumination unit. The first difference from the pixel value of the image of the object photographed with the transmitted light, and the pixel value of the image of the object photographed with the transmitted light of the second filter in the irradiation state with the first amount of light from the illumination means And the second difference between the pixel value of the image of the object photographed by the transmitted light of the second filter in the irradiation state with the second light amount from the illumination means, and the first difference and the first difference Computing means for computing the difference between the two,
Detecting means for detecting a skin area of the object and a face area of the object based on the difference calculated by the calculating means;
Setting means for setting a part of the face area detected by the detection means as a search area;
For the search area set by the setting means, a calculation means for calculating a ratio between the skin area and a specific area composed of pixels having pixel values different from the pixel values of the skin area;
Discrimination means for discriminating a predetermined portion of the object based on the ratio calculated by the calculation means;
Discriminating device for an object including   The said setting means calculates | requires the center axis | shaft of the face area detected by the said detection means, and sets a part around the calculated | required center axis | shaft as a search area | region. An apparatus for discriminating a part of an object according to item.   The region of the object according to any one of claims 2 to 8, wherein the first wavelength region and the second wavelength region are set to wavelengths in a near infrared region of 800 nm to 1100 nm. Discriminator.   The first wavelength range is a wavelength range between 800 nm and 970 nm, and the second wavelength range is a wavelength range between 970 nm and 1100 nm. An apparatus for discriminating a part of an object according to item.   The control means sets the state in which light from the illuminating means is irradiated to the state in which light of the first light quantity is emitted, and the state in which no light is emitted from the illuminating means as the second less than the first light quantity. 11. The object region discriminating apparatus according to claim 1, wherein on / off control of off control in which a light amount of light is irradiated is executed as the light amount control. A part discriminating device according to any one of claims 1 to 11, comprising:
A gender determination device for an object including a determination unit for determining the gender of the person based on a determination result of the determination unit.

Images