JP4798042B2 - Face detection device, face detection method, and face detection program - Google Patents

Description

  The present invention relates to a face detection apparatus that detects a face from an image.

  A face detection device that detects a face from an image is desired to detect the face with high accuracy.

  As a conventional technique in view of such a problem, there is an object detection device that compares an estimated distance with a face calculated from an input image and a measured distance measured by a distance sensor to determine whether it is a face or a non-face ( Patent Document 1).

  However, since this apparatus requires a distance sensor in addition to the face detection unit, the apparatus configuration becomes large and the apparatus cost also increases.

  On the other hand, there is an image processing apparatus that three-dimensionally restores images taken by two cameras (see Patent Document 2). This apparatus detects a face using the three-dimensional information of the image.

However, since this apparatus requires two cameras, the apparatus configuration becomes large and the apparatus cost increases.
JP-A-2005-78376 JP 2005-301833 A

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a technique for executing highly accurate face detection processing with a simple configuration.

  In order to achieve the above object, the present invention employs the following configuration.

  The face detection apparatus according to the present invention includes an imaging unit, a face candidate detection unit that detects a face candidate from an image captured by the imaging unit, and a determination criterion that can be different for each position coordinate in the image. And a criterion storage means for storing the criterion for determining whether the face candidate is a face or a non-face, and whether the face candidate is a face or a non-face And non-face exclusion means for outputting a face candidate excluding the non-face as a face.

  The determination criterion set in advance in the present invention can be different for each position coordinate in the image. Since the non-face exclusion unit of the present invention can determine whether a face candidate is a face or a non-face using a determination criterion suitable for each position coordinate, a face detection process in which the entire image is performed with the same reference Accuracy can be improved. Further, while the conventional technique requires a distance sensor and a second camera, the present invention does not require an additional device, and thus has a simple configuration.

  The determination criterion includes information for defining a range of a face size to be determined as a face, and the non-face exclusion unit compares the determination criterion with the face size of the face candidate to thereby determine the face candidate. It is good to determine whether is a face or a non-face. As a result, only face candidates having an appropriate face size for each position coordinate are detected as faces, so that false detection can be reduced and accuracy can be improved.

  The face candidate detection means has a plurality of face detectors corresponding to different face orientations, and the face detector outputs a score representing the face-likeness of the face candidate together with the face candidate, and the determination criterion Includes information for defining a score range to be determined as a face for each face orientation, and the non-face exclusion means compares the determination criterion corresponding to the face orientation of the face candidate with the score of the face candidate. By doing so, it may be determined whether the face candidate is a face or a non-face. As a result, only face candidates in an appropriate direction for each position coordinate are detected as faces, so that false detection can be reduced and accuracy can be improved.

  The determination criterion storage unit further stores partial region information for dividing an image captured by the imaging unit into a plurality of partial regions and illumination information for each partial region, and the face candidate detection unit includes the face candidate detection unit, Before detecting a face candidate from an image, it is preferable to perform illumination correction based on the illumination information for each partial region. As a result, even when the illumination environment is different for each partial region, the illumination environment of the entire image can be made uniform by performing the illumination correction as a preprocessing, so that false detection can be reduced and accuracy can be improved.

  The determination criterion storage unit may store different illumination information for each time zone, and the face candidate detection unit may determine illumination information used for illumination correction based on the shooting time of the image. Thereby, since it can respond to the change of the illumination environment for every time slot | zone, reduction of false detection and improvement of accuracy can be aimed at.

  Further, the present invention may be regarded as a face detection device having at least a part of the above means, a face detection method including at least a part of the above process, a face detection program for realizing such a method, and It can also be understood as a storage medium storing a program. Each of the above means and processes can be combined with each other as much as possible to constitute the present invention.

  According to the present invention, when executing the face detection process, a determination criterion that can be changed for each position coordinate in the image can be set in advance, and an additional device is not required. It is possible to perform highly accurate face detection processing.

  Exemplary embodiments of the present invention will be described in detail below with reference to the drawings.

<First Embodiment>
In the first embodiment, an example in which the present invention is applied to a monitoring device for monitoring a passerby or an intruder is given. In this type of monitoring apparatus, as shown in FIG. 1, the imaging unit 11 is fixed to a ceiling, a wall, or the like that is higher than a person's head, and in many cases, a captured image overlooking the monitoring area is obtained. Under this circumstance, the image captured by the imaging unit 11 tends to appear as a small face at the top of the image and a large face at the bottom of the image as shown in FIG. Focusing on such a tendency, in this embodiment, the range of the face size to be determined as a face is changed according to the position in the image. Note that the present invention can be preferably applied to a fixed camera such as the monitoring apparatus of the present embodiment, but the present invention is not limited to a fixed camera and can be applied to any photographing apparatus.

<Device configuration>
FIG. 3 is a block diagram showing a functional configuration of the face detection apparatus according to the embodiment of the present invention. The face detection apparatus includes an outline, an imaging unit 11, an image storage unit 12, a face candidate detection unit 13, a non-face determination unit 14, a display unit 15, a result storage unit 16, and a determination criterion storage unit 17. In the present embodiment, these functional elements are realized by an arithmetic processing unit of a computer executing software (program) and controlling hardware resources such as a camera, a memory, and a display as necessary. However, these functional elements may be configured by a dedicated chip.

  As the imaging unit 11, a digital camera including an optical system and an imaging element (CCD, CMOS sensor, etc.) can be preferably applied.

  The image storage unit 12 is a storage device that temporarily stores an image to be processed. As the storage device, any specific technique such as a volatile memory or a nonvolatile memory may be applied.

  The face candidate detection unit 13 has a function of detecting a face candidate from an image by image processing. Further, the face candidate detection unit 13 outputs the face size (face width) of the face candidate and the position coordinate data in the image as the detection result. Any technique of existing face detection processing may be applied to the face candidate detection processing by the face candidate detection unit 13. For example, a method for detecting a face by template matching using a reference template corresponding to the outline of the entire face, a method for detecting a face by template matching based on a facial organ (eyes, nose, ears, etc.), chroma key processing Detecting vertices such as the head, detecting a face based on the vertices, detecting a region close to the skin color, detecting that region as a face, and learning with a teacher signal using a neural network And a method for detecting a face-like area as a face.

  The non-face determination unit 14 determines whether the face candidate detected by the face candidate detection unit 13 is a face or a non-face using a determination criterion corresponding to the detection position of the face candidate, and removes the non-face This is a function to output the face candidate as a face.

  As the display unit 15, any specific technique such as a liquid crystal display may be applied.

  The result storage unit 16 is a storage device that stores the result output as a face from the non-face determination unit 14. As the storage device, any specific technique such as a nonvolatile memory or a hard disk may be applied.

  The determination criterion storage unit 17 is a storage device that stores determination criteria. As the storage device, any specific technique such as a nonvolatile memory or a hard disk may be applied.

  The determination criterion is for determining whether the face candidate detected by the face candidate detection unit 13 is a face or a non-face. In addition, the determination criterion in the present embodiment can be different for each position coordinate in the image.

  An example of the determination criterion in the present embodiment is shown in FIG. As described above, the image photographed by the imaging unit 11 of the present embodiment has a tendency that the face at the top of the image appears small and the face at the bottom of the image appears large as shown in FIG. Therefore, in the example of FIG. 4, partial area information for dividing the entire image into three partial areas 21 to 23, “upper part of image”, “middle part of image”, and “lower part of image”, as information regarding the position coordinates in the image. Set. Here, a quadrangular partial area is set, and the partial area is defined by two coordinates of a start point coordinate and an end point coordinate of a vertex located at the diagonal of the quadrangle. The shape of the partial area is not a quadrangle, but may be a polygon or a circle. The partial area information may be in any format as long as the shape, position, and size of the partial area can be defined. In addition, information for defining a range of face size to be determined as a face is set as a determination criterion. In the example of FIG. 4, the face size ranges “5 to 20” in the partial area 21 at the top of the image, “20 to 35” in the partial area 22 in the middle of the image, and “35 to 50” in the partial area 23 at the bottom of the image. Is set. Note that the face size range is not a numerical value corresponding to the face size (such as a numerical value corresponding to the upper limit of the face size and a numerical value corresponding to the lower limit), but attributes such as “large”, “medium”, and “small”. It may be specified.

<Face detection function>
The function and processing flow of the face detection apparatus of this embodiment will be described with reference to the flowcharts of FIGS.

  The face detection apparatus according to the present embodiment executes the face detection function at regular time intervals. When the face detection function is activated, an image captured by the imaging unit 11 is captured for use in face detection processing (step S11). The input image is stored in the image storage unit 12 together with shooting time information. If necessary, the user can check the face detection processing image on the display unit 15.

  Next, the face candidate detection unit 13 detects a face candidate (step S12). In the present embodiment, a case where six face candidates 24a to 24f shown in FIG. 2 are detected will be described. Here, the detection positions of the face candidates 24a and 24b belong to the partial area 21, the detection positions of the face candidates 24c and 24d belong to the partial area 22, and the detection positions of the face candidates 24e and 24f belong to the partial area 23. Further, the face size of the face candidate 24a is “44”, the face size of the face candidate 24b is “12”, the face candidate 24c is “25”, the face candidate 24d is “45”, and the face candidate 24e is “7” and the face candidate 24f is “48”. The face candidates 24b, 24c, and 24f are faces, but the face candidates 24a, 24d, and 24e are non-faces (false detection).

  Then, the non-face determination unit 14 reads the partial region information and the determination criterion set in advance for each partial region from the determination criterion storage unit 17 (step S13), and determines whether the face candidate is a face or a non-face. A determination is made using a determination criterion corresponding to the detection position of the face candidate, and the face candidate excluding the non-face is output as a face (step S14). In the example of FIG. 2, a face having a size such as the face candidates 24 a, 24 d, and 24 e cannot exist at each position coordinate due to the tendency of the image. By eliminating the face candidates of such an impossible size in the process of step S14, in the present embodiment, it is possible to reduce false detection and to perform highly accurate face detection processing.

  FIG. 6 shows a specific example of the process of step S14.

  First, the non-face determination unit 14 captures the result of step S12 (in this embodiment, position coordinate data and face size data of a face candidate) (step S21). The face size data in this embodiment is data representing the face size, and the larger the face, the larger the value.

  Next, the non-face determination unit 14 selects one of the face candidates detected in step S12 (step S22), and specifies a partial region to which the detected position of the face candidate belongs from the position coordinate data of the face candidate (step S22). S23).

  Then, it is determined whether the face candidate is a face or a non-face by comparing the determination criterion set for each partial region with the face size of the face candidate (step S24). Specifically, when the face size of the face candidate is within the determination criterion set in the partial region to which the face candidate belongs, the non-face determination unit 14 sets the face candidate as a face and sets the other face as a non-face. For example, the detection position of the face candidate 24a belongs to the partial area 21, and the face size is “44”. However, in the determination criterion, “5 to 20” is set in the partial area 21, and the face size “44” of 24a is outside the range of the determination criterion and thus is not a face (the same reason for 24d and 24e). From the non-face). On the other hand, the detection position of 24c belongs to the partial region 22, and the face size is “25”. In addition, in the determination criterion, “20 to 35” is set in the partial region 22, and the face size “25” of 24c is within the range of the determination criterion, so it is regarded as a face (24b and 24f are also for the same reason). Face).

If the face candidate is a face (step S24; YES), the process proceeds to step S26. If the face candidate is a non-face (step S24; NO), the face candidate is excluded from the face candidates detected in step S12 (step S25), the process proceeds to step S26.

  In step S <b> 26, the non-face determination unit 14 determines whether there is a face candidate that is not used for determining whether it is a face or a non-face among the face candidates detected in step S <b> 12. When it is determined that there is another face candidate (step S26; YES), the face candidate is selected, and it is determined whether the face candidate is a face or a non-face (steps S22 to S24). If it is determined whether all face candidates detected in step S12 are faces or non-faces, the process of step S14 is terminated, and the process proceeds to step S15.

  In step S15, the determination result of the non-face determination unit 14 is stored in the result storage unit 16 as a face detection process result. The information of the face detection process result stored in the result storage unit 16 is preferably an image used for the face detection process, its photographing time, the number of detected faces, and information on the detection position. The user can check the face detection processing result on the display unit 15 by inputting the photographing time and the display method of the face detection processing result according to the purpose. The display method may be the total number of detected faces at the input shooting time, or may display the image shot at the shooting time and mark the detection position.

  As described above, in this embodiment, first, the face candidate detection unit 13 detects a face candidate from the image, and determines whether the non-face determination unit 14 is a face or a non-face. In addition, as a determination criterion of the present embodiment, a determination criterion is set such that a face candidate having a small face size is determined as a face in the upper part of the face candidate and a face candidate having a large face size is determined as a face in the lower part of the image. ing. As a result, only face candidates having an appropriate face size are detected as faces for each partial region in the image, so that it is possible to execute a face detection process with higher accuracy than non-face determination based on the same reference.

  In this embodiment, the face width is used as the face size, but the face length, area, or another parameter obtained by combining them may be used.

In the present embodiment, the case where the partial region information and the determination criterion for each partial region are set in the determination criterion storage unit 17 has been described. However, various methods such as setting the determination criterion by a function as shown in FIG. Is applicable. In FIG. 7, the vertical axis corresponds to the position coordinates in the vertical direction of the image, and the horizontal axis corresponds to the face size. For example, when a face candidate is detected at position coordinates Y11, the face candidate has a face size,
Z11 ≦ face size <Z12
Is determined as a face.

<Second Embodiment>
In the second embodiment, a face detection apparatus will be described in a situation where the imaging unit 11 is at the intersection of the T-shaped passage and there are passages in the front and left and right directions of the imaging unit 11 (FIG. 8). . Under this circumstance, the image captured by the imaging unit 11 tends to show a larger face at the lower part of the image, as in the first embodiment. Furthermore, in the upper part of the image, people pass through the front passage, so there are many faces facing the front. In the lower part of the image, people passing through the left and right passages appear, so there is a tendency for many left and right faces to appear (FIG. 9). Focusing on such a tendency, in the present embodiment, the face size range to be determined as a face and the face orientation are changed according to the position in the image.

  Hereinafter, the configuration and processing different from those of the first embodiment will be mainly described.

<Device configuration>
The face candidate detection unit 13 according to the present embodiment includes a plurality of face detectors (for example, front, left, right, left diagonal, and right diagonal five) corresponding to different face orientations. Further, the face candidate detection unit 13 performs face candidate detection processing using all face detectors at one position coordinate. When each face detector detects a face candidate, the face candidate and the face candidate are detected. Output the score. The score in the present embodiment is a value representing the likelihood of a face, and the face candidate that is more likely to be a face has a higher value.

  Next, an example of the determination criterion in the present embodiment is shown in FIG. As described above, the image photographed by the imaging unit 11 of the present embodiment tends to have a larger face as the lower part of the image as shown in FIG. Many faces are shown, and people in the middle of the image pass through the vicinity of the intersection between the front passage and the left and right passages, so there are many faces that appear diagonally left and right, and people in the lower part of the image pass through the left and right passages. , There is a tendency to see a lot of left and right faces. Therefore, in the example of FIG. 10, as information regarding the position coordinates in the image, the entire image is divided into three partial areas 31 to 33, “upper part of image”, “middle part of image”, and “lower part of image”. Partial area information is set (the definition of the partial area here is the same as in the first embodiment). Further, information for defining a face size range and a score range to be determined as a face for each face direction is set as a determination criterion. In the example of FIG. 10, “5 to 20” and “front”; 60, left and right oblique; 80, left and right; 100 in the upper part of the image, and “20 to 35” and “front; 80, left and right oblique; 60, left and right; 80 ", and a partial size 33 of the lower part of the image," 35-50 "and" front; 100, left and right oblique; 80, left and right; 60 ". Is set. The non-face determination unit 14 determines the face candidate as a face when the score output from the face detector is larger than a preset threshold value. Therefore, in the partial area 31, the front face threshold is low, and thus the front face is easily determined as a face. Similarly, in the partial area 32, the left and right oblique faces are easily determined as faces, and in the partial area 33, the left and right faces are easily determined as faces. Note that the face orientation may be defined by a numerical value corresponding to the face orientation or an identifier of the face detector, instead of attributes such as “front”, “left-right diagonal”, and “left-right”.

<Face detection function>
Since the functions of the face detection apparatus of this embodiment are the same as those of the first embodiment except for the processing of step S14, the description thereof is omitted, and here, step S14 of this embodiment is performed according to the flowchart of FIG. The process will be described in detail. In the present embodiment, a case will be described in which three face candidates 34a to 34c shown in FIG. 9 are detected in step S12. Here, the face candidates 34a and 34c are faces, but the face candidate 34b is a non-face (false detection).

  First, the non-face determination unit 14 captures the result of step S12 (in this embodiment, position coordinate data of face candidates, face size data, face orientation of face candidates and their scores) (step S21).

  Next, the non-face determination unit 14 selects one of the face candidates detected in step S12 (step S22), and specifies a partial region to which the detected position of the face candidate belongs from the position coordinate data of the face candidate (step S22). S23). In the example of FIG. 9, the detection position of the face candidate 34 a belongs to the partial area 31, the detection position of the face candidate 34 b belongs to the partial area 32, and the detection position of the face candidate 34 c belongs to the partial area 33.

Then, it is determined whether the face candidate is a face or a non-face from the determination criteria set for each partial region, the face size, face orientation, and score of the face candidate (step S24). For example, the face direction of the face candidate 34a and its score are “front; 90, left diagonal; 70, right diagonal; 40, left; −, right; −”, and the face candidate 34b is “front; 40, left diagonal; 50, right diagonal; 30, left; 70, right; 30 ", and the face candidate 34c is"front;-, left diagonal;-, right diagonal;-, left; 100, right;-"(here , “-” Means that no face candidate was detected.) In step S24, since the detection position of the face candidate 34a belongs to the partial region 31, it is compared with “front; 60, left and right diagonal; 80, left and right; 100”. Since the score of “front” exceeds the threshold, “face” (Similarly, 34b is "non-face" and 34c is "face"). Since the face sizes of the face candidates 34a to 34c are within the face size range set in the partial area to which each of the face candidates 34a to 34c belongs, the face candidate 34b is regarded as a non-face according to the determination criterion only for the face size range (first embodiment). It cannot be judged. However, in the present embodiment, the face candidate 34b can be determined as a non-face by setting a threshold value for each face direction as a determination criterion.

  When the face candidate is a face (step S24; YES), the process proceeds to step S26. When the face candidate is a non-face (step S24; NO), the face candidate is excluded from a plurality of face candidates (step S25), and step Proceed to S26.

  In step S <b> 26, the non-face determination unit 14 determines whether there is a face candidate that is not used for determination among the plurality of face candidates detected in step S <b> 12. When it is determined that there is another face candidate (step S26; YES), the face candidate is selected, and it is determined whether the face candidate is a face or a non-face (steps S22 to S24). If it is determined whether all face candidates detected in step S12 are faces or non-faces, the process of step S14 is terminated, and the process proceeds to step S15.

  In step S15, the determination result of the non-face determination unit 14 is stored in the result storage unit 16 as a face detection process result. The information of the face detection process result stored in the result storage unit 16 is preferably an image used for the face detection process, its photographing time, the number of detected faces, and information on the detection position. The user can check the face detection processing result on the display unit 15 by inputting the photographing time and the display method of the face detection processing result according to the purpose. The display method may be the total number of detected faces at the input shooting time, or may display the image shot at the shooting time and mark the detection position.

  As described above, in this embodiment, first, the face candidate detection unit 13 detects a face candidate from the image, and determines whether the non-face determination unit 14 is a face or a non-face. In addition, as a determination criterion of the present embodiment, among the face candidates, a face candidate with a small face size and facing the front is easily determined as a face in the upper part of the image, and a face candidate with a large face size and left and right facing is determined as a face in the lower part of the image. Judgment criteria that make it easy to do are set. As a result, only face candidates with appropriate face sizes and face orientations are detected as faces for each partial region in the image, so that it is possible to perform face detection processing with higher accuracy than non-face determination based on the same reference.

  In the present embodiment, the case where the partial area information and the determination standard for each partial area are set in the determination reference storage unit 17 has been described. However, the determination standard is set by a function as shown in FIGS. Various methods can be applied. In FIG. 11, the vertical axis corresponds to the position in the vertical direction of the image, and the horizontal axis corresponds to the threshold value. A function is set for each face detector. FIG. 11 means that when a face candidate is detected at the position coordinate Y21, the front face threshold is Z22, the left and right oblique face threshold is Z21, and the left and right face threshold is Z23. In the present embodiment, information that defines the face size range and information that defines the face range for each face orientation are set in the same partial area. May be set.

  In the present embodiment, the method for determining whether or not the face candidate is determined in step S12 is not specifically described, but a score having the same definition as the score in step S14 may be used in step S12. In that case, the threshold used for determining whether or not the face is a candidate may be set lower than the threshold used for determining whether or not the face is a candidate. Thereby, it is possible to prevent face detection omission.

<Third Embodiment>
In the third embodiment, a face detection apparatus will be described in a situation where the imaging unit 11 is installed so as to face indoors and outdoors, and an image captured by the imaging unit 11 includes both indoors and outdoors. (FIG. 12). Under this circumstance, the image captured by the imaging unit 11 tends to show a larger face at the lower part of the image, as in the first embodiment. Further, as shown in FIG. 13, the lighting environment tends to be different between the indoor face and the outdoor face (for example, when the outdoor is brighter than the indoor, the outdoor face appears bright, and the indoor face becomes dark due to backlight. Etc.) Such a difference in lighting environment adversely affects the face detection process. Focusing on such a tendency, in this embodiment, the range of the face size to be determined as a face is changed according to the position in the image. Further, the entire image is divided into a plurality of partial areas, and illumination correction is performed for each partial area in order to align the illumination environment of the entire image as a pre-process for the face candidate detection process.

  Hereinafter, the configuration and processing different from those of the first embodiment will be mainly described.

  The face candidate detection unit 13 in the present embodiment has a function of performing illumination correction as preprocessing in addition to a function of detecting face candidates from an image by image processing.

  The determination criterion storage unit 17 in the present embodiment is a storage device that further stores, in addition to the determination criterion, partial region information for dividing an image into a plurality of partial regions and illumination information for each partial region.

  Next, FIG. 14 shows an example of determination criteria, partial area information, and illumination information for each partial area in the present embodiment. As described above, the image captured by the imaging unit 11 of the present embodiment tends to have a larger face at the lower part of the image as shown in FIG. 13, and the illumination environment includes an indoor face and an outdoor face. Tend to be different. Therefore, in the example of FIG. 14, as the partial area information, information for dividing the entire image into three partial areas 41 to 43 of “upper part of image”, “middle part of image”, and “lower part of image” in the determination criterion storage unit 17. (The definition of the partial area here is the same as in the first embodiment). In addition, as determination criteria (information for defining the face size range) and illumination information, “5 to 20” and “morning; bright, daytime; bright, night; "20-35" and "morning; dark, daytime; dark, night; bright" in the partial area 42 of the image, "35-50" and "morning; dark, daytime; dark, night; bright" in the partial area 43 below the image Is set. The illumination information may be defined not by attributes such as “bright” and “dark” but by numerical values corresponding to the illumination environment and parameters used for illumination correction.

<Face detection function>
The function and processing flow of the face detection apparatus of this embodiment will be described with reference to the flowchart of FIG.

  The face detection apparatus according to the present embodiment executes the face detection function at regular time intervals. When the face detection function is activated, an image captured by the imaging unit 11 is captured for use in face detection processing (step S31). The input image is stored in the image storage unit 12 together with shooting time information. If necessary, the user can check the face detection processing image on the display unit 15.

  Next, the face candidate detection unit 13 reads the partial region information and the illumination information set in advance for each partial region from the determination criterion storage unit 17 (step S32), and performs illumination correction based on the illumination information for each partial region. (Step S33). Here, the face candidate detection unit 13 adjusts the brightness and contrast so as to brighten the image when the illumination information is set to “dark”, and darkens the image when set to “bright”. Adjust the brightness, contrast, etc. to make the entire image as bright as possible. That is, when the face candidate detection unit 13 performs illumination correction for each partial region, the image of FIG. 13 becomes an image as shown in FIG.

  Then, the face candidate detection unit 13 detects a face candidate from the image that has been subjected to illumination correction (step S34).

  Next, the non-face determination unit 14 reads the partial region information and the determination criterion set in advance for each partial region from the determination criterion storage unit 17 (step S35), and determines whether the face candidate is a face or a non-face. Judgment is made using a judgment criterion corresponding to the detection position of the face candidate, and the face candidate excluding the non-face is output as a face (step S36).

  In step S37, the determination result of the non-face determination unit 14 is stored in the result storage unit 16 as the face detection processing result. The information of the face detection process result stored in the result storage unit 16 is preferably an image used for the face detection process, its photographing time, the number of detected faces, and information on the detection position. The user can check the face detection processing result on the display unit 15 by inputting the photographing time and the display method of the face detection processing result according to the purpose. The display method may be the total number of detected faces at the input shooting time, or may display the image shot at the shooting time and mark the detection position.

  As described above, in this embodiment, before the face candidate detection unit 13 detects a face candidate from the image, illumination correction is performed for each partial region. Thereby, since the illumination environment of the whole image can be prepared, a highly accurate face detection process can be performed rather than the face detection process which does not perform illumination correction.

  In this embodiment, the information for defining the face size range and the illumination information are set in the same partial area, but the partial area may be set for each of various types of information. The information for defining the face size range may be a function as described in the first embodiment.

  The lighting information may be set by the user, or the lighting information (such as weather information) may be automatically acquired and set via the Internet or the like. For example, in this embodiment, illumination information for each time zone is set for each partial region. However, when the lighting state varies depending on the time zone depending on the weather (for example, the outdoors is bright on a clear day but dark on a cloudy day), it is necessary to set illumination information each time. In such a case, the lighting information (weather information) is automatically acquired via the Internet, and the brightness of the image is corrected accordingly. Appropriate lighting information can be set for each.

  In the first to third embodiments, the size of the face detected in advance is determined for each partial region of the image, and the direction of the face detected in advance is determined in advance. This can be preferably applied when the lighting environment of the face is determined. For example, in the third embodiment, an example in which a difference between indoor and outdoor illumination environments is assumed is described, but the present invention can be preferably applied to a case where a part of an image is illuminated with illumination.

FIG. 1 is a diagram illustrating an installation mode of an imaging unit according to the first embodiment. FIG. 2 is a diagram illustrating an example of an image captured by the imaging unit according to the first embodiment. FIG. 3 is a block diagram illustrating a functional configuration of the face detection apparatus. FIG. 4 is a diagram illustrating an example of the determination criterion according to the first embodiment. FIG. 5 is a flowchart showing the flow of processing of the face detection apparatus in the first embodiment and the second embodiment. FIG. 6 is a flowchart showing the process flow of step S14. FIG. 7 is a diagram illustrating an example of determination criteria. FIG. 8 is a diagram illustrating an installation mode of the imaging unit in the second embodiment. FIG. 9 is a diagram illustrating an example of an image captured by the imaging unit according to the second embodiment. FIG. 10 is a diagram illustrating an example of a determination criterion according to the second embodiment. FIG. 11 is a diagram illustrating an example of a determination criterion. FIG. 12 is a diagram illustrating an installation state of the imaging unit in the third embodiment. FIG. 13 is a diagram illustrating an example of an image photographed by the imaging unit according to the third embodiment. FIG. 14 is a diagram illustrating an example of partial region information, determination criteria, and illumination information in the third embodiment. FIG. 15 is a flowchart illustrating a process flow of the face detection apparatus according to the third embodiment. FIG. 16 is a diagram illustrating an example of an image after illumination correction.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Image pick-up part 12 Image memory | storage part 13 Face candidate detection part 14 Non-face determination part 15 Display part 16 Result storage part 17 Judgment reference | standard storage part 21-23, 31-33, 41-43 Partial region 24a-24f, 34a-34c Face Candidate

Claims (5)

Imaging means;
Face candidate detection means for detecting a face candidate from an image taken by the imaging means;
A determination criterion for determining whether a face is a face or a non-face based on a face size value of a face candidate in the photographed image, and the face is determined as the vertical position coordinates in the photographed image are upward. Determination criterion storage means for storing determination criteria such that both the upper limit value and the lower limit value of the face size to be reduced are reduced ;
Non-face determination means for determining whether the face candidate is a face or a non-face by comparing the determination criterion and the face size of the face candidate;
Non-face removing means for outputting face candidates excluding face candidates determined as non-faces as faces,
A face detection apparatus comprising: The determination criterion storage means further stores partial area information for dividing an image photographed by the imaging means into a plurality of partial areas and illumination information for each partial area,
The face candidate detection means performs illumination correction based on the illumination information for each partial area before detecting a face candidate from the image.
The face detection apparatus according to claim 1 . The determination criterion storage means can store different illumination information for each time zone,
The face detection apparatus according to claim 2 , wherein the face candidate detection unit determines illumination information used for illumination correction based on a shooting time of the image. A computer that performs face detection processing for detecting a face from an image captured by an imaging unit,
Detecting a face candidate from the image taken by the imaging means;
A determination criterion for determining whether a face is a face or a non-face based on a face size value of a face candidate in the photographed image, and the face is determined as the vertical position coordinates in the photographed image are upward. Judgment criteria are stored so that the upper and lower limits of the face size to be reduced are both small ,
Determining whether the face candidate is a face or a non-face by comparing the criterion and the face size of the face candidate ;
A face detection method of outputting face candidates excluding face candidates determined as non-faces as faces. To a computer that performs face detection processing for detecting a face from an image taken by an imaging means,
Detecting a face candidate from an image photographed by the imaging means;
A determination criterion for determining whether a face is a face or a non-face based on a face size value of a face candidate in the photographed image, and the face is determined as the vertical position coordinates in the photographed image are upward. Storing a determination criterion such that both the upper limit value and the lower limit value of the face size to be reduced are reduced ;
A step of determining by the face candidate whether a non-face or a face, and compares the face size of the face candidate and the criterion,
Outputting face candidates excluding face candidates determined as non-faces as faces;
Face detection program for running.

Images