JP2004259215A - Face detection system and its method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a robust face detection system capable of accurately detecting a face even when a person greatly changes his/her face direction to the right/left or wears glasses. <P>SOLUTION: A face direction computing system 1 is provided with a part detecting a right nose hole and a left nose hole by a resolution filter and a part detecting a face by a plurality of face direction templates inside the system. The face direction computing system is constructed of a nose detection part 2, an image input part 3, a face detection part 10, a face direction template storage part 11, and a face direction computing part 12. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a face detection system based on a face image and a method for appropriately cutting out a face image of a person from an image for a wide range of face orientation angles.
[0002]
[Prior art]
(1) Application of face detection
Face detection based on an image is a technique for examining whether or not a human face exists in two-dimensional image data, and when present, finding the position of the face, and can be used in various fields. However, it is difficult to detect a face in various states (various face orientations, various lighting conditions, various expressions, etc.) robustly in real time using a computer from images taken by a camera. Even with the state-of-the-art technology, it cannot be said that sufficient performance has yet been achieved. For this reason, it is considered that some face detection application fields have not yet become apparent.
[0003]
At present, face detection technology is mainly used in face recognition systems as applications to security systems such as door locks. However, if the performance of face detection technology can be further improved and face detection and face direction calculation can be performed for various face orientations, drivers of vehicles, audiences of exhibits, operators of console terminals, games Face detection technology can be applied to a terminal operator or the like.
[0004]
For the driver of the vehicle, face detection and face direction calculation can detect inattentive driving hindrance and sound an alarm, or present appropriate information for safe operation according to the face direction. .
[0005]
For the audience of the exhibit, it is possible to identify the exhibition that the audience is paying attention to by face detection and face orientation calculation, and to automatically present information on the exhibition.
[0006]
For the operator of the console terminal, the work efficiency can be improved by specifying the display window to be operated by face detection and face direction calculation and moving the cursor.
[0007]
For game terminal operators, face detection and face direction calculation can be used to change the viewpoint of the game environment, determine shooting targets, and use swinging motions to help communicate with characters. it can.
[0008]
(2) Conventional face detection technology
As one of the promising conventional face detection technologies, there is a technology that combines detection of a feature point of a face such as an eye and a nose with a separation filter and detection of a pattern of an area including the feature point (for example, see Patent Literature 1, Non-Patent Literatures 1 and 2). Although the method of this combination is complicated, the face is detected only by the feature point detection using the separability filter, or the face is detected only by the comparison with the pattern of the entire face area. More robust face detection than the method can be performed. However, this method sometimes failed in face detection for a large change in the face direction.
[0009]
As another method of face detection, there is a method in which face direction templates having different face directions are registered in advance and used (for example, see Non-Patent Document 3). Although this method can detect a face even with a large change in the face direction, the accuracy of feature point detection is inferior to a method using a separability filter.
[0010]
Furthermore, in the conventional face detection technology, eyes are used as feature points of the face, and face detection sometimes fails when wearing spectacles.
[0011]
[Patent Document 1]
Japanese Patent No. 3279913
[0012]
[Non-patent document 1]
Yamaguchi, Fukui, "Face Recognition System" Smartface "Robust to Changes in Face Direction and Expression", Transactions of the Institute of Electronics, Information and Communication Engineers, June 2001, J84-D-II, No. 6, p. 1045-1052
[0013]
[Non-patent document 2]
Fukui, Yamaguchi, "Face feature point extraction by combination of shape extraction and pattern matching", IEICE Transactions, August 1997, J80-D-II, No. 8, p. 2170-2177
[0014]
[Non-Patent Document 3]
A. Pentland, and two others, View-based and modular eigenspaces for face recognition, Proceedings of IEEE Computer Computer Society, Union of Computers, Republic of the Union, United States, USA 84-91
[0015]
[Problems to be solved by the invention]
The present invention has been made in consideration of the above circumstances, and its object is to provide a robust and high-precision face detection system and a highly accurate face detection system, even when the face direction of a person changes significantly from side to side or wearing glasses. It is to provide a method.
[0016]
Note that “robustness” refers to “a property that is unlikely to cause a large failure even under bad conditions” when performing face detection processing or the like using an image. For example, when the detection process is unlikely to fail even if the face image is deformed, the lighting conditions change, or the color of the face image changes due to makeup, the method is referred to as a “robust face detection method”.
[0017]
The present invention mainly provides an excellent technique regarding the robustness of the face detection processing when the face is largely turned sideways.
[0018]
[Means for Solving the Problems]
The invention according to claim 1 is a face detection system for detecting a person's face from an image, comprising: an image acquisition unit that acquires the image; and a separation filter that extracts a feature point candidate from the acquired image using a separation degree filter. Degree detection means, face direction template storage means storing a plurality of face direction templates, right nostril template storage means storing a right nostril template, left nostril template storage means storing a left nostril template, and the extracted features. Right nostril detection means for performing right nostril detection using the point candidates and the stored right nostril template, left nostril detection means for performing left nostril detection using the extracted feature point candidates and the stored left nostril template, Based on the detection state of the right nostril, or the detection state of the left nostril, a predetermined Select the orientation template, a face detection system characterized by comprising a face detection unit for performing face detection using the selected face orientation template and the acquired image.
[0019]
The invention of claim 2 comprises a nostril template storage means for storing a nostril template, and a nostril detection means for performing a nostril detection using the extracted feature point candidate and the stored nostril template, The face detecting means selects a predetermined face direction template from the plurality of stored face direction templates based on the detection state of the both nostrils, the detection state of the right nostril, or the detection state of the left nostril. The face detection system according to claim 1, wherein face detection is performed using the selected face orientation template and the acquired image.
[0020]
The face detecting means may limit the face search range to a predetermined range based on the detected right nostril or left nostril, and the face direction may be determined in the limited face search range. The face detection system according to claim 1, wherein the face is detected by moving the template.
[0021]
The invention according to claim 4 is characterized in that the face detecting means limits the face search range to a predetermined range based on the detected both nostrils, right nostrils, or left nostrils, and The face detection system according to claim 2, wherein the face detection is performed by moving the face orientation template.
[0022]
The invention according to claim 5 is characterized in that the face detection means performs face direction calculation based on the similarity between the acquired image and a plurality of face direction templates. 2. A face detection system according to claim 1.
[0023]
The invention of claim 6 is a face detection method for detecting a person's face from an image, comprising: an image obtaining step of obtaining the image; and a separation step of extracting a feature point candidate from the obtained image by a separation degree filter. Degree detection step, a right nostril detection step of performing right nostril detection using the extracted feature point candidates and a pre-stored right nostril template, and a left nostril using the extracted feature point candidates and a pre-stored left nostril template. A left nostril detecting step of performing detection, a right nostril detection state, or a predetermined face orientation template is selected from a plurality of pre-stored face orientation templates based on the left nostril detection state; A face detection method comprising: a face detection step of performing face detection using an orientation template and the acquired image.
[0024]
The invention according to claim 7 is a program for realizing a face detection method for detecting a face of a person from an image by a computer, the program comprising: an image acquisition function for acquiring the image; and a feature point candidate extraction from the acquired image. A separation degree detection function performed by a separation degree filter, a right nostril detection function of performing right nostril detection using the extracted feature point candidates and a pre-stored right nostril template, and a left nostril stored in advance with the extracted feature point candidates. Based on the left nostril detection function for performing left nostril detection using a template, the right nostril detection state, or the left nostril detection state, a predetermined face orientation template is selected from a plurality of pre-stored face orientation templates. And a face detection function for performing face detection using the selected face orientation template and the acquired image. A gram.
[0025]
The inventions of claims 1, 6, and 7 use a single nostril template and a face orientation template.
[0026]
The inventions of claims 1, 6, and 7 combine a separation filter, right nostril detection using a right nostril template, left nostril detection using a left nostril template, and face detection using a plurality of face orientation templates. For this reason, even if the face direction of a person greatly changes left and right, a robust and highly accurate face detection system can be provided.
[0027]
A second aspect of the present invention is obtained by adding both nostrils detection processing to the first aspect of the present invention.
[0028]
The invention of claim 2 provides a right nostril detection using a separation filter and a right nostril template, a left nostril detection using a left nostril template, a double nostril detection using a both nostril template, and a face using a plurality of face orientation templates. Since detection is combined, a robust and high-accuracy face detection system can be provided even when the face direction of a person greatly changes from side to side.
[0029]
According to a third aspect of the present invention, a face search range setting process is added to the third aspect of the present invention.
[0030]
According to the third aspect of the present invention, since the face search range is appropriately set by the right nostril detection and the left nostril detection using the separability filter, even if the face direction of the person greatly changes from side to side, it is robust and high. An accurate face detection system can be provided.
[0031]
According to a fourth aspect of the present invention, a face search range setting process is added to the second aspect of the present invention.
[0032]
According to the fourth aspect of the present invention, since the face search range is appropriately set by the right nostril detection, the left nostril detection, and the both nostrils detection using the degree of separation filter, even if the face direction of the person greatly changes from side to side. A robust and highly accurate face detection system can be provided.
[0033]
The invention of claim 5 uses the face detection system of claims 1 to 4.
[0034]
According to the fifth aspect of the present invention, since the face direction calculation is performed using a robust and high-accuracy face detection system, a robust and high-accuracy face direction calculation can be performed.
[0035]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0036]
(1st Embodiment)
(1) Definition of terms
In this specification, a prerequisite term is defined.
[0037]
(1-1) The definition of “left and right” is performed.
[0038]
The definition of left and right is based on the anatomical left and right of a subject (mainly a human face). The right eye in a normal (not upside down) face image taken from the front with a normal camera will be on the left side from us toward the face image when we look at the face image. The same applies to the face orientation and the nostrils. In addition, the left-facing face described in FIG. 3 faces right from us toward the face image. On the other hand, the right-facing face is facing left from us toward the face image. In addition, the right nostrils 23, 33 in FIGS. 2 and 3 are on the left from us in the figure, and the left nostrils 24, 34 are on the right from us in the figure. By defining the left and right as described above, even if the subject is upside down, there is no ambiguity in the definition of the right eye or the right direction.
[0039]
The right and left nostrils are collectively referred to as "single nostrils" for both nostrils.
[0040]
(1-2) Definition of “front direction of face direction”, “upward direction”, “downward direction”, “rightward direction”, “leftward direction”, “upward degree”, and “horizontal degree” is performed.
[0041]
The appearance of the face image captured by the camera changes depending on the posture of the head. Therefore, the posture of the head with respect to the camera (that is, the posture of the face) will be referred to as “facial orientation”.
[0042]
As shown in FIG. 2, the face direction when the face is pointed straight at the camera is referred to as “front direction”.
[0043]
When facing upward from the front, it is referred to as "upward", and the greater the angle change, the greater the "upward degree".
[0044]
When facing downward from the front, it is called "downward".
[0045]
It is called “rightward” when facing rightward rather than frontward, and it is assumed that the greater the angle change, the greater the degree of rightward facing.
[0046]
It is called “leftward” when facing leftward rather than frontward, and the greater the angle change, the greater the degree of leftward facing.
[0047]
As shown in FIG. 4, the rightward degree and the leftward degree are collectively referred to as “lateral degree”.
[0048]
(1-3) Template matching will be described below.
[0049]
"Matching" (for example, see Non-Patent Document 4) means that when there are two images of the same size, they are overlapped and the difference is determined to determine whether they are the same. “Matching” is also called “collation”.
[0050]
“Template matching” is described in, for example, Non-Patent Document 4, “Image Processing Standard Textbook Editing Committee (ed.), Image Processing Standard Textbook, Image Information Education Promotion Association (CG-ARTS), 1997”. A standard image called “template” is prepared in advance, and it is determined whether or not the target image is the same as the template by matching the target image with the template.
[0051]
As a scale for measuring a difference between two images, for example, a sum of absolute values (SAD: Sum of Absolute Difference) is used. In the SAD, a target image having an image size of M × N is I (m, n), (m = 1,..., M, n = 1,..., N), and a template which is a standard pattern prepared in advance is T ( m, n), (m = 1,..., M, n = 1,..., N), the absolute value of the difference I (m, n) −T (m, n) is m = 1,. , N = 1,..., N are obtained by calculating M × N sums.
[0052]
Other measures used include normalized cross-correlation and sum of squared differences. Here, a measure using the subspace method or the incremental code correlation method is also included. That is, if the matching scale is defined, the case where the template used is not the original image itself but a converted one is also included. The scale may be called "similarity" or "difference" in another word. At this time, the smaller the difference between the two images, the larger the value is called “similarity”, and the smaller the difference, the smaller the value is called “dissimilarity”.
[0053]
If the image input to the face detection system has a size larger than M × N, and it is not known in advance where the target image to be compared is included in the image, the user moves both templates in the image while moving the template. Check the similarity of the images. Then, when the maximum similarity is larger than a preset threshold, it is determined that the same one as the template has been detected at the position giving the maximum similarity. If the maximum similarity is smaller than a preset threshold, it is determined that the same image as the template has not been detected in the input image.
[0054]
It may be necessary to detect a target image having an image size larger than the image size of the template. In that case, matching is performed by moving the template in the image and simultaneously reducing the image. For example, there is a case where the image size of the face template for detecting a face is 15 × 15 and the size of the face region in the image is 100 × 100. In this case, both processing for reducing the image at a reduction ratio of 15/100 and processing for moving the template in the image are performed. If the size of the face area is not known in advance, it is necessary to change the reduction ratio, and matching is performed for various values of the reduction ratio. By making the size of the template smaller than the size of the target area in the image, computer processing time and memory amount can be saved.
[0055]
In this way, by the template matching process, it is possible to know whether or not the same thing as the template exists in the image, and if so, its detection position and size. If it is determined that it exists, it is said to have been detected. The detection position is represented by a method convenient for the subsequent processing, such as the center position of the template when it is superimposed on the template, the position of the upper left corner, the lower right corner of the template, and the like.
[0056]
(2) Configuration of face direction calculation system 1
FIG. 1 is a configuration diagram showing a face orientation calculation system 1 according to the first embodiment of the present invention.
[0057]
Upon input of an image, the face direction calculation system 1 finds a face in the image, and outputs the head position, the nose position, and the face direction. It has a feature of performing detection and a portion of performing face detection using a plurality of face orientation templates. As shown in FIG. 1, a nose detection unit 2, an image input unit 3, a face detection unit 10, It comprises a face direction template storage unit 11 and a face direction calculation unit 12.
[0058]
(2-1) Image input unit 3
The image input unit 3 receives an image input from the camera, stores the image input, and sends the image to the nose detection unit 2 and the face detection unit 10 as necessary.
[0059]
(2-2) Nose detector 2
The nose detection unit 2 receives the image input to the image input unit 3, determines whether or not a nose exists in the image, and sends the determination result and information on the detected nose and nostril position to the face detection unit 10. As shown in FIG. 1, as shown in FIG. 1, a separation degree detection unit 4, a right nostril detection unit 5, a right nostril template storage unit 6, a left nostril detection unit 7, a left nostril template storage unit 8, It comprises a detection determination unit 9.
[0060]
The degree-of-separation detection unit 4 receives the image from the image input unit 3, extracts feature point candidates for the face using a degree-of-separation filter, and extracts information and image of the feature point candidates from the right nostril detection unit 5 and the left nostril detection unit 7. Send to
[0061]
The right nostril detection unit 5 receives the information and the image of the feature point candidate from the separation degree detection unit 4, receives the right nostril template from the right nostril template storage unit 6, and based on the information, the right nostril exists in the image. A search is performed to determine if the right nostril exists. Then, the right nostril detection result information is sent to the nose detection determining unit 9.
[0062]
Here, the “right nostril template” is obtained by converting an image of a region near the position of the right nostril as it is or by performing principal component analysis and the like, and is data used in image matching by template matching described above. It is.
[0063]
The right nostril template storage unit 6 stores the right nostril template in advance, and sends the right nostril template to the right nostril detection unit 5 as necessary.
[0064]
The left nostril detection unit 7 receives the information and the image of the feature point candidate from the separation degree detection unit 4, receives the left nostril template from the left nostril template storage unit 8, and has a left nostril in the image based on the information. A search is performed to determine if there is a left nostril. Then, the left nostril detection result information is sent to the nose detection determining unit 9.
[0065]
The left nostril template storage unit 8 stores the left nostril template in advance, and sends the left nostril template to the left nostril detection unit 7 as necessary.
[0066]
Here, the “left nostril template” is obtained by converting an image of a region near the left nostril position as it is or by performing principal component analysis, and is data used in image matching by template matching.
[0067]
The nose detection determination unit 9 receives the right nostril detection result information from the right nostril detection unit 5, receives the left nostril detection result information from the left nostril detection unit 7, comprehensively determines these information, and determines the nose from the image. It is determined whether or not it has been detected, and the determination result and the detected nose and nostril position information are sent to the face detection unit 10.
[0068]
(2-3) Face detection unit 10
The face detection unit 10 receives the image from the image input unit 3, receives the nose detection determination result and the detected nose and nostril position information from the nose detection unit 2, and stores a plurality of face direction templates from the face direction template storage unit 11. receive.
[0069]
Then, using these pieces of information, a face is detected from the image, and a comprehensive determination is made to determine whether a face has been detected. The determination result, similarity information for a plurality of face orientation templates, and detection The obtained face, nose, and nostril position information are sent to the face direction calculation unit 12.
[0070]
The determination as to whether a face has been detected is performed, for example, as follows.
[0071]
If the nose detection determination result is “not detected”, it is determined that the face has not been detected.
[0072]
If the result of the nose detection determination is "detection", matching is performed by moving the face direction template to a range in the vicinity of the detected nose and the position of the nostril. If the maximum similarity with the face orientation template is larger than the set threshold value, it is determined that the face is detected at the position giving the maximum similarity. At this time, if only the right nostril is detected, it is considered that the left-facing face is more likely to be detected than the right-facing face. It is possible to comprehensively determine that the contradiction is caused by erroneous detection, and determine that the face has not been detected.
[0073]
(2-4) Face orientation template storage unit 11
The face orientation template storage unit 11 stores a plurality of face orientation templates in advance, and sends the plurality of face orientation templates to the face detection unit 10 as necessary.
[0074]
(2-5) Face direction calculation unit 12
The face direction calculation unit 12 receives the face detection determination result from the face detection unit 10, similarity information for a plurality of face direction templates, and position information of the detected face, nose, and nostril.
[0075]
Then, the face direction of the detected face is determined based on the similarity information, and the face detection determination result, the position information of the face, the nose and the nostrils, and the face direction information are output to the outside of the system.
[0076]
(3) Operation contents of the face orientation calculation system 1
FIG. 2 is a conceptual diagram for explaining the positional relationship between the face orientation template corresponding region 21, the double nostril template corresponding region 22, the right nostril 23, and the left nostril 24.
[0077]
The face direction of this face image is the front direction, both nostrils are present in the image, and both nostrils can be detected.
[0078]
FIG. 3 is a conceptual diagram for explaining the positional relationship between the face orientation template corresponding region 31, the right nostril template corresponding region 32, the right nostril 33, and the left nostril.
[0079]
The face direction of this face image is to the left, the right nostril is present in the image, and the right nostril can be detected. The left nostril is difficult to see due to hiding, and it is difficult to detect the left nostril.
[0080]
FIG. 4 is a conceptual diagram illustrating the relationship between the degree of lateral orientation of the face and the nostril detectable range.
[0081]
If the face is facing forward, both nostrils and single nostrils can be detected, but as the degree of laterality becomes right, the right nostrils become hidden, and as a result, only the left nostrils can be detected. When the rightward degree further increases, finally the left nostril cannot be detected. Conversely, the left nostrils become hidden as the horizontal degree becomes leftward, so that only the right nostrils can be detected. When the degree of leftward orientation further increases, the right nostril can no longer be detected.
[0082]
Thus, in order to successfully detect the face even if the face is facing right or left, both the left nostril detection and the right nostril detection should be performed when the nostrils that are the feature points of the face are detected. FIG. 4 shows that if the nose is detected by detecting at least one of them, the nose detection and the face detection using the nose can be performed for a wide range of lateral degrees.
[0083]
(4) Face direction calculation processing procedure
FIG. 5 is a flowchart illustrating a processing procedure of the face orientation calculation system 1 according to the present embodiment.
[0084]
First, an image is input (ST1), and the separability detecting unit 4 applies a separability filter to the input image, and determines a plurality of positions on the image where the value of the separability is large and has a maximum value (separation peak). Individuals are extracted (ST2).
[0085]
Next, the right nostril detection unit 5 collates the right nostril template with the image on the separability peak (ST3), and the left nostril detection unit 7 collates the left nostril template with the image on the separability peak (ST4). These steps ST3 and ST4 can be processed simultaneously.
[0086]
Next, the nose detection determination unit 9 determines whether a nose is detected in the image (ST5). If detected, the process proceeds to step ST7, and if not detected, the process proceeds to step ST1 (ST6).
[0087]
In step ST7, a face search range in which the face direction template is moved is set based on the information on the detected nose position. The face search range may be set to a certain range that does not significantly impair the consistency between the detected nose position and the face position, or may be limited to only one face region determined based on the nose position. is there.
[0088]
Next, the face detection unit 10 detects a face from the image, and the face direction calculation unit 12 determines the face direction of the face (ST8).
[0089]
Next, the face direction calculation unit 12 outputs the head position, the nose position, and the face direction (ST9). If the face direction calculation is to be continued, the procedure goes to step ST1, and if the face direction calculation is ended, the face direction calculation is ended. (ST10).
[0090]
(Second embodiment)
FIG. 6 is a configuration diagram showing a face orientation calculation system 101 according to the second embodiment of the present invention. The same parts as those in FIG. Only different parts will be described.
[0091]
In the present embodiment, the detection performance for a front-facing face is improved by utilizing the detection of both nostrils.
[0092]
In particular, when the threshold value for detecting the right nostril or the left nostril is set higher than the threshold value for detecting the both nostrils, a single nostril of the right nostril or the left nostril cannot be detected, but both nostrils may be detected. Therefore, in such a case, it is effective to detect a front-facing face using the detection state of both nostrils. Note that the reason for setting the threshold for detecting the right nostril and the left nostril higher than the threshold for detecting both nostrils is that both nostrils are unique feature point candidates in which two holes are lined up. This is because it can be detected.
[0093]
(1) Configuration of face direction calculation system 101
The nose detection unit 102 includes a separation degree detection unit 104, a right nostril detection unit 105, a right nostril template storage unit 106, a left nostril detection unit 107, a left nostril template storage unit 108, a double nostril detection unit 113, It comprises a nostril template storage unit 114 and a nose detection determination unit 109.
[0094]
The degree-of-separation detection unit 104 receives the image from the image input unit 103, extracts a feature point candidate of the face using a degree-of-separation filter, and extracts the information and image of the feature point candidate from the right nostril detection unit 105 and the left nostril detection unit 107 Is sent to the both nostrils detection unit 113.
[0095]
Both nostrils detection unit 113 receives feature point candidate information and image from separation degree detection unit 104, receives both nostrils templates from both nostrils template storage unit 114, and has both nostrils in the image based on these information. A search is performed to determine if both are present. Then, both nostrils detection result information is sent to the nose detection determination unit 109.
[0096]
The nostril template storage unit 114 stores both nostril templates in advance, and sends both nostril templates to the nostril detection unit 113 as necessary.
[0097]
Here, the “bore nostril template” is obtained by converting an image of a specific region including the right nostril and the left nostril as it is or by performing principal component analysis, and is data used in image matching by template matching. is there.
[0098]
The nose detection determination unit 109 receives right nostril detection result information from the right nostril detection unit 105, receives left nostril detection result information from the left nostril detection unit 107, receives both nostrils detection result information from the both nostrils detection unit 113, Then, it is determined whether or not the nose is detected from the image by comprehensively determining the information of the nose and the position information of the detected nose and nostril is sent to the face detection unit 110.
[0099]
(2) Face direction calculation processing procedure
FIG. 7 is a flowchart showing a processing procedure of the face orientation calculation system 101 according to the embodiment. The same parts as those in FIG. Only mention.
[0100]
In step ST111, a determination is made by imposing a geometric constraint on whether or not both nostrils can correspond to all peak pairs of the resolution peak.
[0101]
Next, the peak pair whose determination result is OK is collated with both nostril templates (ST112). Step ST103, step ST104, and “step ST111 and step ST112” can be performed simultaneously.
[0102]
【The invention's effect】
As described in detail above, according to the present invention, it is possible to provide a robust and highly accurate face detection system even when the face direction of a person changes greatly from side to side or wearing glasses.
[Brief description of the drawings]
FIG. 1 is a configuration diagram illustrating an example of a face orientation calculation system according to a first embodiment of the present invention.
FIG. 2 is a conceptual diagram illustrating a nostril template according to the embodiment.
FIG. 3 is a conceptual diagram illustrating a single nostril template according to the embodiment.
FIG. 4 is a conceptual diagram illustrating the relationship between the degree of lateral orientation of the face and the nostril detectable range according to the embodiment.
FIG. 5 is a flowchart illustrating an example of a face direction calculation process according to the embodiment.
FIG. 6 is a configuration diagram illustrating an example of a face direction calculation system according to a second embodiment of the present invention.
FIG. 7 is a flowchart showing an example of a face direction calculation process according to the embodiment.
[Explanation of symbols]
1,101 Face orientation calculation system
2,102 Nose detector
3,103 Image input unit
4,104 Separation degree detector
5, 105 Right nostril detector
6,106 Right nostril template storage
7, 107 Left nostril detector
8,108 Left nostril template storage
9,109 Nose detection determination unit
10,110 Face detection unit
11,111 Face orientation template storage
12, 112 Face direction calculation unit
21, 31 Face orientation template corresponding area
22 Nostril template corresponding area
23, 33 right nostril
24, 34 Left nostril
32 Right nostril template corresponding area
113 Nostril detector
114 Nostril template storage

Claims (7)

A face detection system that detects a human face from an image,
Image acquisition means for acquiring the image,
Separation degree detecting means for extracting a feature point candidate from the obtained image by a separation degree filter,
A face orientation template storage means for storing a plurality of face orientation templates,
Right nostril template storage means for storing a right nostril template;
Left nostril template storage means for storing a left nostril template;
Right nostril detection means for performing right nostril detection using the extracted feature point candidates and the stored right nostril template,
Left nostril detection means for performing left nostril detection using the extracted feature point candidates and the stored left nostril template,
Based on the detection state of the right nostril or the detection state of the left nostril, a predetermined face orientation template is selected from the plurality of stored face orientation templates, and the selected face orientation template and the acquired image are used. Face detection means for performing face detection by
A face detection system comprising: Nostril template storage means for storing both nostril templates,
Nostril detecting means for detecting both nostrils using the extracted feature point candidates and the stored nostril template,
With
The face detection means,
Based on the detection state of the both nostrils, the detection state of the right nostril, or the detection state of the left nostril, a predetermined face direction template is selected from the plurality of stored face direction templates, and the selected face is selected. The face detection system according to claim 1, wherein face detection is performed using an orientation template and the acquired image. The face detection means,
The detected right nostril, or limiting the face search range to a predetermined range based on the left nostril, performing the face detection by moving the face orientation template in the limited face search range. The face detection system according to claim 1, wherein The face detection means,
Limiting the face search range to a predetermined range based on the detected nostrils, right nostril, or left nostril, and performing face detection by moving the face orientation template in the limited face search range The face detection system according to claim 2, wherein: The face detection means,
The face detection system according to claim 1, wherein a face direction is calculated based on a degree of similarity between the obtained image and a plurality of face direction templates. A face detection method for detecting a human face from an image,
An image acquisition step of acquiring the image,
A separation degree detecting step of extracting a feature point candidate from the obtained image by a separation degree filter,
A right nostril detection step of performing a right nostril detection using the extracted feature point candidates and a pre-stored right nostril template;
A left nostril detection step of performing left nostril detection using the extracted feature point candidates and a previously stored left nostril template,
Based on the detection state of the right nostril or the detection state of the left nostril, a predetermined face orientation template is selected from a plurality of pre-stored face orientation templates, and the selected face orientation template and the acquired image are used. A face detection step of performing face detection by
A face detection method comprising: A computer-implemented program for detecting a face of a person from an image by a computer,
An image acquisition function for acquiring the image,
A separation degree detection function of extracting a feature point candidate from the acquired image by a separation degree filter,
A right nostril detection function for performing right nostril detection using the extracted feature point candidates and a pre-stored right nostril template,
A left nostril detection function for performing left nostril detection using the extracted feature point candidates and a pre-stored left nostril template,
Based on the detection state of the right nostril or the detection state of the left nostril, a predetermined face orientation template is selected from a plurality of pre-stored face orientation templates, and the selected face orientation template and the acquired image are used. A face detection function that performs face detection
A program for a face detection method characterized by realizing the following.

Images