JP2012238111A - Face image recognition device and face image recognition program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a face image recognition device and a face image recognition program capable of processing face image recognition jobs faster and performing registration work more simply.SOLUTION: A face image recognition device 20 that recognizes a face of a person in an image comprises: template storage means 24 in which person specification templates of face images of persons to be recognized keeping their eyes front and person unspecified templates of average face images facing desired orientations other than front of a lot of people are registered; face area extracting means 22 that extracts from an image a face area having a face of a person; and face recognition means 23 that extracts features of the extracted face area, compares the features of the face area and the person specification templates stored in the template storage means 24 to recognize the face of a person in the face area while tracing the orientation of a face of a person recognized in the face area by comparing the features of the face area and the person unspecified templates registered in the template storage means 24.

Description

  The present invention relates to a face image recognition device and a face image recognition program, and more particularly to a face image recognition device and a face image recognition program for recognizing a human face in a video.

  In the conventional face image recognition method, images in which the faces of persons to be recognized are shown in a plurality of directions are collected, a variable template is constructed from each face image, and registered in a database (DB). In the conventional face image recognition method, a face area in which a person's face in the input video is reflected is detected, and the feature extracted from the face area is compared with each variable template to calculate the degree of similarity. And which face is reflected in each face area reflected in each frame (for example, see Non-Patent Documents 1 and 2).

  FIG. 1 is a configuration diagram of an example of a conventional face image recognition system. The face image recognition system 100 of FIG. 1 has a configuration including a face area detection unit 101, a face part tracking unit 102, and a variable template DB 103. A face area detection unit 101 detects a face area in which a person's face is captured in the input video. As a technique for detecting a face area from an input video, an existing technique such as a Viola-Jones face detector provided by the OpenCV library can be used (for example, see Non-Patent Document 3).

  The face part tracking unit 102 extracts the features of the face area detected by the face region detection unit 101, recognizes who the face is by comparing the extracted features with a variable template, and displays it in each frame. In addition, it was estimated which face was reflected in each face area.

Simon CLIPPINGDALE, Takayuki Ito, “A unified approach to face detection, tracking, and recognition of moving images”, IEICE PRMU 98-200, January 1999. S. Clippingdale, M. Fujii, M. Shibata, Multimedia Databases for VideoIndexing: Toward Automatic Face Image Registration, Proc. IEEE International Symposium on Multimedia 2009, pp. 639-644, 2009. S. Clippingdale, “Development of Real-time Tracking System for Facial Parts”, IEICE Winter Conference, 2010.

  In the conventional face image recognition system 100, each frame of the input video is subjected to face area detection processing, variable template matching is initialized in the detected face area, and a person is recognized. In the conventional face image recognition system 100, even if the face direction is away from the front and the face area cannot be detected, the face image is not detected based on the variable template matching result in the previous frame. Continue tracking and recognizing people.

  In the conventional face image recognition system 100, variable templates in each direction for the number of persons to be recognized are necessary even in directions other than the front. Therefore, the conventional face image recognition system 100 shown in FIG. 1 has the following problems.

  As a first problem, in the conventional face image recognition system 100, in order to initialize collation with a variable template, it is necessary to first detect an area that is likely to be a face in the input video as a face area. Since the face area detection processing by the face area detection unit 101 scans the entire screen of the frame of the input video, there is a problem that the calculation amount is large and time is required.

  As a second problem, in the conventional face image recognition system 100, it is necessary to collect images in which a person's face to be recognized is reflected in a plurality of directions, construct a variable template from each face image, and register it in the DB. is there. However, it is not easy to collect images in which the faces of persons to be recognized are reflected in a plurality of directions.

  As a third problem, even if the conventional face image recognition system 100 can collect images in which a person's face to be recognized is reflected in a plurality of directions, the orientation of the face shown in the input video is determined. The closer the face orientation registered in the variable template DB 103 is, the higher the similarity calculated by collation tends to be.

  Therefore, in the conventional face image recognition system 100, the orientation of the face of the variable template of another person (other person) is different from that of the face of the person's variable template in the direction of the face of the person (person) reflected in the input video. If it is closer than the direction, the person may be misrecognized as another person.

  As a fourth problem, in the conventional face image recognition system 100, a plurality of images of the face of a person to be recognized appear in the same direction (images of faces of all persons to be recognized in a matching direction). If there is, the third problem does not occur, but there is a problem that a large amount of calculation is required for collation because there are variable templates for the number of persons to be recognized for each of a plurality of orientations.

  SUMMARY An advantage of some aspects of the invention is that it provides a face image recognition apparatus and a face image recognition program capable of speeding up face image recognition processing and simplifying registration work.

  In order to solve the above-mentioned problems, the present invention is a face image recognition device for recognizing a person's face in a video, and a person identification template of a face image facing the front of a person to be recognized and a number of person's targets A template storage means in which a person unspecified template of an average face image in a direction other than the front is registered, a face area extraction means for extracting a face area in which a person's face is shown from an image, and an extracted face A feature of the region is extracted, the feature of the face region is compared with the person identification template registered in the template storage means, and the face of the person in the face region is recognized, and the face region A face recognizing unit that tracks the direction of the face of the recognized person in the face region by comparing the feature with the person unspecified template registered in the template storage unit. And wherein the door.

  In addition, what applied the component, expression, or arbitrary combination of the component of this invention to a method, an apparatus, a system, a computer program, a recording medium, a data structure, etc. is also effective as an aspect of this invention.

  According to the present invention, it is possible to provide a face image recognition apparatus and a face image recognition program capable of speeding up the face image recognition process and simplifying the registration work.

It is a block diagram of an example of the conventional face image recognition system. It is a hardware block diagram of an example of PC. It is a functional block diagram of an example of the face image recognition apparatus of a present Example. It is explanatory drawing of an example which shows the outline | summary of skin color area | region extraction. It is explanatory drawing of an example which shows the content of a variable template. It is explanatory drawing of an example which shows the construction method of the person unspecified variable template in a certain direction. It is a function block diagram of an example of a face component tracking part.

  Next, modes for carrying out the present invention will be described based on the following embodiments with reference to the drawings.

<Outline of the present invention>
According to the present invention, in face image recognition by a computer, the face of a person in a video is tracked, and the face of the person registered in the face image database is compared to recognize who the face is. The present invention copes with the increase in the amount of calculation associated with the correspondence between a plurality of people and the orientation of a plurality of faces and the complexity of the registration work. The face of a person registered in the database is compared to recognize who the face is.

  Further, the present invention uses a specially prepared face template (person unspecified variable template) when a person's face appears in a direction away from the front, tracks the face area, and holds the recognition result. Although details of the person unspecified variable template will be described later, it is sufficient to use one person unspecified variable template for each orientation regardless of the number of persons to be recognized. The calculation amount is greatly reduced.

  As described above, since the calculation amount of the tracking technique for tracking the face area is smaller than the calculation amount for recognizing faces of a plurality of persons in various directions, the robustness corresponding to various face directions is achieved. High-performance face image recognition processing can be realized.

<Hardware configuration>
The face image recognition apparatus of the present embodiment can be realized by a PC, a workstation, or the like. Here, an example in which the face image recognition apparatus of the present embodiment is realized by a PC will be described. It should be noted that the face image recognition device does not necessarily indicate that it is composed of one housing. In addition, the face image recognition apparatus according to the present embodiment may be configured such that functions are distributed to a plurality of apparatuses like a face image recognition system.

  The face image recognition apparatus of this embodiment is realized by a PC having a hardware configuration as shown in FIG. FIG. 2 is a hardware configuration diagram of an example of a PC. The PC 10 includes an input device 11, an output device 12, a recording medium reading device 13, an auxiliary storage device 14, a main storage device 15, an arithmetic processing device 16, and an interface device 17 that are connected to each other via a bus 19.

  The input device 11 is a keyboard or a mouse. The input device 11 is used for inputting various signals. The output device 12 is a display device or the like. The output device 12 is used to display various windows and data. The interface device 17 is a modem, a LAN card, or the like. The interface device 17 is used for connecting to a network.

  The face image recognition program installed in the face image recognition device is at least a part of various programs for controlling the PC 10. The face image recognition program is provided by, for example, distribution of the recording medium 18 or downloading from a network or the like.

  The recording medium 18 is a recording medium that records information optically, electrically, or magnetically, such as a CD-ROM, a flexible disk, or a magneto-optical disk, or a semiconductor that electrically records information, such as a ROM or flash memory. Various types of recording media such as a memory can be used.

  When the recording medium 18 on which the face image recognition program is recorded is set in the recording medium reader 13, the face image recognition program is installed from the recording medium 18 to the auxiliary storage device 14 via the recording medium reader 13. The face image recognition program downloaded from the network or the like is installed in the auxiliary storage device 14 via the interface device 17.

  The auxiliary storage device 14 stores the installed face image recognition program, necessary files, data, and the like. The main storage device 15 reads and stores the face image recognition program from the auxiliary storage device 14 when the face image recognition program is started. The arithmetic processing unit 16 implements various processes as described later according to the face image recognition program stored in the main storage device 15.

<Functional configuration>
FIG. 3 is a functional configuration diagram of an example of the face image recognition apparatus according to the present embodiment. The face image recognition apparatus 20 of FIG. 3 has a configuration including a skin color area extraction unit 21, a face area detection unit 22, a face part tracking unit 23, and a variable template DB 24.

  The skin color area extracting unit 21 extracts a skin color area from the input video. Details of the process of extracting the skin color area from the input video will be described later. The face area detection unit 22 extracts a face area in which a person's face is shown from the skin color area extracted by the skin color area extraction unit 21. As the technique for extracting the face area, the existing technique described above may be used.

  The face part tracking unit 23 extracts the features of the face area detected by the face region detection unit 22, and collates the extracted features with the variable template registered in the variable template DB 24. Estimate which face is reflected in which direction. Details of the processing of the face part tracking unit 23 will be described later.

  Note that the variable template DB 24 used in this embodiment is different from the conventional variable template DB 103 as follows. The variable template DB 24 registers a variable template of the front direction (hereinafter referred to as a person specific variable template) and a variable template of a direction other than the front (hereinafter referred to as a person unspecified variable template). The person specifying variable template is a variable template of a face image in the front direction of the person to be recognized. The person unspecified variable template is a variable template of an average face image constructed by collecting face images of target orientations of a large number of persons that are not necessarily recognition targets.

  The variable template DB 24 is obtained by replacing the person specifying variable templates corresponding to the number of persons to be recognized in one direction other than the front with one person unspecified variable template. Details of the variable template DB 24 will be described later.

  The face part tracking unit 23 extracts the features of the face region detected by the face region detection unit 22, and collates the extracted features with the person specifying variable template registered in the variable template DB 24, thereby detecting each detected face region. It is estimated which face is reflected in each face area by collating the extracted feature with the person unspecified variable template registered in the variable template DB 24. presume.

  The face component tracking unit 23 recognizes a person by collating the face area in which the face is reflected in the direction close to the front with the person specifying variable template, and then the face is rotated even if the face is rotated away from the front. By tracking the face by comparing the face area in which the face appears in a direction other than the face and the person unspecified variable template, the recognition result of the person can be held in association with the face area.

  The face image recognition apparatus 20 of the present embodiment solves the first problem described above by narrowing the target range of the process for extracting the face area to the skin color area. Further, the face image recognition apparatus 20 of the present embodiment solves the above second to fourth problems by using a person unspecified variable template.

<Skin color area extraction>
The skin color area extraction unit 21 determines whether or not the RGB value of each pixel of the input frame corresponds to the skin color by a normalized CbCr method (see, for example, Non-Patent Document 3). The skin color region extraction unit 21 converts the RGB values into the YCbCr space, and normalizes the values of Cb and Cr by the equation (1).

なお、Ｙは輝度信号である。ＣｂはＢ信号から輝度信号Ｙを差し引いた色差信号Ｂ−Ｙに特定の定数を掛けた値である。また、ＣｒはＲ信号から輝度信号Ｙを差し引いた色差信号Ｒ−Ｙに特定の定数を掛けた値である。

Y is a luminance signal. Cb is a value obtained by multiplying the color difference signal BY obtained by subtracting the luminance signal Y from the B signal by a specific constant. Cr is a value obtained by multiplying the color difference signal RY obtained by subtracting the luminance signal Y from the R signal by a specific constant.

  If the position of (Cb ′, Cr, ′) (x) corresponding to a certain pixel x is within a certain distance from the skin color representative point, the skin color region extraction unit 21 will obtain that pixel as shown in Expression (2). x is determined to be a skin color.

肌色領域抽出部２１は一般化されたメディアンフィルタにより、肌色または肌色ではないと判定され孤立した画素を除去する。以下の式（３）ではａ＝０．５の場合に、メディアンフィルタとなる。以下の式（３）では、ａの値により少なめ又は多めに画素がＯＮ（ｑ（ｘ）＝１）になる。

The skin color region extraction unit 21 removes an isolated pixel that is determined not to be a skin color or skin color by a generalized median filter. In the following formula (3), a median filter is obtained when a = 0.5. In the following formula (3), the number of pixels is turned ON or less (q (x) = 1) depending on the value of a.

肌色領域抽出部２１は連接領域抽出（connected component labeling）の処理と連接領域の外周ボックスに条件（閾値）をかけることにより、安定した肌色領域を抽出する。

The skin color area extraction unit 21 extracts a stable skin color area by applying a condition (threshold value) to a connected area labeling process and a peripheral box of the connected area.

  FIG. 4 is an explanatory diagram of an example showing an outline of skin color region extraction. The skin color area extraction unit 21 converts the RGB value of each pixel into a YCbCr space, normalizes the values of Cb and Cr by equation (1), and projects them onto the Y = 128 plane. If the skin color area extraction unit 21 is close to the skin color representative point, the pixel is determined to be a skin color.

  The skin color area extraction unit 21 outputs an area from which independent pixels are removed by a generalized median filter. Further, the skin color area extracting unit 21 outputs a skin color area list by the process of extracting the connected area. The skin color area extraction unit 21 extracts a stable skin color area by a median filter and a connected area extraction process.

<Construction of variable template>
The variable template used in the conventional face image recognition system 100 has components as shown in FIG. FIG. 5 is an explanatory diagram of an example showing the contents of the variable template. The variable template used in the conventional face image recognition system 100 includes, as components, the coordinates of up to nine feature points that are visible according to the orientation of the face and the Gabor wavelet features measured at each feature point in the registered face image. Have.

In FIG. 5, the number of feature points is N _FP = 9, but other numbers may be used. Since the feature points are specified for facial parts such as the eyes and the corners of the eyes, the two-dimensional shape of the face image is reflected.

  Also, in FIG. 5, the Gabor wavelet feature uses 5 resolutions × 8 orientations, but other numbers may be employed. The Gabor wavelet feature expresses how much signal energy is present in the face image in the vicinity of the feature point at each resolution and each orientation, and therefore reflects the luminance pattern of the image.

  In the conventional face image recognition system 100, it is necessary to prepare variable templates corresponding to the number of registered persons (the number of persons to be recognized) for each direction. In the present invention, the variable templates for the registered number of persons in directions other than the front are replaced with one variable variable template.

FIG. 6 is an explanatory diagram illustrating an example of a method for constructing a person unspecified variable template in a certain direction. First, in the construction method of the person unspecified variable template in a certain direction, face images I ^k (x), 0 in each target direction of a large number of persons k, 0 ≦ k <K that do not necessarily include the person to be authenticated. ≦ k <K, and feature points

の位置（座標）を手動で指定する。各特徴点の平均位置

Manually specify the position (coordinates). Average position of each feature point

と各特徴点のガボールウェーブレット特徴

And Gabor wavelet features of each feature point

を計測する。次に、ある向きにおける人物不特定可変テンプレートの構築方法では各特徴点ｎ，０≦ｎ＜Ｎ_ＦＰについて、各画像でその特徴点ｎを平均位置にずらしてアラインした平均画像Ａ_ｎ（ｘ）を算出する。

Measure. Next, in the method for constructing a person unspecified variable template in a certain direction, for each feature point n, 0 ≦ n <N _FP , an average image A _n (x) in which the feature point n is shifted to the average position in each image and aligned. Is calculated.

次に、ある向きにおける人物不特定可変テンプレートの構築方法ではガボールウェーブレット特徴ｃ_nを、平均画像Ａ_ｎ（ｘ）での各平均特徴点位置

Next, in the method for constructing the person unspecified variable template in a certain direction, the Gabor wavelet feature c _n is converted into each average feature point position in the average image A _n (x).

で計測する。次に、ある向きにおける人物不特定可変テンプレートの構築方法では人物ｋの顔画像Ｉ^ｋと各特徴点ｎとで、ガボールウェーブレット特徴ｃ_nを平均画像Ａ_ｎ（ｘ）において、各平均特徴点位置

Measure with Next, in the method for constructing a person unspecified variable template in a certain direction, a Gabor wavelet feature c _n is represented in the average image A _n (x) with each face image I ^k of the person k and each feature point n.

で計測する。次に、ある向きにおける人物不特定可変テンプレートの構築方法では顔画像Ｉ^ｋと各特徴点ｎとで、ガボールウェーブレット特徴

Measure with Next, in the method for constructing a person unspecified variable template in a certain direction, a Gabor wavelet feature is represented by a face image I ^k and each feature point n

とｃ_ｎとの位相の差分より、顔画像Ｉ^ｋの特徴点nが平均画像Ａ_ｎ（ｘ）での平均特徴点

Average feature point between the phase difference between c _n, the face image I ^k feature points n is the average image A _n of _(x)

から、どれだけずれているか、その位置ずれ

The amount of deviation from the position

を推定する。次に、ある向きにおける人物不特定可変テンプレートの構築方法では推定した位置ずれを新たに顔画像Ｉ^ｋ（ｘ）に適用して特徴点ごとに、更にずらされた平均画像を算出する。

Is estimated. Next, in the method for constructing a person unspecified variable template in a certain direction, the estimated positional deviation is newly applied to the face image I ^k (x), and an average image further shifted is calculated for each feature point.

次に、ある向きにおける人物不特定可変テンプレートの構築方法では、新たに

Next, in the method of constructing a person unspecified variable template in a certain direction,

において、平均特徴点位置

Mean feature point position

でガボールウェーブレット特徴

With Gabor wavelet features

を計測する。人物不特定可変テンプレートは各平均特徴点位置

Measure. Individual unspecified variable template is the average feature point position

とガボールウェーブレット特徴

And Gabor wavelet features

を有する。

Have

  Next, in the construction method of the person unspecified variable template in a certain direction, the variable templates for the registered number of persons other than the front direction are replaced with one person unspecified variable template. As in the conventional face image recognition system 100, the variable templates for the registered number of persons in the front direction remain the same as the variable templates for the registered number of persons.

<Configuration of Facial Parts Tracking Unit 23>
FIG. 7 is a functional configuration diagram of an example of the face part tracking unit. 7 has a configuration including a recognition unit 31 and a tracking unit 32.

  The recognizing unit 31 extracts the coordinates of the feature points of the face area detected by the face area detecting unit 22 and the Gabor wavelet features measured at each feature point, and the coordinates of the extracted feature points and the measured feature points are measured. By collating the Gabor wavelet feature with the person specifying variable template registered in the variable template DB 24, it is estimated which face is reflected in which direction in each detected face area.

  The tracking unit 32 extracts the coordinates of the feature points of the face region detected by the face region detection unit 22 and the Gabor wavelet features measured at each feature point, and the coordinates of the extracted feature points and the feature points are measured. By collating the Gabor wavelet feature with the person unspecified variable template registered in the variable template DB 24, it is traced in which direction the face appears in each detected face area.

  By the way, the face area detection unit 22 extracts a face area when a face in a direction close to the front is shown in the skin color area. In other words, the face area extracted by the face area detection unit 22 shows a face in a direction close to the front. As a result, the coordinates of the feature points of the face area extracted by the face area detection unit 22 and the Gabor wavelet features measured at each feature point most closely match the person specifying variable template.

  After that, when the face in the face area is rotated away from the front direction, the coordinates of the feature points of the face area and the Gabor wavelet features measured at each feature point match the person unspecified variable template most. become. However, the person unspecified variable template is an average variable template constructed by collecting face images of a large number of persons. Therefore, when matching most with the person unspecified variable template, the recognition unit 31 maintains the estimation (recognition) result already estimated by the person specifying variable template.

  As described above, the face component tracking unit 23 in FIG. 7 tracks the face by the tracking unit 32 and maintains the recognition result by the recognition unit 31 even if the face shown in the face area rotates away from the front direction. (Holding) makes it possible to recognize people with various face orientations.

  In other words, the face image recognition apparatus 20 of this embodiment can track the face direction using a person unspecified template that cannot be used for recognition if the face is recognized in the front direction even if there are no variable templates for the number of registered persons. The recognition result of the person can be given to the face images of various orientations by continuously taking over the already recognized recognition result.

<Effect>
The face image recognition apparatus 20 of the present embodiment reduces the detection time of the face area by the face area detection unit 22 to the skin color area extracted by the skin color area extraction unit 21, thereby reducing the processing time of the face area detection to almost (skin color). Area / screen area).

  Further, the face image recognition apparatus 20 of the present embodiment can greatly reduce the calculation amount of the face part tracking unit 23 as compared with the previous face image recognition system 100. The calculation amount of the face part tracking unit 23 can be reduced to almost (1 / registered number of people) when the face is shown in a direction other than the front.

  Further, since the face image recognition apparatus 20 of the present embodiment only has to collect images in which the faces of the person to be recognized are reflected in the front direction, the registration work of the variable template DB 24 is compared with the face image recognition system 100 that is still used. It can be greatly simplified.

  Further, since the variable template DB 24 of the face image recognition apparatus 20 of the present embodiment replaces the person specifying variable templates for the number of persons to be recognized with respect to the direction other than the front, the variable template DB 24 is replaced with one person unspecified variable template. The third problem does not occur. That is, the face image recognition apparatus 20 of the present embodiment can reduce the performance difference due to the creation of the variable template DB 24.

  The present invention is not limited to the specifically disclosed embodiments, and various modifications and changes can be made without departing from the scope of the claims. The template storage means described in the claims corresponds to the variable template DB 24, the face area extraction means corresponds to the face area detection section 22, the face recognition means corresponds to the face part tracking section 23, and the skin color area extraction. The means corresponds to the skin color area extraction unit 21, the recognition means corresponds to the recognition unit 31, the tracking means corresponds to the tracking unit 32, the person specifying template corresponds to the person specifying variable template, and the person unspecified template It corresponds to a specific variable template.

10 PC
DESCRIPTION OF SYMBOLS 11 Input device 12 Output device 13 Recording medium reading device 14 Auxiliary storage device 15 Main storage device 16 Arithmetic processing device 17 Interface device 18 Recording medium 19 Bus 20 Face image recognition device 21 Skin color area extraction unit 100 Face image recognition system 22, 101 Face Area detection unit 23, 102 Face component tracking unit 24, 103 Variable template DB
31 recognition unit 32 tracking unit

Claims (5)

A face image recognition device for recognizing a person's face in a video,
A template storage unit in which a person identification template of a face image in the front direction of a person to be recognized and a person unspecified template of an average face image in a direction other than the front of a number of persons are registered;
A face area extracting means for extracting a face area in which a person's face is reflected from an image;
Extracting the features of the extracted face area, comparing the features of the face area with the person identification template registered in the template storage means, and recognizing the face of a person appearing in the face area; Face recognition means for tracking the orientation of the face of the recognized person in the face area by comparing the feature of the face area with the person unspecified template registered in the template storage means. A face image recognition device characterized by the above. A skin color area extracting means for extracting a skin color area from the video;
2. The face image recognition apparatus according to claim 1, wherein the face area extracting unit extracts a face area in which the face of the person is reflected from the extracted skin color area. Recognizing means for recognizing a face of a person appearing in a front direction in the face area by comparing the feature of the face area with the person identification template;
Tracking means for comparing the characteristics of the face area and the person-unspecified template to track the orientation of the face of the recognized person appearing in the face area in a direction other than the front;
3. The face image recognition apparatus according to claim 1 or 2, wherein the recognition means holds a recognition result of the recognized person appearing in the face area in a direction other than the front. The person specifying template and the person non-specific template have, as components, coordinates of a plurality of feature points and Gabor wavelet features measured at each feature point,
The person unspecified template calculates, for each feature point of a plurality of face images in a direction other than the target front, a first average image adjusted by shifting the feature point to an average position in the face image,
Estimating a positional shift by a phase difference between a Gabor wavelet feature measured at each average feature point of the first average image and a Gabor wavelet feature measured at each feature point of the plurality of face images;
The estimated positional deviation is newly applied to the plurality of face images, and for each feature point of the plurality of face images, a second average image further shifted for each feature point is calculated, and the second average image is calculated. 4. The face image recognition apparatus according to claim 1, further comprising a Gabor wavelet feature measured at each average feature point of the image as a component. Computer
A template storage unit in which a person identification template of a face image in the front direction of a person to be recognized and a person unspecified template of an average face image in a direction other than the front of a number of persons are registered;
A face area extracting means for extracting a face area in which a person's face is reflected from an image;
Extracting the features of the extracted face area, comparing the features of the face area with the person identification template registered in the template storage means, and recognizing the face of a person appearing in the face area; To collate the feature of the face area with the person unspecified template registered in the template storage means to function as face recognition means for tracking the orientation of the face of the recognized person in the face area. Face image recognition program.

Images