JP4670521B2 - Imaging device - Google Patents

Description

  The present invention relates to an imaging apparatus that recognizes a specific pattern such as a human face from an electronic image.

  Various information such as the contour shape of a person's face, the distance between each part such as eyes, nose, mouth, ears, skin color, etc. is acquired from an electronic image taken with an electronic camera, etc. There is known a technique for recognizing a human face by analyzing using a computer. For example, FaceIt (registered trademark) is currently used as a face recognition algorithm in practical use.

  Moreover, as an imaging device having a face recognition function, for example, there is one described in Patent Document 1. This is a technique that intentionally causes discoloration such as red eyes in the subject person in an electronic image by photographing with flash emission, and recognizes the presence of a face in the image based on the discolored portion. .

JP 2004-206688 A

  However, it is not always possible to recognize the presence of a face by one flash photography, and Patent Document 1 does not consider this point. Moreover, the method of patent document 1 is limited to face recognition, and cannot be used for other pattern recognition.

An imaging apparatus according to the present invention includes an imaging unit that images a subject to generate electrical image data, a pattern recognition unit that recognizes a specific pattern in the image data, an illumination unit that illuminates the subject, and an illumination unit. perform imaging by illuminating the object, and a recognition operation unit attempting recognition operation by the pattern recognition means for the obtained image data, in the case of inability to recognize the pattern, to change the illumination characteristics of the illumination means If the pattern is recognized successfully by the recognition operation that is repeatedly executed while changing the illumination characteristics of the illumination means, auto-focus processing is performed using the recognized part as the focus area. To perform an operation based on the recognition result for the recording image acquired in the main imaging by the imaging means executed after success And features.

  According to the present invention, when the pattern recognition is impossible, the recognition operation is repeated while changing the illumination characteristics of the illumination means. Therefore, even if the face recognition is impossible with one light emission, the face recognition is finally performed. Can increase your chances of success.

An embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a control block diagram of the digital camera in the present embodiment. A subject light beam that has passed through a photographing lens (not shown) forms an image on a light receiving surface of an image sensor 1 such as a CCD. The image sensor 1 is controlled in its operation timing by the timing generator 2 and outputs an electrical image signal corresponding to the light intensity of the formed subject image. The image signal is converted into a digital signal by the A / D converter 3 and then subjected to various image processing in the image processing circuit 4 to generate image data. The image data is compressed as necessary and recorded on a recording medium such as a recording circuit 5 memory card.

  The timing generator 2 and the image processing circuit 4 are controlled by the CPU 6. Also connected to the CPU 6 are a face recognition calculation unit 7, a flash device 8 for illuminating a subject, an operation unit 9 including a plurality of switches, and the like.

  The face recognition calculation unit 7 recognizes a human face from an image using a known method as described in the background art, and inputs a recognition result to the CPU 6. If the face recognition is successful, not only can the face be found in the image, but it is also possible to automatically determine who the face is. Based on the face recognition result, the CPU 6 can perform various operations. For example, the recognized face is collated with face data for a plurality of persons registered in the database in the camera, and if there is a match, the image data is recorded in a folder prepared for that person. According to this, by preparing folders for a plurality of people in advance on the image recording medium, it is possible to automatically sort the taken images by individual.

In addition, when a plurality of faces are recognized in a group photo or the like, individual face portions can be cut out and distributed to individual folders. Further, by performing AF processing with the face recognition portion in the image as the focus area, it is possible to automatically focus on the face (face recognition AF).
Note that the face recognition and the operation based on the result are performed only when the face recognition mode is set on the camera.

  By the way, if there is a shadow or the like on the face portion of a person at the time of imaging, there are inconveniences such that the success rate of face recognition by the face recognition calculation unit 7 is reduced, and recognition takes time. In addition, when photographing at low luminance, face recognition may be difficult due to the influence of noise or the like. Therefore, it is conceivable to illuminate the subject and take an image, and to perform face recognition using the image signal obtained by this, but there is no guarantee that face recognition will be successful with one light emission. In the present embodiment, in such a case, the success rate of face recognition is increased by changing the illumination characteristics of the flash device 8 and trying light emission, imaging, and face recognition operations again. Hereinafter, a description will be given with reference to FIG.

FIG. 2 is a flowchart showing shooting control when the face recognition mode is set.
If a release operation is performed when the face recognition mode is set, imaging for face recognition is performed in step S1 to obtain image data. In step S2, face recognition by the face recognition calculation unit 7 is performed on the image data, and it is determined in step S3 whether the face recognition has succeeded or failed. In this case, the human face is used as the specific pattern, but the photographer may be able to select a recognition pattern (other than the face) in advance. By selecting the recognition pattern, the recognition speed and the recognition success rate can be improved.

  If the face recognition is successful, main imaging is performed in step S11 to obtain image data to be recorded on the recording medium. In step S12, an operation based on the face recognition result is performed. For example, the obtained face data is collated with data in the database, and if there is a match, the image data is recorded in a folder prepared for the person.

  On the other hand, if it is determined in step S3 that face recognition is impossible, the process proceeds to step S4. Here, face recognition failure means that the face cannot be recognized in the image or the presence of the face can be recognized, but data sufficient to identify the person cannot be obtained, or the recognition result is output even after a certain period of time. Refers to no state. A possible cause of the failure is that the luminance pattern of the object scene is below the threshold of face recognition ease.

  Therefore, in step S4, the flash device 8 emits light to illuminate the subject, and imaging is performed in step S5. In step S6, face recognition is attempted again on the obtained image data, and the result is determined in step S7. If the face recognition is successful, the process proceeds to step S11. If the face recognition is unsuccessful, the elapsed time from the first face recognition is determined in step S8. If the predetermined time has not been reached, the process proceeds to step S9 to try face recognition again. In step S9, the illumination characteristics of the flash device 8 are changed.

  Here, as an illumination characteristic of the flash device 8, there is, for example, a light emission amount. Since it is considered that the face recognition has failed because the light emission amount is too small, next, for example, the threshold value of the light control level is adjusted in step S9 to increase the light emission amount, and then the process returns to step S4. In addition, the image data obtained by flash imaging is subjected to contrast processing and contour enhancement processing, and tone correction and color correction are performed on the image data in consideration of the color characteristics of illumination light and the characteristics of the image sensor. This can improve the success rate of face recognition.

  On the other hand, if it is determined in step S8 that the predetermined time has elapsed, the face recognition is abandoned, a warning indicating that the face cannot be recognized (for example, a warning sound or a warning display) is given in step S10, and the main imaging is performed in step S11. I do. Note that instead of time, face recognition may be abandoned when the number of face recognitions reaches a predetermined number and is still unrecognizable.

  Here, the illumination characteristics of the flash device 8 are not limited to the light emission amount, but also include the light emission angle, the light emission direction, and the like. For example, in a group photo, even if the flash device is turned on, the light does not turn to the corner of the screen and the face of a person located in the corner may not be recognized. In such a case, it is possible to make everyone's face recognition possible by widening the light emitting angle (light distribution angle) of the flash device and irradiating again. The change of the light distribution angle can be realized with a zoom flash device. Furthermore, when the person is not in the center of the screen, the face cannot be recognized even if the center is illuminated. In this case, irradiation may be performed by changing the illumination direction (angle). When a face recognition area is designated in advance, illumination light may be emitted in that direction. The change of the illumination direction can be realized by using a flash device in which the direction of the light emitting unit is variable. The light emission angle (light distribution angle) may be zoomed according to the zooming of the imaging lens.

  As described above, in order to make face recognition possible, it is desirable to repeat light emission, imaging, and face recognition while sequentially or simultaneously changing the light emission amount, light emission angle, and light emission direction of the flash device. Even if face recognition is impossible due to light emission, the possibility of finally succeeding in face recognition can be increased. Since how much the illumination characteristic should be changed depends on the face recognition algorithm, it is desirable to obtain an appropriate amount by experiment in advance.

  The present invention can also be used for a personal authentication system. Further, the present invention is not limited to human face recognition, and can be widely used in the case of recognizing specific patterns of pets, plants, and other objects. The illumination light is not limited to visible light, but may be infrared light, and a type of light suitable for a pattern recognition algorithm may be selected. Furthermore, an optimal image can be obtained by adjusting the light emission amount, light emission angle, and light emission direction of the flash device according to the registered pattern characteristics based on the pattern recognition result. For example, wrinkles on the face can be made inconspicuous.

The block diagram of the camera in one Embodiment of this invention. The flowchart explaining operation | movement of one Embodiment.

Explanation of symbols

1 Image sensor 6 CPU
7 Face recognition calculation unit 8 Flash device

Claims (4)

Imaging means for imaging a subject and generating electrical image data;
Pattern recognition means for recognizing a specific pattern in the image data;
Illumination means for illuminating the subject;
Recognizing the object with the illuminating means to perform imaging, and having a recognition operation means for trying the recognition operation by the pattern recognition means on the obtained image data,
In the case of non recognition of the pattern, repeat the recognition operation while changing the illumination characteristics of said illumination means,
When said recognition of the pattern by the recognition operation repeatedly executed while changing the illumination characteristics of the illumination means is successful, a moiety recognizing the pattern autofocus process as a focus area, the recognition of the pattern has been successfully An image pickup apparatus that performs an operation based on a recognition result with respect to a recording image acquired in the main image pickup by the image pickup means executed later. If the recognition of the first pattern in the inability to recognize patterns even predetermined time has elapsed, or when the recognition number of the pattern is unavailable recognition of patterns reaches the predetermined number of times, a truncating the pattern recognition The imaging apparatus according to claim 1, characterized in that:   The imaging apparatus according to claim 1, wherein the illumination characteristic of the illumination unit includes at least one of a light emission amount, an irradiation angle, and an irradiation direction. Recognition of the pattern by the serial recognition operation means is a face recognition to identify a person,
The operation based on the recognition result is an operation of recording an image acquired in the main imaging in association with a person specified based on the face recognition. The imaging device described in 1.

Images