JP2001330882A - Camera with subject recognizing function - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make operability excellent by facilitating the mounting of a subject recognizing means in a camera and improving subject information recognizing accuracy and recognizing speed. SOLUTION: This camera is provided with a photographing mode setting means for setting the photographing mode of the camera, an area sensor consisting of plural pixels and picking up the image of a subject, the subject recognizing means for recognizing the photographed subject information based on output from the area sensor, and a control means for controlling the photographing function of the camera in accordance with the result of the recognition by the subject recognizing means. The subject recognizing means changes (#108 to #111) recognition algorithm for recognizing the subject information in accordance with the set photographing mode.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement of a camera having an object recognizing function having an object recognizing means for recognizing an object to be photographed using an output of an area sensor comprising a plurality of pixels.

[0002]

2. Description of the Related Art In recent cameras, functions have been advanced automatically. For example, a photographing mode dial 43 as shown in FIG. 2 is provided with a plurality of settable photographing modes, and the photographer responds to the scene according to the scene to be photographed. If the photographing mode dial 43 is set at the position of the picture display, the camera can be operated by a previously programmed control method. Therefore, even if the user does not have any particular knowledge of the functions of the camera, the user simply selects one of the picture displays according to the shooting scene or situation desired to be taken, and then simply presses the shutter.

[0003] However, even if the shooting scene can be set, it is not known what the subject is and where it is, specifically, the position, size and number of the subject on the screen. Exposure control and autofocus that were optimal for the camera could not be performed accurately. Therefore, AE exposure according to a predetermined program diagram, such as fixing an aperture with a certain depth of field so that focusing does not fail and setting the aperture,
Only the AF mode which seems to be suitable has been selected. Also, when there is a difference in brightness between the background and the subject such as a backlight scene, it has been difficult to adjust the exposure only to the subject. On the other hand, it has been difficult to constantly track a moving subject to adjust exposure and autofocus.

Therefore, a camera provided with a subject recognizing means for recognizing a subject to be photographed has been desired. For example, JP-A-6-160944 proposes a camera provided with control means for automatically operating a subject tracking device when a shooting mode suitable for shooting a moving object is selected from the shooting modes. The color difference signal of the scene in the designated area is detected, and the position of the main subject is detected by this change.

[0005]

However, the tracking method based on the color difference signal alone cannot accurately grasp the subject, and there is a problem that it is not effective especially in a photograph of a person occupying a large part of the subject. Was. On the other hand, there has been considerable research on face recognition technology,
As a method of extracting a face from a color image, Japanese Patent Laid-Open No.
Although many proposals have been made, such as -63597, the amount of calculation required to recognize each of them is considerable, and it is difficult to achieve these in the limited space of a camera, microcomputer capability, and cost. there were. Particularly at the recognition speed, processing must be performed in a short time until the shutter is released, and high-speed processing is absolutely necessary.

(Object of the Invention) A first object of the present invention is to make it easy to mount an object recognizing means in a camera, and to improve accuracy and speed of recognizing object information.
An object of the present invention is to provide a camera with a subject recognition function with good operability.

A second object of the present invention is to provide a camera with a subject recognizing function capable of accurately recognizing subject information according to a photographing mode without complicating a subject information recognizing algorithm. Things.

A third object of the present invention is to provide a camera with a subject recognizing function capable of reducing the load for recognizing subject information and making it easier to mount subject recognizing means on the camera. What you want to do.

[0009]

According to a first aspect of the present invention, there is provided a photographing mode setting unit for setting a photographing mode of a camera, and a plurality of pixels for photographing a subject. An area sensor, subject recognition means for recognizing subject information to be shot based on the output of the area sensor, and control means for controlling a shooting function of the camera in accordance with a recognition result by the subject recognition means. The subject recognizing means is a camera with a subject recognizing function for changing a recognition algorithm for recognizing subject information according to a set photographing mode.

In order to achieve the second object,
According to a second aspect of the present invention, there is provided a photographing mode setting means for setting a photographing mode of a camera, an area sensor including a plurality of pixels for photographing a subject, and recognition of subject information to be photographed based on an output of the area sensor. And a control unit for controlling a photographing function of the camera in accordance with a result of the recognition by the subject recognition unit. The subject recognition unit recognizes the subject information in accordance with a set photographing mode. And a camera with a subject recognition function for changing parameters constituting the recognition algorithm.

Further, in order to achieve the third object,
According to a fifth aspect of the present invention, there is provided a photographing mode setting unit for setting a photographing mode of a camera, an area sensor including a plurality of pixels for photographing a subject, and object information to be photographed based on an output of the area sensor. At least a subject recognizing unit having a face recognizing algorithm for recognizing a face, and a control unit for controlling a photographing function of a camera according to a recognition result by the subject recognizing unit; The camera with the object recognition function that does not perform the operation of the face recognition algorithm depending on the shooting mode.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail based on illustrated embodiments.

FIG. 1 is a configuration diagram of a main part when the present invention is applied to a single-lens reflex camera. In FIG. 1, reference numeral 1 denotes a photographing lens (only four lenses are shown for simplicity), and a part 1a of the constituent lens is a focus lens for adjusting a focal position, and 1b is a blur correction lens. The blur correction lens 1b has an optical axis K
It is possible to move in a plane perpendicular to the image plane, and thereby to perform a known camera shake correction by changing an image forming position in the image forming plane of the photographing lens 1. Reference numeral 2 denotes a main mirror which is inclined or retracted to the photographing optical path according to the state of observation of the subject image by the finder system and the state of photography of the subject image. Reference numeral 3 denotes a sub-mirror, which reflects a light beam transmitted through the main mirror 2 toward a later-described focus detection device 6 below the camera body. Reference numeral 4 denotes a shutter, and reference numeral 5 denotes an image recording member which is a film or a solid-state image sensor located at a focal plane of the taking lens 1.

Reference numeral 6 denotes a focus detection device, which includes a field mask 61, a field lens 62, reflection mirrors 63 and 66, a secondary imaging lens 65, an aperture 64, and a well-known phase difference detection method. It is composed of a focus detection line sensor 67 and the like. Recently, the focus detection device 6 is provided with a plurality of focus detection points (regions for detecting sudi focus information) not only in the center of the screen but also in the periphery thereof. In this embodiment as well, a focus detection device having such a plurality of focus detection points is assumed, but this point is a known technique, and a detailed description thereof will be omitted. Reference numeral 7 denotes a focusing plate arranged on a predetermined image forming plane of the photographing lens 1, and 8 denotes a pentaprism. 9 and 10
Are an imaging lens and a photometric sensor for measuring the luminance of the subject in the viewfinder observation surface, respectively.
Is composed of a plurality of photodiodes so that a plurality of areas in the screen can be measured. Reference numeral 11 denotes a semi-transmissive mirror provided in the finder optical path, and a part of the finder light is reflected upward in FIG. 1 and is imaged on the focus plate 7 on the known area sensor 13 by the imaging lens 12. Re-image the image. On the other hand, the finder light transmitted through the semi-transmissive mirror 11 is guided to the eyepiece lens 14 so that the photographer can enlarge and observe the subject image. Reference numeral 15 denotes a strobe unit arranged on the front surface of the pentaprism 8, which has a retractable mechanism that pops up when emitting light. Reference numeral 16 denotes a shake detection sensor such as a vibrating gyroscope arranged to detect angular velocities in the pitch direction and the yaw direction with respect to the optical axis K in order to detect camera shake of the photographer.

Reference numeral 31 denotes an aperture provided in the taking lens 1;
Reference numeral 32 denotes an aperture driving device, 33 denotes a focus lens driving motor, and 34 denotes a focus lens driving device including a driving gear and the like. An encoder 35 includes a photocoupler and a pulse plate attached to the focus lens driving device 34. The encoder 35 detects a driving amount of the focus lens driving device 34 and transmits the detected amount to the lens control circuit 104. The focus lens drive motor 33 is driven by a predetermined amount based on the information and the information on the focus lens drive amount from the camera, and the focus lens 1a is moved to the in-focus position.

Reference numeral 36 denotes a subject distance detecting device for detecting the absolute position of the photographic lens 1 and obtaining the distance from the camera to the subject. The subject distance detecting device 36 comprises, for example, a code pattern of about 4 bits from the closest position to infinity. The subject distance at the in-focus position can be detected using the brush contact. Reference numeral 37 denotes a focal length detecting device that detects the focal length of the photographing lens 1, and is capable of detecting focal length information according to the zooming lens using a brush contact (not shown). Reference numeral 38 denotes a shift lens driving device that moves the blur correction lens 1b on a plane orthogonal to the optical axis K. Reference numeral 39 denotes a mounting contact serving as an electrical interface between a known camera and a lens.

FIG. 2 is a top view of the single-lens reflex camera having the above configuration.

In FIG. 1, reference numeral 41 denotes a release button,
Reference numeral 2 denotes an LCD for an external monitor; 43, a shooting mode dial for selecting a main switch of the camera and a shooting mode;
Reference numeral 4 denotes a self-timer setting button.

The photographing mode dial 43 has settable photographing modes. Among them, reference numeral 43a denotes an image zone in which a photographing mode suitable for a scene represented by pictograms is set. There are four modes: portrait mode suitable for landscape photography, landscape mode suitable for landscape photography, close-up mode suitable for close-up photography, and sports mode suitable for moving body photography. In the present embodiment, an operation when the photographer sets any one of the photographing modes in the image zone will be described. Reference numeral 43b denotes a shooting mode in which the camera sets a predetermined shooting function in advance assuming general shooting called a green mode.

FIG. 3 is a block diagram showing a configuration of an electric circuit built in the single-lens reflex camera having the above-described configuration. The same members as those in FIG. 1 are denoted by the same reference numerals.

A microcomputer central processing unit (hereinafter, referred to as a CPU) 101 of a microcomputer built in the camera body includes a recognition information processing unit 101 therein.
a and the EEPROM 101b. And the C
Various control circuits are connected to the PU 101 to control various functions of the camera.

The photometric circuit 102 amplifies the signal from the photometric sensor 10, performs logarithmic compression and A / D conversion on the amplified signal, and provides the CPU 10 with luminance information of each sensor corresponding to a plurality of areas in the screen.
Send to 1. The focus detection circuit 103 performs A / D conversion of outputs from a plurality of sets of line sensors 67 so that phase difference type focus detection can be performed at a plurality of positions in the screen, and performs CP conversion.
Transmit to U101. The lens control circuit 104 disposed on the lens 1 is connected to the CPU 101 via the mount contact 39.
It controls the focus lens driving motor 33 based on the control information from the camera, controls the aperture driving device 31, and controls the shift lens driving device 38.

The flash control circuit 105 includes a CPU 101
The built-in flash 15
Is popped up, the built-in strobe 15 is caused to emit light in synchronization with the release, and a signal from the light control sensor 55 is received to control the amount of strobe light emission. SW1 is a switch for turning on the first stroke of the release button 41 to start photometry, AF, and a subject recognition operation which is one of the features of the present invention. SW2 is a camera for turning on the second stroke of the release button 41. Release switch. SW-DIAL
Is a dial switch provided in the photographing mode dial 43. The status signals of the above switches and various switches (not shown) arranged at various places of the camera are SW.
The signal is input to the signal input circuit 106,
It is transmitted to PU101.

The LCD drive circuit 107 has a known configuration for driving a liquid crystal display (LCD), and displays an aperture value, a shutter time, a set photographing mode, and the like in accordance with a signal from the CPU 101. , Monitor LC
It is displayed on D42 or a not-shown display LCD in the viewfinder. The shutter control circuit 108 controls the shutter magnet Mg-1 for running the front curtain and the shutter magnet Mg-2 for running the rear curtain when energized, and exposes the photosensitive member to a predetermined amount of light. The motor control circuit 109 controls the motor M1 for winding and rewinding the film, and the motor M2 for charging the main mirror 2 and the shutter 4. The shutter control circuit 108 and the motor control circuit 109 execute a series of camera release sequences.

The shake detection sensor 16 outputs an angular velocity due to camera shake, and a shake detection circuit 111 converts the angular velocity into an actual shake amount and sends it to the recognition information processing unit 101a. At the same time, it is sent to the camera shake correction circuit 110, which calculates the drive amount of the shift lens corresponding to the actual camera shake correction amount and sends it to the CPU 101. In response to this, the CPU 101 sends control information to the lens control circuit 104 as described above, and controls the blur correction lens 1b via the shift lens driving device 38 to perform camera shake correction.

Here, a description will be given of the connection to the recognition information processing unit 101a formed inside the CPU 101, which is described for easy understanding.

First, the aforementioned shake detection circuit 111 is connected, and the amount of camera shake is input as an information signal. Next, the subject position information and the focal length information of the photographing lens are input by the lens position detection circuit 112 from the signals of the subject distance detection device 36 and the focal length detection device 37 provided in the lens.

The area sensor 13 is a color area sensor having the number of pixels (approximately several hundred thousand pixels) necessary to mainly detect face information of a subject, and is a known CCD or CM.
The sensor control circuit 113 is controlled by a known driving method. The subject image input to the sensor control circuit 113 is sent to the image processing circuit 114 and
/ D conversion and stored in a memory unit (not shown) by the recognition information processing unit 101a as necessary, and image data required by the subject recognition circuit 115 and the subject image motion vector detection circuit 116 in accordance with an algorithm described later. Supplied as Image data corresponding to a shooting screen is output to the subject recognition circuit 115, and image data having a predetermined time difference is output to the subject image motion vector detection circuit 116.

The subject recognition circuit 115 detects the size and number of the face, for example, if the subject has a face. Also,
If it is a landscape, it detects what kind of landscape scene it is, and details will be described in an algorithm described later. The subject image motion vector detection circuit 116 detects a motion vector by performing a two-dimensional correlation operation from image data having a time difference for each of a plurality of regions, and the details will be described later in an algorithm.

As described above, the shake detection circuit 111, the lens position detection circuit 112, the object recognition circuit 115, and the motion vector detection circuit 116 are connected to the CPU 101, and the camera shake information and the focus of the lens are transmitted to the recognition information processing unit 101a. Distance information, subject distance information, subject recognition information,
The motion vector information of the subject image is input.

Reference numeral 117 denotes a recognition information recording circuit including a recording magnetic head, which records the subject information finally detected by the recognition information processing unit 101a as recognition information on the magnetic recording layer of the film. Of course, the present invention is not limited to this magnetic recording method, but may be a semiconductor memory such as an EEPROM or a flash memory, an optical recording means such as an MO or DVD-ROM, or other removable recording media.

Next, the operation of the single-lens reflex camera configured as described above will be described.

FIG. 4 is a flowchart showing up to the photographing operation of the single-lens reflex camera having the above configuration.

First, in step # 101, when the photographing mode dial 43 is released from the LOCK position and one of the photographing modes in the image zone 43a is set, power is supplied, and the camera is ready for photographing. Next, in step # 102, it is detected whether the switch SW1 which is turned on in the first stroke of the release button 41 is turned on,
If it is ON, the process proceeds to step # 103, and if it is OFF, the detection of step # 102 is repeated. Step # 10
When the process proceeds to step 3, what is the set photographing mode is read. If the photographing mode is the portrait mode, step # 1 is executed.
04, in the case of the landscape mode, to step # 105, in the case of the close-up mode, to step # 106, and in the case of the sports mode, to step # 107.

In the portrait mode, step #
The process proceeds to step # 108 via 104, where the subject recognition circuit 115 performs face recognition processing. That is, the subject recognition circuit 115 sequentially captures the subject images at the predetermined frame rate by the area sensor 13 and performs face recognition processing on the captured subject images to determine the size and number of faces of the subject images. At this time, since the mode is the portrait mode, the subject recognition processing is executed by an algorithm suitable for face recognition. Details of face recognition will be described later. Next, the process proceeds to step # 112, where the CPU 101
Fetches this result from the object recognition circuit 115 into the recognition information processing unit 101a, and determines the AE control or the control operation of the AF operation in the AE program diagram corresponding to the object according to the face recognition result.

FIG. 5 shows an example of this control content. The vertical item in FIG. 5 indicates the number of recognized faces, and the horizontal item indicates the size of the face area.

First, when the number of faces is one and the size of the face area is larger than a predetermined size, the AE (automatic exposure control) stops down a little when it is on the long focal length side based on the open F number of the attached lens. It is assumed that the AE program has a narrow aperture priority. This is the most human-centered photo, and we give priority to taking pictures so that only the face is in focus. However, if the attached lens becomes a long focus, it will focus only on a part of the face, so it is a program to narrow down a little. In the AF (auto focus control), the lens is driven so as to match the average focus position of the focus detection points included in the face area.

When the size of the face area is smaller than the predetermined size, the AE is a program for giving priority to the open FNo (F number) of the mounted lens, and the AF focuses on the closest face area in the face area. The lens is driven so as to match. This is also set so that the background is blurred as much as possible for the size of the face.

In the case of one person, the built-in strobe 15 is forcibly fired in any case, and the dimming correction amount is one step of minus. This is performed in order to generate a catchlight on the subject's pupil.

Next, if the number of faces is from two to three, the AE is controlled by the aperture value at which each person is focused irrespective of the size of the face as described above. Depth priority program. Further, the AF is set so that the lens is driven to a position that internally divides the near point and the far point by “7:10”. This focuses on both the near-point and far-point persons, and blurs the other distant background as much as possible, depending on the amount of defocus between multiple focus detection points. The aperture is set to be within the depth of field, and the aperture value is prioritized. This is not a so-called portrait photograph, but a photographing mode that is focused on a person is set.

Next, if the number of faces is four or more, or unspecified, regardless of the size of the face, AE is determined by the aperture value = F8
Aperture priority AE program for narrowing down according to the focal length based on In the AF, the lens is driven so that the average position of each face is focused. This is based on focusing on all the people, but with a setting that is narrowed down to a certain extent so that both the person and the background are sharply captured, and further narrowed down as the focal length becomes longer. . A so-called group photographing mode is set such that a photograph in which both the person and the background are harmonious overall can be taken.

As described above, even if the portrait mode is selected in the case of the conventional camera, the AE is fixed to the aperture priority AE at the open FNo and the AF is fixed to the one-shot AF.
As described above, camera control can be further optimized in accordance with the situation of various subjects in photographing a person.

Returning to FIG. 4, when the content of the camera control based on the face recognition result when the portrait mode is set as described above is determined, the process proceeds to step # 116 to actually perform the control.

If the mode is the landscape mode, the process proceeds to step # 109 via step # 105, where the object recognition circuit 115 performs a scene recognition process. That is,
The subject recognition circuit 115 sequentially captures the subject images at a predetermined frame rate by the area sensor 13,
A scene recognition process is performed from the captured subject image to recognize what kind of landscape scene the subject is. At this time, since the mode is the landscape mode, the subject recognition processing is executed by an algorithm suitable for recognizing what kind of landscape is. The details of the scene recognition will be described later. Next, the process proceeds to step # 113, where the CPU 101 outputs the result to the recognition information processing section 101a by the object recognition circuit 115.
To determine the AE control and AF operation control contents for the scene according to the scene recognition result.

For example, if there is a green area having an area larger than a predetermined area, the green area is recognized as a landscape of a forest. (For example, 0.3 to 0.5 step plus). If there is a blue area or a high-brightness area with an area larger than a predetermined area at the top of the screen, it will be recognized as a landscape including the sky. Try to take steps to reduce it. In AF, a limiter is provided at a position closer than a predetermined distance so as not to focus.

As described above, in the case of a conventional camera, even if the landscape mode is selected, the photometry method is controlled directly by the photometry value of the normal evaluation photometry or the average photometry. The camera control can be further optimized.

If the mode is the close-up mode, the process proceeds to step # 110 via step # 106, where the object recognition circuit 115 performs object recognition processing. That is, the subject recognition circuit 115 sequentially captures the subject images at a predetermined frame rate by the area sensor 13 and performs object recognition processing on the captured subject images to recognize the size and depth of the main subject among the subjects. .
At this time, since the mode is the close-up mode, the subject recognition processing is executed by an algorithm suitable for recognizing the size of the object. Details of the object recognition will be described later. Next, the process proceeds to step # 114, where the CPU 101 fetches the result from the object recognition circuit 115 into the recognition information processing unit 101a, and determines AE control or AF operation control content corresponding to the scene according to the object recognition result.

For example, if there is a region of an object having an area larger than a predetermined area in the screen and information is recognized that the object has a predetermined depth or more, the depth is increased by about 2/3 from the closest point in the depth direction. The aperture is set to have a depth of field such that focus is achieved so far, and the AF also drives the lens to a position that satisfies it. In the case of an object having a size and depth other than that, the aperture is set to have a depth of field that can focus on the entire depth, and the AF is also driven to a position that satisfies the condition. This means that if the subject to be taken close-up becomes larger than a certain level, it will be necessary to narrow down considerably in order to focus on the whole, and it will exceed the limit of actual focusing, and shoot subjects that occupy a lot of such screens In such a case, focusing on the whole is often not very suitable as a finish of a close-up photograph.

As described above, in the case of the conventional camera, even if the close-up mode is selected, the aperture priority AE control with the fixed depth of field having a large depth of field is merely performed. The camera control can be further optimized according to the subject to be photographed.

If the mode is the sports mode, the process proceeds to step # 111 via step # 107, and the subject image motion vector detecting circuit 116 performs a process of detecting the motion of the subject image. That is, the subject image motion vector detection circuit 11
Numeral 6 sequentially captures the subject image at a predetermined frame rate by the area sensor 13 and performs a motion detection process of the subject image from the captured subject image. At this time, since the mode is the sport mode, the subject recognition processing is executed by an algorithm suitable for detecting the motion of the subject, not by what the subject is. The details of the motion detection will be described later. Next, the process proceeds to step # 115, where the CPU 1
01 fetches the result into the recognition information processing unit 101a from the image movement detecting circuit 116,
6. The information processing unit 1 recognizes information obtained by converting the angular velocity due to camera shake into an actual shake amount by the shake detection circuit 111.
01a. Thereby, the recognition information processing unit 101
“a” subtracts the camera shake information obtained by the blur detection circuit 111 from the motion vector information of the subject image obtained by the subject motion vector blur detection circuit 116 to calculate the actual motion amount of the subject. This is because the former motion vector of the subject image is detected including the camera shake amount of the photographer. The control content of the AE control and the AF operation corresponding to the scene is determined according to the obtained motion detection result of the subject.

For example, AE is a shutter speed priority AE that is automatically set to a shutter speed that does not cause subject blur according to the amount of movement of the subject, and AF always performs defocus detection and focuses continuously in accordance with it. In addition to setting the servo mode to match
Move the focus detection point according to the moving direction of the subject,
It is set to execute a so-called tracking mode. This is to supplement a moving subject in the sports mode as a top priority target.

As described above, in the conventional camera, even if the sports mode is selected, the combination of the aperture and the shutter speed is set to the shutter speed priority exposure program which is set to the higher shutter speed side than the standard AE program. Although the AF mode is set uniformly and the AF mode is merely set to the servo mode, the control of the camera can be further optimized according to the subject at the time of moving object shooting as described above.

As described above, steps # 112 to # 1
AE by recognition information processing corresponding to each of the 15 shooting modes
When the control contents of the control and the AF operation are determined, the camera is actually controlled in accordance with the determined control contents.

When any of steps # 112 to # 115 is executed, the process proceeds to step # 116, where the output of the focus detection line sensor 66 is fetched, and the defocus amount at each focus detection point is calculated. Then, the defocus amount is calculated in order to finally bring the subject into focus according to the control contents determined in steps # 112 to # 115. Then, proceeding to step # 117, the CPU 101 sends a signal to the lens control circuit 104 to drive the focus lens 1a by a predetermined amount according to the defocus amount, and adjusts the focus of the lens. Also, a so-called AF mode of one shot or servo is set. Then step #
At 118, focal length information and subject distance information are fetched from the lens position detection circuit 112 as lens position information.
Next, proceeding to step # 119, the CPU 101 causes the photometric circuit 102 to perform photometry. Then, the CPU 101 detects the luminance of the entire photometry area, performs a photometry operation for weighting the specific area and performs exposure correction in accordance with the control contents determined in steps # 112 to # 115, and the CPU 101 takes in the photometry calculation.

Next, at step # 120, the CPU 101
Drives the shake detection sensor 16 and the shake detection circuit 111
The information converted into the actual blur amount is taken into the photographing mode setting circuit 101a. The information converted into the actual blur amount is also sent to the camera shake correction circuit 110,
Here, the driving amount of the shift lens corresponding to the actual amount of camera shake correction is calculated, and the CPU 101 also takes in the driving amount. Next, in step # 121, the CPU 101 sends control information to the lens control circuit 104 based on the shift lens driving amount, controls the shake correction lens 1b via the shift lens driving device 38, and performs camera shake correction.

Next, in step # 122, the exposure value determined based on the photometric value obtained in step # 119 and the set photographing mode, that is, the shutter speed and the aperture value, are stored in the LCD drive circuit 107. Is displayed on the monitor LCD 42. In the following step # 123, the above step # 1
It is determined whether or not the situation requires flash photography in accordance with the control contents determined in steps 12 to # 115. If necessary, the built-in flash 15 is popped up in step # 124.

In the next step # 125, the release button 4
It is detected whether the switch SW2, which is turned on in the second stroke of No. 1, is turned on.
And the above operation is repeated. On the other hand, if it is ON, the flow advances to step # 126 to execute the camera release sequence. Specifically, first, the CPU 101 supplies power to the motor M2 via the motor control
And drives the aperture driving device 31 via the lens control circuit 103 to stop down to a predetermined aperture. Next, the shutter is driven at a predetermined shutter time by the shutter control circuit 108, and if necessary, at the time when the front curtain has been driven, the built-in flash 15 is transmitted via the flash control circuit 105.
Is emitted, and the exposure to the image recording member 5 is completed. Thereafter, the motor M2 is energized again to perform mirror down and shutter charging, and is also energized to the motor M1 to feed the film frame, thereby completing a series of release sequences.

In the case of a silver halide film camera, during the feeding, the object recognition operation of steps # 108 to # 111 and the object information obtained by the recognition information processing of steps # 112 to # 115 are performed. Perform magnetic writing. In addition,
In semiconductor memories other than silver halide films, such as EEPROM and flash memory, optical recording means such as MO and DVD-ROM, and other removable recording media, recording may be performed immediately after the switch SW2 is turned on. The recording of the photographed image may be performed after the recording of the photographed image is completed, as long as it is not a film but an electronic image sensor.

As described above, in the first embodiment, the object recognition (including the motion detection of the object) is performed in accordance with the selected photographing mode, and the recognition is performed in accordance with the characteristics of each mode. Processing is in progress.

Next, steps # 108 to # 111 in FIG.
The object recognition operation in each shooting mode will be described in detail with reference to FIGS.

FIG. 6 shows an object recognition algorithm when the portrait mode is set. Step # 1
It is a flowchart which performs face recognition in 08.

When this operation starts, the object recognizing circuit 115 determines in step # 201 that the EEPROM 101b
The various parameters used for the face recognition algorithm are read from the memory and stored in a predetermined address of a memory (not shown). The parameters used for this recognition algorithm are shown together with other recognition algorithms in FIG. Next, the process proceeds to step # 202, in which variables generated in the process are initialized. Next, the process proceeds to step # 203, in which an initial setting of an area for performing face recognition is performed. In this case, since the portrait mode is set, the subject is almost at the center and in the vicinity thereof, and in most cases is located at the upper part of the screen. In particular, depending on the posture of the camera, for example, in vertical position shooting, the ratio of the upper part of the screen is considerably high. Therefore, the parameters are set so that about 50% of the area including the upper area in the screen is limited as the initial recognition target area.

Next, the process proceeds to step # 204, where the sensors are preliminarily stored in the limited area in step # 203, and then read out in blocks of several pixels. Next, the process proceeds to step # 205, where it is determined whether the luminance of each block is low, medium, or high with a predetermined threshold. Only the surrounding area is further limited as an area for performing face recognition. This is because the face portion of a person is a region of relatively medium luminance, the eyes and hair are regions of low luminance, and the background includes a region of high luminance. By limiting the area in which the recognition processing is performed in two stages, the processing load of the recognition is considerably reduced. This threshold is used as a parameter in the EEPROM 101.
b and read in step # 201.

Next, proceeding to step # 206, the area sensor 13 performs an accumulation operation to perform face recognition, and reads an object image in an area where face recognition is performed. Then, in the next step # 207, the brightness, hue, saturation, chromaticity, and brightness are calculated from the read R, G, and B values of each pixel of the area sensor 13.
A color feature amount such as a color difference is obtained. The feature amount to be obtained here is to determine a feature amount in a color space in order to determine a skin color area in the next step. Next, proceed to step # 208,
If the feature values of these colors fall within a predetermined skin color range, the target pixel is determined to be a skin color, and a skin color region is extracted. Here, the area of the color regarded as a flesh color is set as a parameter. For example, if flicker is detected by the photometric sensor 10, it is determined as a fluorescent light and the parameters of the flesh color area are changed, or the flicker is changed by the luminance itself. In addition, even if the light source changes, the skin color can be extracted.

Then, the process proceeds to a step # 209, wherein an edge detection process is performed on the image composed of the pixel region determined to be flesh color and its peripheral region. There are several methods for the edge detection process, but in this case, it is performed by a method based on the second-order difference.
The second order difference is equivalent to a second derivative on a continuous space, and is typically represented as follows.

▽ ² f _ij = f _{i + 1, j} + f _{i-1, j} + f _{i, j + 1} + f _{i, j-1} -4 * f _{i, j} (1) where f _ij Represents the pixel output value (lightness) of the i-th row and j-th column, and ▽ ² f _ij represents the Laplacian (second derivative) of f _ij .

This is represented by a weight matrix as shown in FIG. 8. In this embodiment, eight neighboring Laplacians are calculated centering on a target pixel and surrounding eight pixels. If this value is larger than a predetermined threshold, binarization is performed using the target pixel as an edge pixel. As described above, when the edge is detected by performing the second-order difference, the zero-cross point of the difference output is obtained. Thus, even a sharp edge or a gentle edge is hardly affected by the edge strength, and the edge is easily detected. Be sensitive to However, in the present algorithm, the target region is considerably limited by skin color detection, and thus such a method is effective.

Next, the process proceeds to step # 210, where a matching process for determining whether the face is a face is performed. Here, it is assumed that template matching is used, in which a partial image (template) in the input image is moved and a matching position is detected. First, a plurality of elliptical face templates having different sizes and different ratios of the major axis and the minor axis are created. The threshold of the size of the face is set to such a size that the user is sufficiently recognized on the screen and is considered to be a photograph centered on a person (for example, the whole body is about の of the screen). These templates may be created in advance and stored in a memory (not shown). The face template is binarized by an ellipse, and the template outline has a width of several pixels in order to increase the matching degree with the actual face outline. Further, the matching process is continued and the above step # 20 is performed.
The degree of matching between the edge image of the face candidate obtained in step 9 and the face template is determined. There are some things as a measure for determining the degree of matching, for example, an edge image f of the face candidate (x, y), template t (x, y), ∬ and R is the entire area of the template _R (ft) ² dxdy... (2) A matching process is performed using several types of face templates in this manner, and a template that best matches the target pixel is obtained.

Next, proceeding to step # 211, if the calculated squared error value is smaller than a predetermined threshold value, it is determined that the area surrounded by the face template that best matches the target pixel is the face area. Then, the process proceeds to step # 212. Then, in step # 212, the number and size of the face area determined as the face area are obtained and output, and the face recognition processing in step # 108 in FIG. 4 ends.

If there is no area determined to be a face in step # 211, the flow advances to step # 213 to determine the number of face recognition operations performed so far. If the number is less than the predetermined number, the process proceeds to step # 214, and the parameters of the recognition algorithm are changed so as to be different from the previous detection and to perform the recognition process again. For example, the limited range of the recognition area is widened, the width of the flesh color feature amount is widened, and the threshold value of the edge detection is changed.
In step 02, the face recognition process is performed using the new recognition parameters.

If face recognition has been performed a predetermined number of times in step # 213, it is determined that the subject is not suitable for portrait mode, and the camera control mode is automatically changed to green mode. After setting, the process proceeds to step # 116 in FIG. The green mode is a so-called "camera entrusting mode", which is the same setting as the shooting mode shown at 43b in FIG. 2, and which is set to a control mode suitable for a very general shooting. FIG. 5 is a flowchart illustrating a scene recognition algorithm in step # 109 in FIG. 4, which is a subject recognition algorithm when set.

When this operation starts, the object recognizing circuit 115 determines in step # 301 that the EEPROM 101b
, Various parameters used for the scene recognition algorithm are read and stored at predetermined addresses in a memory (not shown). Note that the parameters used in this recognition algorithm are collectively shown in FIG. 7 as described above. Next, the process proceeds to step # 302, where variables generated in the process are initialized. Then, in the next step # 303, an initial setting of an area for performing scene recognition is performed. Here, since the mode is the landscape mode, the recognition area is not particularly limited, and almost the entire area that can be observed with the viewfinder is targeted.

Next, the process proceeds to step # 304, where the sensors are preliminarily stored in the limited area in step # 303, and then read out in blocks of several pixels. Next, in step # 305, whether the luminance of each block is low, medium, or high is compared with a predetermined threshold value.
In the landscape mode, since the subject scene extends from low luminance to high luminance, there is no limitation based on this luminance area, and therefore, parameters are set so that there is no particular limitation even through this step. Next step #
In step 306, the area sensor 13 performs an accumulation operation to perform scene recognition, and reads a subject image in an area where scene recognition is performed. Then, in the next step # 307,
From the read R, G, and B values of each pixel of the area sensor 13, color feature amounts such as lightness, hue, saturation, chromaticity, and color difference are obtained. The feature amount to be determined here is to determine a feature amount in a color space in order to determine a natural color area in the next step. Next, proceeding to step # 308, if the feature amounts of these colors fall within the predetermined color range, it is determined that the target pixel is a natural color, and a color region is extracted. Here, the area of the color considered as a natural color is set in the parameter,
For example, here, the parameters are set so as to determine a region from green to dark green to extract a forest scenery and a blue region to extract the sky.

Then, the process proceeds to a step # 309, wherein an edge detection process is performed on an image composed of the pixel region determined to be a natural color and its peripheral region. There are several methods for edge detection. Here, edge detection is performed using the gradient of an image, and the Roberts gradient is used as a well-known method.

[(F _{i, j} −f _{i + 1, j + 1} ) ² + (f _{i, j + 1} −f _{i + 1, j} ) ² ] ^1/2 ... (3) As can be seen, the Roberts gradient is suitable for oblique edge detection, which is also appropriate in natural scenes such as landscapes. If this value is larger than a predetermined threshold, binarization is performed using the target pixel as an edge pixel. The gradient as the threshold is set relatively gentle, and parameters are set so as to detect even a gentle change.

Then, the process proceeds to a step # 310, wherein a region extracting process for recognizing a scene is performed. Although there are several methods, it is assumed that connected component processing is performed. A binarized image created by edge processing often contains a noise component. These noise-like small components, protrusions, holes, etc. can be processed by graphic fusion. Alternatively, a characteristic amount (area or width) of each connected component is measured, and a connected component is created by eliminating components whose values are equal to or less than a certain threshold, and a characteristic region is extracted as a scene. Note that the contour may be extracted by further converting to a line figure by thinning processing.

Next, the process proceeds to step # 311. If the extracted region is larger than a predetermined threshold value (here, the size to be displayed on the screen is 5% or more), it is determined that there is a region serving as a landscape scene, and step # 311 is performed. Proceed to 312. In step # 312, the position of the area determined to be the scene area (here, the area considered to be a forest, the area considered to be the sky) is determined on the screen, and is output.
The scene recognition processing of 109 is completed.

If there is no area determined to be a scenery scene in step # 311, the flow advances to step # 313 to determine the number of scene recognition operations performed so far. The parameters are changed so that the parameters of the recognition algorithm are different from those of the previous detection and the recognition processing is performed again. For example, the width of the natural color feature amount is increased, the threshold value of the edge detection is changed, and the like, and the process proceeds to step # 302, where the scene recognition processing is performed again with the new recognition parameters.

If a predetermined number of scenes have already been recognized in step # 313, it is determined that the subject is not suitable for the landscape mode, and the camera control mode is automatically changed to the green mode. Then, the process proceeds to step # 116 in FIG.

As described above, the scenery scene recognition algorithm is different from the face recognition algorithm described above in that the recognition algorithm itself is changed to one suitable for scene detection, and the same algorithm is used to change the parameters to perform recognition processing. To increase the accuracy of recognition.

As a method of region extraction, a region forming method (Region Growing) may be performed. This starts from the initial division into small sets of areas, and repeats the merging process while judging whether or not adjacent areas can be regarded as a uniform area even if they are merged. There is a discontinuity of the density value in the part, a distribution range of the density value in the area, a maximum deviation from the average density, and the like. However, although the accuracy is improved as a recognition algorithm, it imposes a considerable load on computational power.

FIG. 10 shows an object recognition algorithm when the close-up mode is set.
It is a flowchart which performs the object recognition in 10.

When this operation is started, the object recognizing circuit 115 reads various parameters used for the object recognizing algorithm from the EEPROM in step # 401 and stores them at predetermined addresses in a memory (not shown). Next, the process proceeds to step # 402, in which variables generated in the process are initialized. Next, in step # 403, an area for performing scene recognition is initialized. Since it is a close-up mode here,
Most subjects are located almost at the center, but the proportion of the subject on the screen is quite large. Therefore, the parameters are set so as to limit an area of about 70% at the center of the screen as an initial recognition target area.

Next, the process proceeds to step # 404, where the sensors are preliminarily stored in the limited area in step # 403, and then read out in blocks of several pixels. Then, in the next step # 405, whether the luminance of each block is low, medium, or high is compared with a predetermined threshold value. The area to be recognized is further limited. Basically, the low-luminance area is excluded, and this is limited because the object to be subjected to close-up photography has almost no low luminance when compared with the luminance.

Next, the process proceeds to step # 406, in which the area sensor 13 performs an accumulation operation to perform object recognition, and reads a subject image in an area where object recognition is performed. Then, in the next step # 407, the read area sensor 13
From the R, G, and B values of each pixel, color feature amounts such as brightness, hue, saturation, chromaticity, and color difference are obtained. In the next step, the feature amount to be obtained in the color space is determined in order to determine the same color area. Next, the process proceeds to step # 308, in which the target pixels are determined to be the same color if the feature amounts of these colors are within the predetermined range of some types of colors, and the same color area is determined. Extract. Here, the area of the color regarded as the same color is set in the parameter.

Next, the process proceeds to step # 409, where an edge detection process is performed on the image composed of the pixel region determined to be the same color and its peripheral region. Here, the Roberts gradient is set in the same manner as in step # 309 for scene recognition. Used. If this value is larger than a predetermined threshold, binarization is performed using the target pixel as an edge pixel. The gradient as the threshold is set tighter than in step # 309, and the parameters are set so as to detect a change that is somewhat clear.

Then, the process proceeds to a step # 410, wherein a region extracting process for recognizing the object is performed. Here, it is the same as step # 310, and the description is omitted. As a method of region extraction, edge processing (threshold processing) was performed after extracting the same color region. However, at the stage of extracting the color feature amount and extracting the region of the same color, the above-described region forming method (Region Growing) was used. May be performed.

Next, proceeding to step # 411, if the extracted region is larger than a predetermined threshold value (here, the size to be displayed on the screen is 10% or more), it is determined that there is a region of the object to be photographed in close proximity, and the process proceeds to step # 411. Proceed to # 412. Then, in step # 412, the position and size of the area determined to be the close-up object on the screen are obtained and output. In addition, a plurality of defocus amounts of the target area are obtained from the focus detection circuit 103, and an approximate depth is detected and output from this. This completes the object recognition processing in step # 110 of FIG.

If there is no area determined to be a close-up object in step # 411, the flow advances to step # 413 to determine the number of times of object recognition performed so far. Proceeding to 414, the parameters of the recognition algorithm are changed so as to be different from the previous detection and to perform the recognition process again. For example, the width of the feature amount regarded as the same color is widened, the threshold value of the edge detection is changed, and the like.

If the object recognition has been executed a predetermined number of times in step # 413, it is determined that no subject is suitable for the close-up mode, and the control mode of the camera is automatically changed to the green mode. Then, the process proceeds to step # 116 in FIG.

As described above, the close-up scene recognition algorithm uses almost the same algorithm as the above-described scene recognition algorithm. However, by changing the parameters for recognition and performing the recognition process, the object can be more accurately recognized. Specialized to recognize.

FIG. 11 is a flowchart showing the subject recognition algorithm in step # 111 in FIG. 4, which is the subject recognition algorithm when the sport mode is set.

When this operation starts, the object recognizing circuit 115 reads various parameters used for the motion detection algorithm from the EEPROM in step # 501,
It is stored at a predetermined address in a memory (not shown). And
In the next step # 502, variables generated in the process are initialized. Next, the process proceeds to step # 503, in which an area for performing scene recognition is initialized. In this case, since the mode is the close-up mode, the subject is mostly located almost at the center. Although the proportion of the subject on the screen is not so large, the subject moves considerably.
Are set so as to limit the region as an initial recognition target region.

Next, the process proceeds to step # 504, in which the area sensor 13 performs an accumulation operation to detect a motion, and reads a subject image in an area where the motion is to be detected. Here, in the above three algorithms, preliminary storage of the area sensor 13 is performed, but in the sports mode, this step is omitted because it is desired to minimize the time until detection. Next, the process proceeds to step # 505, where the read area sensor 13 is read.
The edge detection process is performed on the image using only the G component as the luminance signal component from the R, G, and B values of each pixel. Also in this case, the gradient of Roberts is used as in step # 309. If this value is larger than a predetermined threshold, binarization is performed using the target pixel as an edge pixel. The gradient as the threshold is set tighter than in step # 309, and the parameters are set so as to detect a change that is somewhat clear.

Then, the process proceeds to a step # 506, wherein the detected edge image is stored in a predetermined memory (not shown). Next, in step # 507, the image of the next frame is read. In the following step # 508, the above-described step # 5
As in 05, an edge detection process is performed on the image using only the G component as a luminance signal component. Next, in step # 509, the edge image of the previous frame is read from the memory, the correlation with the image of the current frame is calculated, and the motion of the image is detected based on the correlation value. As calculation formula previous frame image f _i-1 of the (x, y), the image f _i of the current frame
Assuming that (x, y), the shift amount ξ, η is min ΣΣ [f _i−1 (x−ξ, y−η) −f _i (fx, y)] ² (4) or min ΣΣ | f _{i− 1} (x−ξ, y−η) −f _i (fx, y) | (4) ′, which is to be obtained.

More specifically, two consecutive frames of images are compared. This comparison is performed while shifting the reading position of the image data stored in the memory by one pixel. The sum of the difference between the two pixels is obtained, and the value is stored in a different area each time the value is shifted by one pixel. The sum of the differences of each pixel is bi
This is done by comparing t and counting the number of different bits. The same comparison is performed on the next line and thereafter, and the sum of the differences between the pixels at that time is added to the area where the value of the first line is stored, and this becomes the correlation value.

Then, the process proceeds to a step # 510, where a point where the correlation value becomes the smallest is obtained and is set as a motion vector (motion amount) from the previous time in the area. That is,
The number of pixels shifted in the vertical and horizontal directions when the correlation value becomes minimum is the amount of movement in the vertical and horizontal directions. Instead of such means, projected images may be formed in the horizontal direction and the vertical direction, and the motion vector may be obtained from the correlation value (movement amount of the position of the center of gravity).

Next, proceeding to step # 511, if the motion vector is larger than the predetermined amount, proceeding to step # 512,
It determines that there is a moving subject and outputs a motion vector amount.
This completes the motion detection processing of step # 111 in FIG.

If the motion vector is smaller than the predetermined amount in step # 511, the flow advances to step # 513.
It is determined whether the switch SW2 is ON. If the switch SW2 is ON, the control mode of the camera is automatically set to the green mode, and the process proceeds to step # 116 in FIG. ON
If not, the process returns to 502 to repeat the motion detection.

As described above, in the sports mode, a moving object is detected rather than recognizing what it is. The detection algorithm is performed by changing a recognition parameter.

(Second Embodiment) Next, a second embodiment of the present invention will be described with reference to the flowchart of FIG. The circuit configuration and the like of the single-lens reflex camera are the same as those in the first embodiment.

In the first embodiment, the subject recognition processing is performed in all shooting modes. However, there are some difficulties due to the limited arithmetic capacity of the CPU that can be mounted on the camera. Therefore, in the second embodiment, the subject recognition process is performed only in a photographing mode in which a person who is a general photographer is considered to be photographed.

In FIG. 12, in step # 601,
The photographing mode dial 43 is removed from the LOCK position, one of the photographing modes in the image zone 43a is set, power is supplied, and the camera is ready for photographing.
Next, in step # 602, it is detected whether or not the switch SW1 which is turned on in the first stroke of the release button 41 is turned on. If the switch SW1 is turned on, the process proceeds to step # 603 and turned off.
If so, the detection in step # 602 is repeated.

At step # 603, it is determined whether the self-timer setting button 44 has been pressed and the self-timer mode has been set. If the self-timer mode has been set, the process proceeds to step # 609, where The subject recognition circuit 115 performs face recognition processing. That is, the subject recognition circuit 115 sequentially captures the subject images at the predetermined frame rate by the area sensor 13 and performs face recognition processing on the captured subject images to determine the size and number of faces of the subject images. This is the same as when the portrait mode of the first embodiment is set. When the self-timer shooting is performed, the face recognition processing is forcibly performed because the photograph is almost a person, and the self-timer shooting is performed. The functions are enhanced and the operability of the photographer is simplified.

If not set, step # 6
The process proceeds to step 04 to read what the set shooting mode is, and to select one of the portrait mode (# 605), the landscape mode (# 606), the close-up mode (# 607), or the sports mode (# 608). Proceed to.

If the mode is the portrait mode, the flow advances to step # 609 via step # 605, and the same operation as in the first embodiment is performed. In landscape mode,
If the mode is the close-up mode or the sports mode, step # 6
Steps # 611 to # 611 through # 06, # 607 and # 608
In step # 613, a known photographing program is set.

For example, in the case of the landscape mode, a program that changes at the ratio of “Tv: Av = 1: 2” from the point of “AE program = open F number” and “1 / focal length”, “AF mode = A landscape mode program such as "one-shot AF", "feed mode = 1 frame shooting", and "photometry mode = average photometry" is set in step # 611.

In the close-up mode, "AE program =
Aperture priority program with deeper depth of field "," AF mode = one-shot AF "," feed mode = 1 frame shot ",
A close-up mode program such as "photometry mode = partial photometry" is set in step # 612. In the case of the sports mode, the shutter speed priority exposure program set to the high-speed shutter second side, the AF mode
Mode = servo AF ”,“ feed mode = continuous shooting ”,
A sports mode program such as "photometry mode = center-weighted photometry" is set in step # 613.

When the settings are made as described above, the process proceeds to step # 614, and the following steps # 614 to # 625 are executed. These are executed in step # 614 of FIG. 4 in the first embodiment. The description is omitted because it is the same as 116 to # 127.

As described above, the subject recognition processing is limited to face recognition only, and the function is activated only when the photographing mode in which a person scene can be photographed is selected. In addition to being able to mount the object recognition on the camera and not being able to exert the effect of the object recognition function in all shooting scenes, satisfactory results can be obtained in many shooting scenes.

The second embodiment may be configured so that the face recognition process is performed not only in the self mode but also in a known remote control shooting mode. Further, in the second embodiment, not only the shooting mode such as the landscape mode but also the AF mode is set to the servo mode or the feeding mode is set to the continuous shooting mode so that the subject recognition process is not performed. May be configured.

According to each of the above embodiments, depending on which photographing mode is set, such as portrait mode, landscape mode, close-up mode, or sports mode,
Since the recognition algorithms in the object recognition circuit 115 and the object image motion vector detection circuit 116 are changed (steps # 108 to # 111 in FIG. 4 and step # 609 in FIG. 12), it is necessary to mount recognition means on the camera. This makes it easy to improve the recognition accuracy and the recognition speed, thereby enabling more advanced camera control and providing a camera with good operability.

Further, according to the set photographing mode, the subject recognition circuit 115 and the subject image motion vector detection circuit 1
Since the parameters (see FIG. 7) constituting the recognition algorithm in step 16 are changed, the subject can be accurately recognized according to the shooting mode without complicating the recognition algorithm.

When a photographing mode other than the self mode or the portrait mode is set, the operation of the face recognition algorithm is prohibited, so that it is possible to easily mount the recognition means on the camera. As a result, the photographer can concentrate on photo opportunities and composition of the screen, and can take better pictures.

(Correspondence Between the Invention and the Embodiment) In each of the above embodiments, the mode dial 43 and the self-timer setting button 44 correspond to the photographing mode setting means of the present invention, and the object recognition circuit 115 and the object image motion vector detection Circuit 116
Corresponds to subject recognition means of the present invention, and the CPU 101 (recognition information processing unit 101a) corresponds to control means of the present invention.

The present invention is not limited to the configurations of these embodiments, but may be any configuration that can achieve the functions described in the claims or the functions of the embodiments. Needless to say, there is no problem.

Although the present invention has been described with respect to an example in which the present invention is applied to a single-lens reflex camera, the present invention can also be applied to various types of imaging devices such as a video camera and an electronic still camera.

[0118]

As described above, according to the first aspect of the present invention, it is easy to mount the object recognizing means on the camera, and the accuracy and speed of recognizing the object information are improved. A camera with a good subject recognition function.

According to the second aspect of the present invention, there is provided a camera with a subject recognition function capable of accurately recognizing subject information according to a shooting mode without complicating a subject information recognition algorithm. It can be provided.

According to the fifth aspect of the present invention, the load for recognizing the subject information can be reduced and the subject recognizing means can be more easily mounted on the camera. It is possible to provide an attached camera.

[Brief description of the drawings]

FIG. 1 is a main part configuration diagram when applied to a camera having a subject recognition function according to an embodiment of the present invention.

FIG. 2 is a top view of the camera of FIG.

FIG. 3 is a block diagram showing a configuration of an electric circuit of the camera in FIG. 1;

FIG. 4 is a flowchart showing a shooting operation of the camera shown in FIG. 1;

FIG. 5 is a diagram illustrating an example of camera control based on a face recognition result in the camera of FIG. 1;

FIG. 6 is a flowchart of subject recognition in a portrait mode in the camera of FIG. 1;

FIG. 7 is a diagram illustrating recognition parameters used in a subject recognition operation in the camera of FIG. 1;

FIG. 8 is a determinant of Laplacian near 8 in the camera of FIG. 1;

FIG. 9 is a flowchart of subject recognition in a landscape mode in the camera of FIG. 1;

FIG. 10 is a flowchart of object recognition in a close-up mode in the camera of FIG. 1;

FIG. 11 is a flowchart of subject recognition in the sports mode in the camera of FIG. 1;

FIG. 12 shows a photographing operation in each photographing mode of the camera according to the second embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 6 focus detection device 10 photometry sensor 13 area sensor 16 blur detection sensor 43 shooting mode dial 104 photometry circuit 101 CPU 101a recognition information processing unit 104 lens control circuit 111 blur detection circuit 112 lens position detection circuit 115 recognition information processing circuit 116 subject image motion Vector detection circuit

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G06T 7/00 300 H04N 5/232 Z 5C066 H04N 5/232 5/235 5L096 9/64 R 5/235 G02B 7/11 N 9/64 G03B 3/00 A F-term (reference) 2H011 AA01 BA21 BB02 BB03 BB04 BB06 DA00 2H051 BA02 CB22 CE10 CE16 DA15 DA17 DA19 DA20 DA21 DB01 EB20 2H054 AA01 2H100 FF01 5C022 AA13 AB02 AB27 AB02 AB27 CA05 CA13 EF00 EF11 GA01 HA01 KD04 KD06 KE05 KE17 KM02 KM13 5L096 CA03 JA11

Claims (9)

[Claims] 1. A photographing mode setting means for setting a photographing mode of a camera, an area sensor comprising a plurality of pixels for photographing a subject, and a subject recognizing means for recognizing subject information to be photographed based on an output of the area sensor. And control means for controlling the photographing function of the camera in accordance with the result of recognition by the object recognizing means. The object recognizing means performs recognition for recognizing object information in accordance with a set photographing mode. A camera with a subject recognition function that changes the algorithm. 2. A photographing mode setting means for setting a photographing mode of a camera; an area sensor including a plurality of pixels for photographing a subject; And control means for controlling a photographing function of a camera according to a recognition result by the subject recognition means, wherein the subject recognition means recognizes subject information in accordance with a set shooting mode, and a recognition algorithm for recognizing subject information. A camera with a subject recognition function, wherein parameters constituting the recognition algorithm are changed. 3. The camera according to claim 1, wherein the recognition algorithm includes at least an algorithm for performing face recognition. 4. The camera with a subject recognition function according to claim 1, wherein the recognition algorithm includes at least an algorithm for detecting a motion of the subject. 5. A photographing mode setting means for setting a photographing mode of a camera, an area sensor comprising a plurality of pixels for photographing a subject, and at least a face among subject information to be photographed is recognized based on an output of the area sensor. A subject recognition unit having a face recognition algorithm to perform, and a control unit for controlling a photographing function of a camera in accordance with a recognition result by the subject recognition unit. A camera with a subject recognition function, wherein the operation of the face recognition algorithm is not performed. 6. The control unit controls the photographing function of the camera based on information corresponding to a preset photographing mode when the operation of the face recognition algorithm by the subject recognizing unit is not performed. The camera with a subject recognition function according to claim 5, wherein 7. A photographing mode in which the object recognizing means does not perform the operation of the face recognition algorithm is a photographing mode other than a photographing mode for photographing a scene mainly composed of a person. The camera with the subject recognition function described in. 8. The camera with a subject recognition function according to claim 1, wherein the control of the photographing function of the camera is exposure control and focus adjustment control. 9. The photographing mode setting means according to claim 1, wherein the photographer manually sets the photographing mode in accordance with the type or state of the subject.
9. The camera with a subject recognition function according to any one of items 1 to 8.