JP2010054883A - Vibration-proofing control device and imaging apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately correct parallel vibration of a photographic image due to translational vibration applied to an imaging apparatus that is a vibration-proofing object. <P>SOLUTION: The vibration-proofing control device has: an acceleration detection means 11p which detects the translational vibration; a shake correction means 95 which eases deterioration caused by image blur on a screen of the imaging apparatus; a movement detection means 93d which detects a movement component of the photographic image from an image signal of the imaging apparatus; translational vibration extraction means 93d, 58 and 59 which extract the translational vibration according to an output signal from the movement detection means; and calibration means 57 and 55 which calibrate a signal from the acceleration detection means based on an output signal from the translational vibration extraction means. The shake correction means is actuated based on the signal from the acceleration detection means calibrated by the calibration means. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image stabilization control apparatus and an imaging apparatus having a shake correction unit that corrects image shake.

  With the current camera, all the important tasks for shooting such as determining exposure and focusing are automated, and it is very unlikely that people who are unskilled in camera operation will fail to shoot. Recently, cameras equipped with a system that prevents image shake due to camera shake have been commercialized, and there are almost no factors that cause a photographer to take a picture.

  Here, a system for preventing image blur will be briefly described. The camera shake at the time of shooting is usually a vibration of 1 Hz to 10 Hz as a frequency. In order to be able to take a picture without image blur even when such a camera shake occurs at the shutter release time, the camera shake due to the camera shake is detected, and a correction optical system is set according to the detected value. The shake correction means provided must be displaced.

  Therefore, in order to take a photograph in which image shake does not occur even when camera shake occurs, first, it is necessary to accurately detect camera vibration, and secondly, to correct optical axis changes due to camera shake. Become. In principle, the vibration (camera shake) can be detected by detecting an acceleration, angular acceleration, angular velocity, angular displacement, and the like, and mounting a calculation unit that appropriately calculates the output on the camera. Based on this detection information, image blur correction is performed by driving a shake correction unit that decenters the photographic optical axis.

  FIG. 10A is a plan view of a single-lens reflex camera, and FIG. 10B is a side view thereof. The image stabilization system mounted on the interchangeable lens 90 attached to the single-lens reflex camera performs shake correction with respect to the camera vertical shake and horizontal shake indicated by arrows 92p and 92y with respect to the optical axis 91. Of the camera body 93, 93a is a release member, 93b is a mode dial (including a main switch), 93c is a retractable strobe, and 93d is a camera CPU.

  Reference numeral 94 denotes an image sensor, and 95 denotes a shake correction unit as a shake correction unit that drives a correction lens 95a, which is a correction optical system, freely in the directions of arrows 95p and 95y to perform shake correction in the directions of arrows 92p and 92y. . Reference numerals 96p and 96y denote angular velocity meters for detecting shakes around the arrows 92p and 92y, respectively, and arrows 96pa and 96ya indicate the respective detection sensitivity directions.

  The outputs of the angular velocity meters 96p and 96y are converted into the shake correction target values of the shake correction unit 95 by calculating the signals by the lens CPU 97. The drive target value controls the drive unit 98 in synchronization with a switch that is turned on when the release member 93a provided in the camera body 93 is half-pressed (hereinafter referred to as switch S1: an operation switch for instructing photometry and focusing for shooting preparation). Via the coil of the shake correction unit 95. Thereby, image blur correction starts.

  In the image stabilization system described with reference to FIG. 10, the angular velocity meters 96p and 96y are used for camera shake detection. The camera body 93 is not only subjected to angular shakes around the arrows 92p and 92y (around the photographing optical axis) but also parallel shakes indicated by arrows 11pb and 11yb (parallel shake within a plane perpendicular to the photographing optical axis). However, under general photographing conditions, the angular shake around the arrows 92p and 92y is dominant, and image degradation due to the parallel shake shown by the arrows 11pb and 11yb is small. For this reason, only the angular velocity meters 96p and 96y need be provided for camera shake detection.

  However, in close-up shooting (shooting conditions with a high shooting magnification), image degradation due to parallel shake indicated by arrows 11pb and 11yb cannot be ignored. For example, let us consider a condition in which a subject is photographed as close as 20 cm as in macro photographing, or a case where the focal length of the photographing optical system is very large (for example, 400 mm) even if the subject is located at about 1 m. In such a case, it is necessary to positively detect the parallel shake and drive the shake correction unit 95. Patent Document 1 discloses a technique in which an accelerometer that detects acceleration is provided, parallel shake is detected by the accelerometer, and the shake correction unit 95 is driven together with the output of an angular velocity meter provided separately.

  Here, there is a possibility that the signal of the accelerometer used for detecting the parallel shake may change due to an environmental change such as temperature. Therefore, it is difficult to correct the parallel shake with high accuracy. Further, in order to correct the parallel shake, it is necessary to detect not only the accelerometer but also the subject distance (or subject magnification) and determine the amount of correction of the parallel shake according to the signal. This is because the influence of image deterioration due to parallel shake appears when the subject is close as described above. However, it is difficult to detect the distance of the subject with high accuracy, and this also causes deterioration of the parallel shake correction accuracy.

  Japanese Patent Application Laid-Open No. 2004-228620 discloses that the motion information of a captured image is captured by an image sensor mounted on a camera, thereby calibrating the angular velocity meter. However, in Patent Document 2, the angular velocity meter is simply calibrated based on the motion information of the image sensor, and the accelerometer cannot be calibrated. This is because the motion information includes angular shake and parallel shake and cannot be separated.

In addition, as shown in Patent Document 3, there has been proposed that the angular velocity meter is not calibrated based on motion information under imaging conditions in which parallel shake occurs.
JP 7-225405 A JP 2002-359769 A JP 2001-356380 A

  As described above, conventionally, the accelerometer cannot be calibrated, and the parallel shake cannot be corrected with high accuracy.

(Object of invention)
SUMMARY OF THE INVENTION An object of the present invention is to provide an image stabilization control device and an image pickup apparatus that can highly accurately correct a parallel shake of a captured image caused by translational vibration applied to an image pickup apparatus subject to image stabilization.

  In order to achieve the above object, the present invention provides an acceleration detection means for detecting translational vibration applied to an image pickup apparatus subject to image stabilization, a shake correction means for reducing deterioration due to image shake on the screen of the image pickup apparatus, Based on the motion detection means for detecting the motion component of the captured image from the image signal of the imaging device, the translational vibration extraction means for extracting the translational vibration by the output signal of the motion detection means, and the output signal of the translational vibration extraction means And a calibration unit that calibrates the signal from the acceleration detection unit, and the image stabilization control device operates the shake correction unit based on the signal from the acceleration detection unit calibrated by the calibration unit. .

  Similarly, in order to achieve the above object, the present invention is an imaging apparatus including the image stabilization control apparatus of the present invention.

  ADVANTAGE OF THE INVENTION According to this invention, the anti-shake control apparatus or imaging device which can correct | amend the parallel shake of the picked-up image by the translational vibration added to the imaging device of anti-shake object with high precision can be provided.

  The best mode for carrying out the present invention is as shown in Examples 1 to 3 below.

  1A and 1B are a top view and a side view of a digital camera which is an example of an imaging apparatus according to Embodiment 1 of the present invention. 10 (a) and 10 (b) is different from the conventional example in that accelerometers 11p and 11y are provided, and the acceleration detection axes of the accelerometers 11p and 11y are arrows 11pa and 11ya.

  FIG. 2 shows the translational vibration detected by the parallel shake detectors (hereinafter referred to as accelerometers) 11p and 11y and the angular vibration of the rotational vibration detected by the angular shake detectors (hereinafter referred to as angular velocity meters) 96p and 96y. It is a block diagram which processes a signal. These detectors are mainly processed in the lens CPU 97. In FIG. 2, only signal processing for correcting image blur due to vertical camera shake (angle shake 92p and parallel shake 11pb in FIG. 1B) is shown. However, in actuality, the image blur correction due to the lateral shake of the camera (the angular shake 92y and the parallel shake 11yb in FIG. 1A) is also subjected to signal processing.

  In FIG. 2, the angular velocity signal of the angular velocity meter 96p is input to the amplifying unit 12p. The amplification unit 12p not only simply amplifies the output of the angular velocity meter 96p, but also includes a DC removal circuit that removes a DC component superimposed on the angular velocity meter 96p and a high-frequency attenuation circuit that removes a high-frequency noise component. The output of the amplifying unit 12p is A / D converted and taken into the lens CPU 97. The captured signal is numerically processed in the lens CPU 97, but here, the processing is divided into blocks for illustration.

  An angular shake calculation unit (hereinafter referred to as an angular velocity integration unit) 13p integrates the camera shake angular velocity signal input from the amplification unit 12p into the first order and converts it into an angular shake correction target value (hereinafter referred to as an angular shake angle). The angular velocity integrating unit 13p integrates a high frequency component of usually about 0.1 Hz or more in the camera shake angular velocity signal and converts it into an angular shake angle. However, at the time of starting the angular velocity integration, the integration band is narrowed (for example, a signal of 2 Hz or less is attenuated) to start signal processing earlier (this is called time constant switching). The obtained angular deflection angle signal is input to the adding unit 14p, added to a parallel deflection displacement signal described later, and converted into a total deflection signal. The adder 14p adds an angular shake angle signal and a parallel shake displacement signal described later based on signals from the release member 93a and the focus detection unit 27. The adder 14p receives the angular shake angle signal in response to the switch S1 that is turned on by half-pressing the release member 93a, and the parallel shake displacement signal is shaken based on the signal input (focusing is completed) of the focus detection unit 27. Add to the angle signal.

  The total shake signal is input to the frequency characteristic changing unit 15p, and the frequency characteristic is changed. The frequency characteristic changing unit 15p mainly attenuates the low frequency component of the shake total signal, determines which frequency (for example, 0.1 Hz or 5 Hz) is to be attenuated, and attenuates the signal component. This is to prevent the shake correction from being performed by increasing the attenuation of the shake total signal (for example, a signal of 5 Hz or less is attenuated) when a large change in shake occurs, such as when changing the framing of the camera. Without this frequency characteristic changing unit 15p, even the framing component of the camera is corrected for shake, so that it is impossible to perform good framing of the camera.

  The output of the frequency characteristic changing unit 15p is input to the sensitivity changing unit 16p. The sensitivity changing unit 16p changes the amplification factor of the signal of the frequency characteristic changing unit 15p in response to the signals of the focal length detecting unit 18 and the photographing distance detecting unit 19 input to the lens CPU 97.

  In general, the shake correction sensitivity of the optical system of the shake correction unit 95 included in the zoom lens varies depending on the zoom state and the focus state. For example, it is assumed that when the shake correction unit 95 is driven 1 mm in zoom wide, the 1 mm image shifts on the image plane. In this case, when the shake correction unit 95 is driven by 1 mm with zoom tele, the 3 mm image is shifted on the image plane. Similarly, the relationship between the amount of drive of the shake correction unit 95 and the amount of image shake changes when the subject is close and infinite. Therefore, in order to correct the sensitivity (for example, in the case of tele, the amplification factor is reduced to 1/3), the amplification factor of the signal of the frequency characteristic changing unit 15p is changed by zoom or focus. The focal length detection unit 18 is an encoder or the like that is provided in the lens and detects the position of the lens that moves by zooming. The focal length is detected by the output of the encoder. An imaging distance detector 19 is also provided in the lens, and is an encoder or the like that detects the position of the lens that is moved by focusing. The imaging distance is detected by the output of the encoder.

  In this way, the sensitivity changer 16p adds the angle shake angle signal (angle shake correction target value) and the parallel shake displacement signal (parallel shake correction target value), and performs the frequency and gain processing to obtain a total shake correction target value. Is output.

  When the switch S1 is turned on by half-pressing the release member 93a, the total shake correction target signal from the sensitivity changing unit 16p is converted into a PWM signal and input to the shake correction drive unit 98p. The shake correction drive unit 98p drives the shake correction unit 95 according to the input PWM signal, and performs image shake correction.

  When the switch S1 is turned on, only the angle shake angle signal is input to the adder 14p, and therefore only the angle shake correction is performed. In response to the switch S1 being turned on, the focus detection unit 27 in the camera CPU 93d drives the focus sensor 32 in the camera main body 93 to detect the focus state of the photographic subject. Then, based on the detection result of the focus sensor 32, the focus detection unit 27 sends a focus shift amount to the lens drive calculation unit 33 in the lens microcomputer 97. The lens drive calculation unit 33 drives the focus drive unit 34 based on the signal to move the focus lens 99. Here, since the shake of the angle is corrected even during the focusing operation described above, a highly accurate focusing operation is possible.

  After the focus lens 99 is driven, the focus sensor 32 detects the focus state of the subject again, and displays an in-focus display when the focus is sufficient (if the focus is insufficient, the lens is moved again). When a sufficient focus state is achieved, the shooting distance detector 19 causes the adder 14p to add the parallel shake displacement signal to the angle shake angle signal via the shooting distance corresponding unit 36p. The payout amount of the focus lens 99 is constantly input to the sensitivity changing unit 16p. The sensitivity change unit 16p does not supply the focus lens 99 when the focus detection unit 27 is in focus (of the shooting distance detection unit 19). Signal) is the anti-vibration sensitivity value.

  Here, the shooting distance corresponding unit 36 p receives the output of the shooting distance detecting unit 19. Then, whether or not the signal (subject distance) adds the parallel shake displacement signal to the angular shake angle signal to correct the shake (adds the parallel shake displacement when the subject distance is close), or the initial value of the parallel shake target value to be described later The initialization unit 35p outputs a signal indicating whether or not initialization is necessary.

  As will be described later, the photographing magnification is calculated from the relationship between the feed amount of the focus lens 99 and the position of the zoom lens. However, the imaging magnification is also calculated based on signals from the focal length detection unit 18 and the imaging distance detection unit 19 with the focus detection of the focus detection unit 27 as a trigger. In other words, when the zoom is determined, the subject is in focus and the amount of extension of the focus lens 99 is known, the sensitivity for shake correction is known, and the total shake correction target value is calculated. It is assumed that the zoom is determined before the photographer performs a half-press operation on the release member 93a. The image magnification is also obtained when the subject is in focus.

  The total shake correction target signal obtained by adding the angle shake angle signal and the parallel shake displacement signal obtained as described above is converted into a PWM signal and input to the shake correction drive unit 98p. The shake correction drive unit 98p drives the shake correction unit 95 according to the input PWM signal, and performs image shake correction. In other words, the parallel shake correction is also performed when the focusing is completed.

  Next, signal processing of the accelerometer 11p will be described. The acceleration signal of the accelerometer 11p is input to the amplifying unit 20p. The amplifying unit 20p not only simply amplifies the output of the accelerometer 11p, but also includes a DC removal circuit that removes a DC component superimposed on the accelerometer 11p and a high-frequency attenuation circuit that removes a high-frequency noise component. The output of the amplifying unit 20p is A / D converted and taken into the lens CPU 97. Although the captured acceleration signal is also numerically processed in the lens CPU 97, the processing is divided into blocks for the sake of explanation.

  First, the acceleration signal is input to the acceleration gravity correction unit 21p, and the gravity component is corrected. Here, the meaning of the correction of the gravity component will be described. Since the camera is horizontal at the photographing position of the camera shown in FIG. 1B, the sensitivity direction 11pa of the accelerometer 11p is the same as the direction of gravity 28 (see FIG. 3A). At this time, the accelerometer 11p always outputs a signal commensurate with the gravitational component, and detects the parallel shake component by superimposing it. Here, since the signal output of the gravity component is a DC component, it can be removed by a DC removal circuit or the like in the amplification unit 20p.

  However, since the position of the accelerometer 11p changes as shown by the broken line in FIG. 3A due to the change in the rotation angle of the hand shake that occurs when the camera is held, the direction of gravity when viewed from the accelerometer 11p. Will change. Therefore, the output of the accelerometer 11p changes due to a change in the hand shake angle.

  FIG. 3C shows changes in the output of the accelerometer 11p with respect to the attitude of the accelerometer 11p. The horizontal axis represents the change in attitude of the accelerometer 11p (camera shake rotation angle θ), and the vertical axis represents the output of the accelerometer 11p. . A waveform 30p indicates the output of the accelerometer 11p. When the attitude angle of the accelerometer 11p changes from zero (a state where 1G is added as shown in FIG. 3A) to ± θ attitude, the output of the accelerometer 11p is output. Change (decrease) with it.

  4A and 4B show the output due to the gravity change of the accelerometer 11p, the horizontal axis represents the elapsed time since holding the camera, and the vertical axis represents the shake rotation angle θ and the output of the accelerometer. Yes.

  Even if it is assumed that no parallel shake has occurred at this time, the accelerometer 11p outputs an error signal 30p due to the influence of the gravity component change caused by the hand shake rotation angle 29p. In close-up shooting, the camera is often pointed down. FIG. 3B is an example of such a case, and the gravity direction 28 is orthogonal to the sensitivity direction 11pa of the accelerometer 11p. In this case, the error signal is a broken line 31p in FIGS. 3 (c) and 4 (b).

  Here, there is a difference in the magnitudes of the error signals 30p and 31p between the arrangement of the accelerometer 11p in FIG. 3A and the arrangement of the accelerometer 11p in FIG. This is because, in the arrangement of FIG. 3 (a), the influence of gravity is caused by the cosine with respect to the change of the camera shake rotation angle, and in the arrangement of FIG. 3 (b), the influence of the gravity is caused by the sign. This is because the sign changes more greatly when it is small. Therefore, in order to correct the influence of the gravity, the rotation angle of the camera shake is detected, and the attitude of the accelerometer 11p (the sensitivity axis with respect to the gravity as shown in the difference between FIGS. 3A and 3B). It is necessary to know which angle is).

  Returning to FIG. 2, the lens CPU 97 receives an ON signal of the switch S1 accompanying the half-pressing operation from the release member 93a via the camera CPU 93d. The half-pressing of the release member 93a is an operation after determining the composition with the camera facing the shooting subject in preparation for shooting, thereby starting photometric and focusing operations on the subject. In FIG. 2, the above operation is not directly related to the invention and is omitted, and the ON signal of the switch S1 is input to the initial posture direction detector 23p. An acceleration amplification signal from the amplification unit 20p is also input to the initial posture direction detection unit 23p, and the posture of the accelerometer 11p is determined based on the magnitude of the acceleration amplification signal when the ON signal of the switch S1 is input. .

  The half-pressing of the release member 93a is performed after the photographer determines the composition, so that there is no significant change in posture thereafter. Therefore, it is effective to determine the attitude of the accelerometer 11p based on the ON signal of the switch S1 accompanying the operation. Of course, after the switch S1 is turned on, the posture may be detected after the camera has focused on the subject. In this case, the time from the switch S1 to the focus is adjusted using the time of the accelerometer 11p. The output cannot be integrated (described later). In order to save time, it is desirable to detect the attitude of the accelerometer 11p when the switch S1 is turned on. Specifically, the attitude determination of the accelerometer 11p is as shown in FIG. 3A when the magnitude of the acceleration is 1G when the switch S1 is turned on, and as shown in FIG. 3B when the acceleration is 11G. At the time of posture and acceleration during that time, it is determined as a posture corresponding to it.

  The angular deflection angle signal from the angular velocity integrating unit 13p is input not only to the adding unit 14p described above but also to the gravity influence calculating unit 25p. The gravitational effect calculation unit 24p performs a calculation for determining a change in gravity applied to the accelerometer 11p based on the input change in angular deflection angle. However, as described above, the calculation method (whether the calculation is performed by sine or cosine) differs depending on the attitude of the accelerometer 11p with respect to gravity. Therefore, the signal of the initial posture direction detection unit 23p is also input to the gravity influence calculation unit 24p, and the calculation coefficient is changed between the posture in FIG. 3A and the posture in FIG. Specifically, as shown in FIG. 3A, when the posture φ when 1G is applied is zero degrees, and the change in posture is θ (which is also set to θ because it is substantially equal to the camera shake rotation angle θ), the acceleration The output change of the total 11p is obtained by G (COSφ−COS (φ + θ)). Therefore, φ is obtained by the initial posture direction detector 23p, θ is obtained by the hand shake angle, and is used for calculating the gravitational effect.

  The acceleration amplification signal from the amplifying unit 20p is input to the acceleration gravity correcting unit 21p, and a difference from the signal change of the accelerometer 11p accompanying the gravity change obtained by the gravity influence calculating unit 24p is calculated. Then, the output error of the accelerometer 11p due to the influence of gravity is removed. The acceleration output from which the error component has been removed is input to a parallel shake calculation unit (hereinafter referred to as an acceleration integration unit) 22p. The acceleration integration unit 22p performs second-order integration of the acceleration signal corrected for the influence of gravity input from the acceleration gravity correction unit 21p and converts it into a parallel shake correction target value (hereinafter referred to as parallel shake displacement). Similar to the angular velocity integrating unit 13p, the acceleration integrating unit 22p integrates a high-frequency component of about 0.4 Hz or higher in the acceleration signal into the second-order displacement and converts it into a parallel shake displacement. The acceleration integration unit 22p narrows the integration band (for example, integrates only components of 1 Hz or more) at the time of starting the integration of the acceleration signal, thereby speeding up the signal processing (time constant switching).

  The parallel shake displacement signal of the acceleration integration unit 22p is input to the image magnification correction unit 25p. The photographing magnification calculator 26p calculates the photographing magnification based on the focal length information from the focal length detector 18 and the photographing distance information from the photographing distance detector 19. As described above, the focal length detection unit 18 is an encoder or the like that is provided in the lens and detects the position of the lens that is moved by zooming. The focal length is detected by the output of the encoder. An imaging distance detector 19 is also provided in the lens. The photographing distance detection unit 19 is an encoder or the like that detects the position of a lens that moves by focusing, and detects the photographing distance based on the output of the encoder.

  As described above, the focus drive unit 34 extends the focus lens 99, and after completion of the extension, when the focus detection unit 27 confirms the in-focus state, the imaging magnification calculator 26p detects the focal length detection unit 18 and the imaging distance detection. The photographing magnification is calculated based on the output of the unit 19. Parallel shakes 11pb and 11yb (see FIG. 1) have a large effect on the screen when the subject is close and the photographing focal length is large (when the photographing magnification is high), and when the subject is far (when the photographing magnification is low). There is almost no effect on the screen. Therefore, it is necessary to amplify the parallel shake displacement detected and calculated by the accelerometers 11p and 11y in accordance with the photographing magnification.

  The image magnification correction unit 25p amplifies the parallel shake displacement of the acceleration integration unit 22p based on the calculation value of the shooting magnification calculation unit 26p (calculates that the shooting magnification is high when the focal distance is long and the subject distance is close). Yes. However, the photographing magnification obtained by extending the focus lens 99 is not very accurate. This is because the imaging magnification calculation unit 26p calculates the imaging magnification based on the outputs of the focal length detection unit 18 and the imaging distance detection unit 19, but outputs of the focus lens 99 and the encoder that detects the position of the zoom lens. This is because the accuracy is not so high.

  Therefore, at this time, the translational shake displacement is only amplified within a predetermined accuracy, and finally the accelerometer output is amplified with high accuracy by the calibration method described later.

  The output of the image magnification correction unit 25p is input to the parallel shake correction target value initialization unit 35p. The parallel shake correction target value initialization unit 35p initializes the parallel shake correction target value in synchronism when an ON signal of the switch S2 is generated by fully pressing the release member 93a, that is, when an instruction to start photographing is issued. In other words, the output of the parallel shake correction target value initialization unit 35p becomes zero when the release member 93a is fully pressed, and thereafter the output starts continuously. Alternatively, the parallel shake correction target value initialization unit 35p limits the parallel shake correction target value to a predetermined value (for example, 90% of the full range of the parallel shake correction target value) when the release member 93a is fully pressed, and then outputs continuously. You may let them.

  The parallel shake correction target value initialization unit 35p also receives a signal from the shooting distance corresponding unit 36p, and initializes the parallel shake correction target value only when the shooting distance is closer to the set value (for example, the shooting distance is 30 cm). The unit 35p is operated. This operation may be controlled not only by the subject distance but also by the photographing magnification, and the parallel shake correction target value initialization unit 35p may be operated only when the photographing magnification becomes equal to or larger than the same magnification.

  The adding unit 14p adds the signal of the angular velocity integrating unit 13p and the signal of the parallel shake correction target value initializing unit 35p. As described above, the output of the angular velocity integrating unit 13p is almost output when the subject is far and the photographing focal length is short. Become only. The operations after the adding unit 14p are as described above, and the frequency characteristic changing unit 15p for facilitating camera framing change, and the sensitivity changing unit 16p for adjusting the effect of image blur correction in accordance with the sensitivity of the optical system. The total shake correction target value is obtained via Then, the shake correction unit 95 is driven.

  FIG. 5 is a flowchart showing the operation of the main part in the above configuration, and this flow starts when the main power supply of the camera is turned on. In order to explain the main configuration of the present invention in an easy-to-understand manner, various control steps (for example, battery check, photometry, distance measurement, lens driving for AF, strobe charging, and photographing are provided in the camera. Operation, movement, etc.) are omitted. Further, in the following flow, the explanation will be made with an example in which the angular shake 92p and the parallel shake 11pb of the camera are detected by the angular velocity meter 96p and the accelerometer 11p. However, the same flow is performed when the angular shake 92y and the parallel shake 11yb of the camera are detected by the angular velocity meter 96y and the accelerometer 11y.

  In FIG. 5, in step # 1001, it waits for switch S1 to turn on by half-pressing operation of release member 93a, and when switch S1 is turned on, it proceeds to step # 1002. In Step # 1002, the initial posture direction detector 23p detects the posture of the camera from the signal from the accelerometer 11p. This detects the gravitational acceleration applied to the accelerometers 11p and 11y. For example, when the camera is held horizontally as shown in FIGS. 1A and 1B, the accelerometer 11p outputs 1G. The accelerometer 11y outputs 0G. In this state, if the camera is held vertically (in the horizontal state but the composition is vertical), the accelerometer 11p outputs 0G and the accelerometer 11y outputs 1G. Further, when the camera is pointed downward or upward, both accelerometers 11p and 11y output 0G.

  The reason why the posture is detected when the release member 93a is half-pressed is that the photographer decides the framing by holding the camera and performs the half-pressing operation of the release member 93a after stabilization, so that the posture change is small thereafter. It is. When the posture of FIG. 1A is determined by the signals from the accelerometers 11p and 11y, gravity correction is performed on the output of the accelerometer 11p. However, the gravitational effect calculation unit 24y determines that the output of the accelerometer 11y is not subjected to gravity correction, and the correction amount of the acceleration gravity correction unit 21y is set to zero. This is because there is no change in gravitational acceleration due to angular deflection. That is, the acceleration gravity correction unit 21y (not shown, but provided with the same configuration as the acceleration gravity correction unit 21p and provided to correct the gravitational effect of the accelerometer 11y) is applied to the amplified signal of the accelerometer 11y. The gravity component is not corrected.

  On the other hand, when the camera is held vertically (accelerometer 11p = 0G, accelerometer 11y = 1G), gravity correction of the accelerometer 11y is performed based on the signal from the angular velocity meter 96y. However, gravity correction of the accelerometer 11p based on the signal of the angular velocity meter 96p is not performed. That is, the gravity influence calculation unit 21p sets the correction amount of the acceleration gravity correction unit 21p to zero. When the camera is held downward or upward (accelerometer 11p = ± 1, accelerometer 11y = ± 1G), gravity correction of the accelerometer 11p is performed based on the signal of the angular velocity meter 96p, and the angular velocity meter 96y Based on the signal, gravity correction of the accelerometer 11y is performed.

  As described above, it is determined whether or not to perform gravity correction according to the posture. Note that not only gravitational acceleration but also acceleration due to parallel shake is superimposed on the signals of the accelerometers 11p and 11y. For this reason, the signals from the accelerometers 11p and 11y are averaged for a predetermined time (for example, 1 second) to extract only the gravity component.

  When posture detection is completed as described above, the process proceeds to step # 1003, and sensitivity correction according to the lens state and frequency correction according to the shake state (panning, etc.) are performed on the camera shake angle signal. Then, in the next step # 1004, angle shake correction based on the angle shake angle signal is started. In subsequent step # 1005, the gravitational effect calculation unit 24p calculates the gravitational acceleration superimposed on the accelerometer 11p based on the camera posture by the initial posture direction detection unit 23p and the angular shake angle information from the angular velocity integration unit 13p, and the acceleration The gravity correction unit 21p corrects the error output. That is, at this time, the accelerometer 11p is driven to start detecting parallel shake. However, parallel shake correction is not performed at this point.

  In the next step # 1006, the focusing operation is started. Then, in the next step # 1007, the process waits until the lens extension for focusing is completed. That is, the focus detection unit 27 detects the focus state of the subject by the focus sensor 32, and the lens drive calculation unit 33 calculates the focus lens feed amount. Thereafter, after the focus lens 99 is driven by the focus driving unit 34, the focus sensor 32 waits again until it confirms that the subject is in focus. Further, when the extension of the focus lens 99 is completed, the amount of extension of the focus lens 99 is read by the shooting distance detector 19 to detect the shooting distance (subject distance).

  In the next step # 1008, output calibration of the accelerometers 11p and 11y is performed. Details of this step will be described later with reference to FIG. In the subsequent step # 1009, it is determined whether or not the switch 2 is turned on by the full pressing operation of the release member 93a. If the switch 2 is turned on, the process proceeds to step # 1010. If not, the process proceeds to step # 1013. Proceed to

  First, the case where the switch 2 is turned on and the process proceeds to step # 1010 will be described. In step # 1010, in addition to the shake correction unit 95 performing angle shake correction in step # 1004, the shake correction unit 95 is driven based on signals from the accelerometers 11p and 11y. In other words, total shake correction is performed so as to mitigate image plane deterioration due to angular shake and parallel shake. In the next step # 1011, photographing is performed, and when photographing is completed, the process proceeds to step # 1012, where it is determined whether or not the release member 93a is fully pressed and the switch S2 is turned off. If the switch S2 is not turned off, the process waits in step # 1012 until the switch S2 is turned off. After that, when the switch S2 is turned off, the process proceeds to step # 1013.

  When the switch S2 is turned off in step # 1009 or step # 1012, the process proceeds to step # 1013 as described above, and here the state of the switch S1 is detected. If the switch S1 is also off, the process proceeds to step # 1014 to stop driving the shake correction unit 95, that is, stop the shake correction operation, return to step # 1001, and wait for the switch S1 to be turned on again. . If the switch S1 remains on in step # 1013, the process returns to step # 1006 and the same operation is continued.

  As described above, in the first embodiment, only the angular shake is corrected first, and the parallel shake is corrected after focusing. As described above, since the angle shake is corrected before the focusing operation, it is possible to improve the focusing accuracy and accurately obtain information (shooting magnification) for performing the parallel shake.

  Hereinafter, features of the first embodiment of the present invention will be described.

  In FIG. 1B, a signal from the camera CPU 93d is input to the lens CPU 97 via a mount contact that detachably holds the camera body 93 and the lens barrel 90. In FIG. 1B, for easy understanding, an arrow is directly shown from the camera CPU 93d to the lens CPU 97. Image information of the image sensor 94 is input to the camera CPU 93d, and the camera CPU 93d detects a motion component in the image from a plurality of pieces of image information provided with a time difference, and sends the signal to the lens CPU 97.

  Here, the motion components detected by the signal from the image sensor 94 include image motion due to camera angular shake, image motion due to parallel shake, and image motion due to subject movement. . However, in general, in still image shooting, since the subject is stationary, the motion component detected by the image sensor 94 is mainly composed of image motion due to camera angular shake and image motion due to parallel shake.

  In the first embodiment, the output calibration of the accelerometer 11p is performed using the motion component.

  FIG. 6 is a block diagram showing a circuit configuration related to output calibration of the accelerometer 11p. The angular velocity signal from the angular velocity meter 96p is converted into an angular shake angle by the integrator 51 and input to the shake correction drive unit 98p. The shake correction drive unit 98p drives the shake correction unit 95 to perform shake correction. As a result, the angular shake is corrected.

  On the other hand, the signal from the accelerometer 11p is converted into a parallel shake velocity by the integrator 54, and the signal is added to the signal from the angular velocity meter 96p and then integrated into the parallel shake displacement by the integrator 51 and input to the shake correction drive unit 98p. Is done. Then, the shake correction driving unit 98p drives the shake correction unit 95 to perform shake correction. As a result, the parallel shake is also corrected.

  Further, as described above, the output of the angular velocity meter 96p is converted into a parallel shake angle by the integrator 52 in order to remove the gravitational effect from the output of the accelerometer 11p. The initial posture gain setting unit 53 sets the amount of gravity removed according to the initial posture of the accelerometer 11p, and subtracts the gravity component from the signal from the accelerometer 11p.

  Here, a parallel vibration velocity signal, which is an output through the integrator 54 of the accelerometer 11p, is extracted by a bandpass filter (BPF) 56 in a predetermined frequency band (for example, around 2 Hz) in the vibration frequency band. The band pass filter 56 is configured to extract only a signal in the vicinity of 2 Hz, for example, and extracts only this frequency band from the parallel shake speed signal. The velocity signal in this frequency band is a large and frequent signal in parallel shakes, and is suitable for comparison with the actual motion velocity.

  On the other hand, a motion component is extracted from the image signal from the image sensor 94 by the camera CPU 93d as described above. Since this motion component becomes a displacement signal, it is converted into a motion speed by the differentiator 58, and only a signal in the vicinity of 2 Hz is extracted by the band pass filter 59 (the same characteristic as the band pass filter 56). The comparator 57 compares the output signal of the bandpass filter 56 with the output signal of the bandpass filter 59. At this time, the gain of the parallel shake velocity signal that is the output of the integrator 54 is set to the image magnification gain so that the output signal (amplitude) of the bandpass filter 56 and the output signal (amplitude) of the bandpass filter 59 have the same amount. Adjustment is performed by the unit 55.

  For example, the comparator 57 compares the difference between the maximum value and the minimum value of the output signal (amplitude) of the bandpass filter 56 and the output signal (amplitude) of the bandpass filter 59 in a predetermined period. Then, the image magnification gain setting is performed so that the difference between the maximum value and the minimum value in the output signal (amplitude) of the bandpass filter 56 is the same as the difference between the maximum value and the minimum value of the output signal (amplitude) of the bandpass filter 59. The output gain of the integrator 54 is adjusted by the unit 55. Alternatively, the rectified value (average value of absolute values) of the output signal (amplitude) of the bandpass filter 56 and the rectified value (average value of absolute values) of the output signal (amplitude) of the bandpass filter 59 are the same. The output gain of the integrator 54 is adjusted by the image magnification gain setting unit 55.

  As a result of the above adjustment, the output gain of the integrator 54 is adjusted, the output amplitude of the bandpass filter 56 gradually approaches the output amplitude of the bandpass filter 59, and finally the mutual amplitude becomes the same amount, thereby completing the gain adjustment. To do.

  Thereby, the gain of the parallel shake is adjusted by the motion component detected by the image sensor 94, and the parallel shake on the image plane is detected with high accuracy.

  In the processing so far, both the differential value of the motion signal (component) in the image sensor 94 and the integration of the acceleration signal are compared as a speed signal. You may compare the displacement of a component. However, the second order integral value of the acceleration signal is not preferable because the period until the output is stabilized becomes long and the low frequency noise of the accelerometer 11p is amplified.

  Further, the motion signal in the image sensor 94 may be second-order differentiated and converted into motion acceleration, and the signal may be compared with the signal of the accelerometer 11p. It becomes impossible to accurately compare the signals.

  Considering the above problem, the differential value of the motion signal of the image sensor 94 is compared with the integral value of the accelerometer 11p. After the gain adjustment is completed, a parallel shake velocity signal that is an output via the integrator 54 of the accelerometer 11p is extracted as a specific signal by a low-pass filter (LPF) 510. The low-pass filter 510 is configured to extract only a frequency signal of 0.4 Hz or less, for example, and only this frequency band is extracted from the parallel shake speed signal. The velocity signal in this frequency band is not actually a parallel shake but a signal offset component superimposed on the accelerometer 11p.

  As described above, the motion signal is extracted from the image signal of the image sensor 94 by the camera CPU 93d. Since this movement becomes a displacement signal, it is converted into a movement speed by the differentiator 58, and only a frequency signal of 0.4 Hz or less is extracted by the low-pass filter 512 (the same characteristic as the low-pass filter 510). The comparator 511 compares the output signal of the low pass filter 510 with the output signal of the low pass filter 512. At this time, the offset component is subtracted from the speed signal that is the output of the integrator 54 so that the output signal (offset) of the low-pass filter 510 and the output signal of the low-pass filter 512 have the same amount.

  The offset component of the integrator 54 is adjusted by the above adjustment, and the offset of the low-pass filter 510 gradually approaches the DC component of the low-pass filter 512. Finally, the mutual DC components become the same amount and the offset adjustment ends. Thereby, the offset component superimposed on the accelerometer 11p is removed, and the parallel shake on the image plane is detected with high accuracy.

  Here, as described above, the motion detected by the image sensor 94 includes image motion due to camera angular shake and image motion due to parallel shake. If only the component of parallel shake is not extracted from these motions, the motion is accelerated. High-precision output calibration of a total of 11p is not possible.

  A technique for extracting only the component of parallel shake will be described below.

  The output calibration of the accelerometer 11p is performed in step # 1008 in FIG. 5, and detailed processing in this step will be described with reference to the flowchart in FIG.

  In FIG. 7, in step # 2001, the positions of the zoom lens and the focus lens 99 are read by the zoom encoder (focal length detection unit 18) and the focus encoder (shooting distance detection unit 19) to obtain the focal length and the subject distance. Approximate the shooting magnification. In the next step # 2002, the output of the accelerometer 11p is amplified from the obtained photographing magnification. The output of the accelerometer 11p is amplified by changing the gain of the integrator 54 in the image magnification gain setting unit 55 in FIG. Note that the accuracy of the amplification gain obtained here is not so high (depending on the accuracy of each encoder), so the accuracy of the amplification gain is not high.

  In addition, since the amplification gain of the accelerometer 11p is finally set by the motion signal of the image sensor 94, it may not be necessary to set the gain of the accelerometer 11p at this point. However, if the shooting magnification is very small and the parallel shake is not affected by the image plane, there is no need for calibration, so that the shooting operation is prioritized without calibration and the parallel shake affects the image plane. Even in this case, the amplification gain of the accelerometer 11p is roughly determined. Thus, in order to speed up the subsequent calibration, the photographing magnification information is obtained and the amplification gain of the accelerometer 11p is set.

  In the next step # 2003, motion detection is performed from the output of the image sensor 94, and after differentiation and conversion into motion speed, only a signal in the vicinity of 2 Hz is extracted by the band pass filter 59. Here, since the shake of the angle has already been corrected in step # 1005 of FIG. 5, the shake remaining on the imaging surface of the image sensor 94 is a very low frequency that could not be removed even during the motion and the shake of the shake. This is the movement caused by the movement of Therefore, if the differentiated motion signal is passed through the band-pass filter 59 of FIG. 6, the very low frequency shake is eliminated and only the motion component due to the parallel shake is extracted. That is, when the motion component is acquired from the output of the image sensor 94, the angular shake is already corrected, and the obtained motion component is subjected to the band-pass filter 59, thereby extracting the parallel shake (translation). Vibration extraction).

  In the next step # 2004, the speed signal obtained from the accelerometer 11p (only the signal in the vicinity of 2 Hz is extracted by the band pass filter 56) and the signal obtained in the above step # 2003 are compared. Then, the gain of the accelerometer 11p is adjusted so that the velocity signal obtained from the accelerometer 11p (only the signal near 2 Hz is extracted by the bandpass filter 56) is the same as the amplitude of the signal obtained in step # 2003. . That is, adjustment is performed by changing the gain of the integrator 54 in the image magnification gain setting unit 55 in FIG.

  In the next step # 2005, the comparison result in step # 2004 (the amplitude ratio of the bandpass filter processing signal of the integral value of the accelerometer 11p and the amplitude of the bandpass filter processing signal of the motion information differential value obtained by the image sensor 94) is obtained. Ask whether it is within the prescribed range. As a result, if the ratio is within a predetermined range, for example, 0.8 times to 1.2 times, the process proceeds to step # 2006. If not, the process returns to step # 2004, and comparison and amplification adjustment are performed again. In this manner, the amplification gain of the integral value of the accelerometer signal is driven.

  In the next step # 2006, the velocity signal obtained from the accelerometer 11p (only the signal around 0.4 Hz is extracted by the low-pass filter 510) and the signal obtained in step # 2003 (the signal around 0.4 Hz by the low-pass filter 512). Only take out). Then, the DC component of the accelerometer 11p is adjusted so that the velocity signal obtained from the accelerometer 11p (only the signal near 0.4 Hz is extracted by the low-pass filter 510) is the same as the signal obtained in step # 2003. . At this time, since the motion information of the image sensor 94 passes through the low-pass filter 512 after differentiation, only the DC component is extracted from the image motion caused by the parallel shake.

  Then, the accelerometer 11p is integrated and converted into a speed signal, and then the signal that has passed through the low-pass filter 510 and the motion signal obtained from the image sensor 94 are differentiated and converted into motion speed, and then the signal that has passed through the low-pass filter 512. Find the difference. Thereby, the DC component superimposed on the accelerometer integral value is extracted. Then, the extracted DC component is subtracted from the accelerometer integral value. In practice, the motion signal is differentiated and converted into a motion speed, and the signal that has passed through the low-pass filter 512 includes a motion component due to the influence of the extremely low frequency angular shake that is the remainder of the angular shake correction. The DC component superimposed on the accelerometer integral value is larger than the error. Therefore, the effect of removing the DC component superimposed on the accelerometer integral value by this operation is great.

  In the next step # 2007, whether or not the DC component remaining in step # 2006 (the difference between the low-pass filtered signal of the integral value of the accelerometer 11p and the low-pass filtered signal of the motion information differential value obtained from the image sensor 94) is within a predetermined range. Ask for. If the difference is within a predetermined range, for example, 0.1 mm / s to 0.3 mm / s, the process proceeds to step # 1009 in FIG. 5; otherwise, the process returns to step # 2006, and the comparison is performed again. Make adjustments. In this way, the DC component of the integrated value (also referred to as the accelerometer integrated value) of the output signal of the accelerometer is driven.

  As described above, the gain of the translational shake speed signal, which is the accelerometer integral value, is adjusted and the DC offset component is removed.

  In this way, even if the detection accuracy of the shooting magnification is low, the correction gain of the parallel shake can be set by using the motion signal of the image pickup element 94 that has already been corrected for the angular shake. Using this, accurate shake correction is performed.

  Even when the accuracy of the accelerometer 11p to be used varies depending on the temperature and environment, it is calibrated with the signal from the image sensor 94, so that accurate shake correction can be performed using the signal from the accelerometer 11p during shooting.

  Further, since the offset component superimposed on the accelerometer integral value is removed, the second-order integrator 51 is not affected by the offset with respect to the accelerometer 11p shown in FIG. The deflection displacement can be obtained.

  As described above, in the first embodiment, in order to extract only the parallel shake component from the shake applied to the camera from the image signal, the image signal after the correction of the angular shake is used and the frequency band is obtained by the band pass filter. We focus on what can be done by limiting. Then, output calibration of the accelerometer 11p is performed based on the extracted parallel shake component, thereby enabling highly accurate translational shake correction.

  Specifically, the camera according to the first embodiment has the following configuration. Acceleration detection means (accelerometers 11p, 11y) for detecting translational vibration applied to the camera (camera body 93, interchangeable lens 90), and shake correction means (shake correction unit 95) for reducing deterioration due to image shake on the shooting screen Have Furthermore, it has a motion detection means (camera CPU 93d) for detecting the motion component of the captured image from the image signal from the image sensor 94. Furthermore, it has translational vibration extraction means (camera CPU 93d, differentiator 58, band pass filter 59, and part for executing the operation of step # 1005 in FIG. 5) for extracting translational vibration based on the output signal of the motion detection means. Further, a calibration unit (comparator 57, image magnification gain setting unit 55, part for executing the operation of step # 1008 in FIG. 5) for calibrating the signal from the acceleration detection unit based on the output signal of the translational vibration extraction unit. Have. Then, the shake correcting means is operated based on the signal from the acceleration detecting means.

  Further, it has angular velocity detection means (angular velocity meters 96p, 96y) for detecting angular vibration applied to the camera, and the shake correction means is operated based on the acceleration signal calibrated by the calibration means and the signal from the angular velocity detection means. ing.

  The motion detection means detects the motion component of the photographed image in a state where the shake correction means is actuated based on the output signal of the angular velocity detection means during preparation for photographing, and the angular shake component is relaxed.

  With the above-described configuration, it was possible to accurately alleviate deterioration due to parallel shake of a captured image caused by translational vibration applied to the camera.

  Next, a digital camera according to Embodiment 2 of the present invention will be described. Note that the configuration of the camera is the same as in the first embodiment.

  In the first embodiment, the angular shake is corrected, the motion component of the remaining parallel shake is detected by the image sensor 94, and the accelerometer 11p is calibrated as a result. On the other hand, in the second embodiment of the present invention, the angular shake is not corrected when the motion component is detected, and only the parallel shake component is extracted from the image information of the image sensor 94 to calibrate the accelerometer 11p.

  FIG. 8 shows image information captured by the image sensor 94, and the screen 71 is divided into a plurality of areas 71a to 71t.

  As a focusing method in the camera, the focus lens is extended, the blur of each area of the screen is evaluated for each extension amount, and the focus lens is set at the lens extension position with the least blur (sharp image). The method is widespread. In the case of this method, the focus lens is scanned at the time of focusing to evaluate the blur of the screen areas 71a to 71t. For example, after focusing on the main subject (person 72a in FIG. 8), the background (buildings 73a and 74a in FIG. 8) is far from the camera and how far away it is from the camera. I can do it.

  That is, in FIG. 8, the main subject 72a is focused. Then, the image areas 71b, 71c, 71d, 71h, 71i, 71m, and 71n are main subject areas, and the image areas 71a, 71e, 71f, 71g, 71j, 71k, 71l, 71o, 71p, 71q, 71r, 71s, and 71t are This is the subordinate subject area.

  Here, the motion component (motion vector) detected from the image signal in each region is indicated by an arrow at the center of each region. The motion vector is obtained by reading time-series changes in the imaging signal. However, when the subject is close to the camera, the shooting magnification increases, and image shift due to angular shake and parallel shake appears as a motion vector. .

  On the other hand, when the slave subject is far from the camera, the photographing magnification is low, and an image shift due to only the angle shake appears as a motion vector.

  In FIG. 8, large motion vectors appear in the main subject areas 71b, 71c, 71d, 71h, 71i, 71m, and 71n due to the influence of parallel shake. In addition, the sub subject areas 71a71e, 71f, 71g, 71j, 71k, 71l, 71o, 71p, and 71t are not affected by the parallel shake, so the motion vectors are small. In addition, since there is no image information regarding the areas 71q, 71r, and 71s, a motion vector cannot be acquired, and this area is not used for subsequent processing.

  The difference between the motion vectors in the main subject areas 71b, 71c, 71d, 71h, 71i, 71m, 71n and the motion vectors in the sub subject areas 71a, 71e, 71f, 71g, 71j, 71k, 71l, 71o, 71p, 71t. Take. And this difference is calculated | required as a motion vector by a parallel shake.

  That is, the object distance for each of a plurality of areas of the image signal obtained two-dimensionally on the imaging screen and the movement of the captured image corresponding to the area are detected. Then, the translational vibration component is extracted from the difference between the motion of the captured image in the region where the subject distance is close and the motion of the captured image in the region where the subject distance is far. Then, the accelerometer 11p is calibrated based on the parallel shake motion signal thus obtained.

  FIG. 9 is a flowchart showing the operation of the main part according to the second embodiment of the present invention. This flow starts when the main power of the camera is turned on. In order to explain the main configuration of the present invention in an easy-to-understand manner, various control steps (for example, battery check, photometry, distance measurement, lens driving for AF, strobe charging, and photographing are provided in the camera. Operation, movement, etc.) are omitted. In addition, steps that perform the same operations as those in the first embodiment are denoted by the same step numbers, and the description thereof is simplified.

  In FIG. 9, in step # 1001, it waits for switch S1 to turn on by half-pressing operation of release member 93a, and when switch S1 is turned on, it proceeds to step # 1006. In the next step # 1006, the focusing operation is started. Then, in the next step # 1007, the process waits until the lens extension for focusing is completed. Here, focusing is based on the TV-AF method described above. First, the focus lens 99 is scanned from infinity to the nearest position, and the blur of the image signal at each focus lens position is evaluated. Then, the focus position where the image is sharpest in the screen area determined as the main subject is searched, the focus lens 99 is moved to that position, and the focusing is finished. At this time, the focus state in the sub subject area is also obtained from the blur evaluation value of the image at the time of scanning the focus lens, and the focus state of the sub subject with respect to the main subject is known.

  In the next step # 1002, the initial posture direction detector 23p detects the posture of the camera from the signal from the accelerometer 11p. In the subsequent step # 1003, sensitivity correction according to the lens state and frequency correction according to the shake state (panning, etc.) are performed on the angular shake angle signal. At this time, the angular shake is detected by the angular velocity meter 96p. In the next step # 1005, the gravitational effect calculation unit 24p calculates the gravitational acceleration superimposed on the accelerometer 11p based on the camera posture by the initial posture direction detection unit 23p and the angular shake angle information from the angular velocity integration unit 13p. The acceleration gravity correction unit 21p corrects the error output. That is, at this time, the accelerometer 11p is driven to start detecting parallel shake. In step # 1015, extraction of parallel shake is started. This is to extract only the motion component due to the parallel shake based on the difference between the motion vectors of the main subject area and the subordinate subject area when the focusing operation is performed in step # 1007 as described above.

  In the next step # 1008, output calibration of the accelerometers 11p and 11y is performed. Details of this step are as described with reference to FIG. In the subsequent step # 1009, it is determined whether or not the switch 2 is turned on by the full pressing operation of the release member 93a. If the switch 2 is turned on, the process proceeds to step # 1010. If not, the process proceeds to step # 1013. Proceed to

  First, the case where the switch 2 is turned on and the process proceeds to step # 1010 will be described. In Step # 1010, the shake correction unit 95 is driven based on the angular shake signal from the angular velocity meter 96p and the signals from the accelerometers 11p and 11y. In the next step # 1011, photographing is performed, and when photographing is completed, the process proceeds to step # 1012, where it is determined whether or not the release member 93a is fully pressed and the switch S2 is turned off. If the switch S2 is not turned off, the process waits in step # 1012 until it is turned off. If the switch S2 is turned off, the process proceeds to step # 1013.

  When the switch S2 is turned off in step # 1009 or step # 1012, the process proceeds to step # 1013 as described above, and here the state of the switch S1 is detected. If the switch S1 is also off, the process proceeds to step # 1014 to stop driving the shake correction unit 95, that is, stop the shake correction operation, return to step # 1001, and wait for the switch S1 to be turned on again. . If the switch S1 remains on in step # 1013, the process returns to step # 1006 and the same operation is continued.

  In the second embodiment, attention is paid to the fact that it is possible to extract only the parallel shake component from the shake applied to the camera from the image signal by using the difference in motion information for each subject distance. Then, the accelerometer 11p is calibrated based on the extracted parallel shake component to enable highly accurate translational shake correction.

  Specifically, the camera of the second embodiment has the following configuration as in the first embodiment. Acceleration detection means (accelerometers 11p, 11y) for detecting translational vibration applied to the camera (camera body 93, interchangeable lens 90), and shake correction means (shake correction unit 95) for reducing deterioration due to image shake on the shooting screen Have Furthermore, it has a motion detection means (camera CPU 93d) for detecting the motion component of the captured image from the image signal from the image sensor 94. Furthermore, it has translational vibration extraction means (camera CPU 93d, a part for executing the operation of step # 1015 in FIG. 9) for extracting translational vibrations based on the output signal of the motion detection means. Furthermore, it has a calibration means (part which performs operation | movement of step # 1008 of FIG. 9) which calibrates the signal from an acceleration detection means based on the output signal of a translational vibration extraction means. Then, the shake correcting means is operated based on the signal from the acceleration detecting means.

  The motion detection means detects a motion component of the captured image from the relationship between the subject distance for each of a plurality of regions of the image signal obtained two-dimensionally and the captured image corresponding to the region. Further, the translational vibration extracting means extracts a parallel shake component from the difference between the motion of the captured image in the region where the subject distance is close and the motion of the captured image in the region where the subject distance is far.

  With the above-described configuration, it was possible to accurately alleviate deterioration due to parallel shake of a captured image caused by translational vibration applied to the camera.

  In each of the embodiments described above, the explanation of the parallel shake countermeasure has been continued by taking the image stabilization control system of the digital camera as an example. However, since the device of the present invention can be combined into a small and highly stable mechanism, it is not limited to a digital camera, but is an example of an optical device having a shooting function for a still picture of a digital video camera, a surveillance camera, a Web camera, and a shooting function It can also be used to shoot still images on a mobile phone.

It is a figure which shows the upper surface and side surface of the camera concerning each Example of this invention. It is a block diagram which shows the circuit structure of the camera concerning each Example of this invention. It is a figure for demonstrating the gravity error added to the accelerometer concerning each Example of this invention. It is a figure which shows the relationship between the hand-shake rotation angle concerning each Example of this invention, the output of an accelerometer, and time. It is a flowchart which shows operation | movement of the principal part of the camera concerning Example 1 of this invention. It is a block diagram which shows the circuit structure regarding the output calibration of the accelerometer concerning each Example of this invention. It is a flowchart which shows the operation | movement at the time of the output calibration of the accelerometer concerning Example 1 of this invention. It is a figure for demonstrating the parallel shake extraction method concerning Example 2 of this invention. It is a flowchart which shows the operation | movement at the time of the output calibration of the accelerometer concerning Example 2 of this invention. It is a figure which shows the upper surface and side surface of the camera of a prior art example.

Explanation of symbols

11p Accelerometer 11y Accelerometer 13p Angular velocity integration unit 13y Angular velocity integration unit 19 Shooting distance detection unit 21p Acceleration gravity correction unit 21y Acceleration gravity correction unit 22p Acceleration integration unit 22y Acceleration integration unit 24p Gravity effect calculation unit 24y Gravity effect calculation unit 25p Image magnification Correction unit 25y Image magnification correction unit 26p Shooting magnification calculation unit 26y Shooting magnification calculation unit 35p Parallel shake target value initialization unit 36p Shooting distance correspondence unit 51, 52 Integrator 55 Multiplication gain setting unit 56, 59 Band pass filter 57 Comparator 58 Differentiator 96p Angular velocity meter 96y Angular velocity meter 97 Lens CPU
510 Low-pass filter 511 Comparator 512 Low-pass filter

Claims (7)

Acceleration detecting means for detecting translational vibration applied to the image pickup apparatus to be image-blind,
Shake correction means for reducing deterioration due to image shake on the screen of the imaging device;
Motion detection means for detecting a motion component of a captured image from an image signal of the imaging device;
A translational vibration extracting means for extracting a translational vibration based on an output signal of the motion detecting means;
Calibration means for calibrating a signal from the acceleration detection means based on an output signal of the translational vibration extraction means;
An image stabilization control apparatus that operates the shake correction unit based on a signal from the acceleration detection unit calibrated by the calibration unit. Angular velocity detection means for detecting angular vibration applied to the imaging device;
2. The image stabilization control apparatus according to claim 1, wherein the shake correction unit is operated based on a signal calibrated by the calibration unit and a signal from the angular velocity detection unit.   3. The motion detection unit detects a motion component of the photographed image in a state in which the shake correction unit is actuated based on an output signal of the angular velocity detection unit to relax the angular shake component. The vibration control device described in 1.   4. The prevention according to claim 1, wherein the translation vibration extracting unit extracts a translational vibration by extracting a predetermined frequency band in a hand vibration frequency band from an output signal of the motion detection unit. Vibration control device.   3. The prevention according to claim 1, wherein the motion detection unit detects a subject distance for each of a plurality of regions of the image signal obtained two-dimensionally and a motion component of a captured image corresponding to the region. Vibration control device.   The translation vibration extracting unit extracts the translational vibration from a difference between a motion component of the photographed image in a region where the subject distance is short and a motion component of the photographed image in a region where the subject distance is far. Item 6. The vibration control device according to Item 5.   An image pickup apparatus comprising the image stabilization control device according to claim 1.

Images