JP4507548B2 - Blur correction device - Google Patents

Description

  The present invention relates to a blur correction device that corrects image blur by moving a part or all of a lens with a photographing device such as a camera, a lens barrel, and a video.

  In recent years, in order to prevent camera shake, there has been a technology of a shake correction device that detects camera shake and moves part of the lens along the camera shake to correct image blur on the film surface. It is being established.

FIG. 16 is a conceptual diagram showing a camera provided with a shake correction apparatus.
The blur of the camera 200 has 6 degrees of freedom, and is divided into a pitching, yawing, and rolling motion that is a rotational motion of 3 degrees of freedom and a motion in the X, Y, and Z directions that is a translational motion of 3 degrees of freedom. It is done. The blur correction of the camera 200 is usually performed for a motion with two degrees of freedom of pitching and yawing.

  The blurring motion of the camera 200 is monitored by the angular velocity sensors 102x and 102y. The angular velocity sensors 102x and 102y are piezoelectric vibration type sensors that detect Coriolis force generated by normal rotation. The angular velocity sensor 102x is an angular velocity meter for pitching blur detection. Similarly, the angular velocity sensor 102y is a yawing blur detection. It is an angular velocity meter for use.

  When performing the shake correction, the shake correction apparatus sends the outputs of the angular velocity sensors 102x and 102y to the CPUs 103x and 103y, and calculates the target drive position of the shake correction lens 101. The CPUs 103x and 103y send instruction signals to the voltage drivers 104x and 104y in order to drive the blur correction lens 101 to the target drive position, and the voltage drivers 104x and 104y supply power to the VCMs 105x and 105y along the instruction signals. I do. The blur correction lens 101 is driven by the VCMs 105x and 105y. In this way, the blur correction can be performed by driving the blur correction lens 101 according to the blur.

However, in recent years, the influence of blurring at the time of fixing a tripod (so-called tripod blurring) is also regarded as a problem. Tripod blur means that, for example, when shooting with a tripod attached to a camera, the tripod shakes due to the influence of the mechanism operation inside the camera body, affecting the shooting result.
Further, the reason that tripod blur has been regarded as a problem is that the allowable range of shake has become narrower with the spread of digital cameras. In the case of a silver salt camera, the permissible circle of confusion is about φ30 μm from the viewing conditions. However, in the case of a digital camera, the permissible circle of confusion is smaller than that of a silver salt camera due to the difference in viewing conditions.

Further, in the conventional shake correction apparatus, the shake is detected, and the shake correction lens 101 is driven according to the shake to optically correct the image blur (for example, see Patent Document 1). This blur correction corresponds to the frequency band of camera shake (about 0.1 to 10 Hz).
On the other hand, the frequency band of the tripod shake differs depending on, for example, the lens weight and the rigidity of the tripod.
In order to follow the frequency band higher than the camera shake and perform the blur correction when the tripod is fixed, for example, problems such as the delay of the angular velocity sensors 102x and 102y and the delay due to control sampling must be taken into consideration. Further, in tripod blur, since the absolute shake amount is smaller than the hand shake, the influence of drift of the angular velocity sensors 102x and 102y cannot be ignored.
Japanese Patent Laid-Open No. 11-38461

  SUMMARY OF THE INVENTION An object of the present invention is to provide a shake correction apparatus capable of performing high-accuracy shake correction in consideration of an angular velocity sensor / control sampling delay.

The present invention solves the above problems by the following means. In order to facilitate understanding, description will be made with reference numerals corresponding to the embodiments of the present invention, but the present invention is not limited to this. That is, according to the first aspect of the present invention, the shake detection unit (3) that detects the shake of the imaging device (2) and the image blur are optically corrected according to the output of the shake detection unit (3). Based on the output of the shake correction unit (4) that operates, the position detection unit (7) that detects the position of the shake correction unit (4), and the shake detection unit (3), the imaging device (2) Based on the output of the support state determination unit (21) for determining whether the support state is handheld or fixed, and the shake detection unit (3), a target speed calculation unit (4) that calculates the target speed of the shake correction unit (4) and 26), on the basis of the position of the blur correction unit calculated output or from the target speed of the position detector (7) (4), wherein the shake correction unit (4) is etc. far Iho from the movable center of the range The amount of change in the target speed is determined so as to give a large centripetal force to the blur correction unit (4) That the speed bias calculation unit (28), wherein the speed bias calculation unit (28), when the supporting state determination unit (21) is determined fixed, than when it is determined that hand, the movable range center wherein to the centripetal force rather large in the vicinity, reducing the centripetal force at a position distant from the movable range center, a motion compensation device according to claim.
According to the second aspect of the present invention, the shake detection unit (3) that detects the shake of the imaging device (2) and the image blur are optically corrected according to the output of the shake detection unit (3). Based on the output of the shake correction unit (4) that operates, the position detection unit (7) that detects the position of the shake correction unit (4), and the shake detection unit (3), the imaging device (2) Based on the output of the support state determination unit (21) for determining whether the support state is handheld or fixed, and the shake detection unit (3), a target speed calculation unit (4) that calculates the target speed of the shake correction unit (4) and 26), an output or said blur correcting unit, which is calculated from the target speed based on the position (4), the target to provide a centripetal force before Symbol blur correction unit (4) of the position detector (7) A speed bias calculation unit (28) for determining a speed change amount, and the speed bias Calculation unit (28), wherein the shake correction unit amount of change in the target speed (4) is significantly farther from the movable center of the range, hand-held when the supporting state determination unit (21) is determined to be fixed The blur correction device is characterized in that the rate of increase in the gradient of the change amount of the target speed is made smaller than when it is determined to be present.

According to a third aspect of the present invention, in the shake correction apparatus according to the first or second aspect , the speed bias calculation unit (28) is configured to calculate an output between the position detection unit (7) and a change amount of the target speed. The relationship includes a first table expressed by a cubic expression and a second table expressed by a linear expression, and the speed bias calculation unit (28) determines that the support state determination unit (21) is handheld. In this case, the change amount of the target speed is determined using the first table, and when the support state determination unit (21) determines that the target speed is fixed, the second table is used to determine the target speed. The blur correction device is characterized in that a change amount is determined.
A fourth aspect of the present invention, the motion compensation device according to any one of claims 1 to 3, for correcting the image blur optically with the output of the shake detecting unit (3) A target speed calculation unit (26) for calculating a target speed, an amplification unit for amplifying the output of the position detection unit (7), and a subtraction for subtracting the output of the amplification unit from the output of the target speed calculation unit (26) And a blur correction device including an integration unit that integrates an output of the subtraction unit.
According to a fifth aspect of the present invention, there is provided a photographing apparatus comprising the blur correction device according to any one of the first to fourth aspects.

  The blur correction device according to the present invention includes: (1) a speed bias calculation unit that determines a change amount of the target speed based on an output of the position detection unit or a position of the blur correction unit calculated from the target speed; Is determined to be fixed, the amount of change in the target speed is increased, so that stable blur correction can be performed even when the blur correction unit is located near the movable center.

(2) The speed bias calculation unit uses the first table in which the relationship between the output of the position detection unit and the amount of change in the target speed is expressed by a cubic equation when the support state determination unit determines that the hand is held. When the speed change amount is determined, and similarly, when it is determined to be fixed, the target speed change amount is determined using the second table represented by the linear expression, so blur correction is performed with high accuracy. be able to.

(3) Since the low-pass filter that extracts the low frequency component of the signal detected by the shake detection unit changes the parameter to the high frequency side when the support state determination unit determines that the signal is fixed, the drift of the shake detection unit Components can be removed.

(4) The phase compensation calculation unit for compensating for the phase lag of the shake detection unit cancels the compensation operation or changes the parameter when the support state determination unit determines that it is fixed. Sometimes an increase in gain can be prevented.

(5) Based on the output of the shake detection unit, the position bias calculation unit that compresses the gain of the target position calculation unit that calculates the target position of the shake correction unit is configured so that when the support state determination unit determines that the gain is fixed, Since the compression is released, the correction error associated with the blur correction can be removed when the tripod is fixed.

(6) a first calculation unit that performs a feedback calculation, and a second calculation unit that performs a feedforward calculation by adding the output value of the first calculation unit to a value obtained by multiplying a signal indicating the target position by a predetermined gain. When the support state determination unit determines that the support state is fixed, the drive control unit is set to a negative gain and controls the shake correction unit to follow the target position. , Gain characteristics and phase characteristics can be improved.

  The best mode for carrying out the invention will be described below with reference to the drawings and the like.

  FIG. 1 is a block diagram showing an outline of an embodiment of a shake correction apparatus 100 according to the present invention. In this specification, blur correction refers to correction in the pitching direction and yawing direction. In FIG. 1, the shake correction device 100 in the pitching direction is described, but the same applies to the shake correction device in the yaw direction. Further, the camera 2 corresponds to the cross-sectional view of the camera 2 in the YZ plane shown in FIG.

The blur correction apparatus 100 according to the present embodiment includes a camera 2, a lens CPU 20, and the like. The camera 2 and the lens CPU 20 are connected via amplifiers and low-pass filters 34 and 35, a PWM (pulse width modulation) driver 36, and the like.
The camera 2 includes a lens barrel 1 and the like. The lens barrel 1 includes lenses 9, 10, and 11 of a photographing optical system, support bases 12, 13, and 14 of the lenses 9, 10, and 11, a CCD 15, an angular velocity sensor 3, a shake correction lens 4, and a shake correction lens. 4 includes a lens chamber 5 for fixing 4, a VCM 6 for driving the shake correction lens 4, a correction lens position detection unit (PSD) 7 for detecting the position of the shake correction lens 4, and the like. Further, the blur correction lens 4 can move up and down so as to correct the inclination of the optical axis 16 due to the blur in the pitching direction. The angular velocity sensor 3 is a vibration gyro-type sensor that detects Coriolis force (vibration), and is floatingly supported on the lens barrel 1 via the connecting portion 3a in order to be hardly affected by acceleration due to mechanism operation. ing.

  The lens CPU 20 includes a support state determination unit 21, a follow-up control unit 22, a target position conversion unit 23, an HPF processing unit 24, and the like. The lens CPU 20 is connected to an EEPROM 31, a focal length encoder 32, a focus (absolute distance) encoder 33, and the like, and is further connected to a body CPU 40, a release SW 41, and the like via a mount contact (not shown).

  The lens CPU 20 communicates with the body CPU 40 via the mount contact. Information on the release SW 41 is input to the body CPU 40, and it can be detected whether the release SW 41 is half-pressed or fully pressed. In synchronization with the half-press ON of the release SW 41, a shake correction start command is sent from the body CPU 40 to the lens CPU 20 in synchronization with the half-press OFF.

Here, the flow of various signals (data) in the lens CPU 20 will be schematically described.
The output of the angular velocity sensor 3 is subjected to the removal of the high frequency noise component via the amplifier and the LPF 35, and then quantized by the A / D converter, and the angular velocity data ω1 is sent to the HPF processing unit 24 and the support state determination unit 21. Entered.
The HPF processing unit 24 calculates the angular velocity data ω3 from which the drift component of the angular velocity sensor 3 has been removed by calculating and subtracting the center level of the angular velocity data ω1 using the LPF, and sends the angular velocity data ω3 to the target position converting unit 23. Output.

Based on the angular velocity data ω1, the support state determination unit 21 determines whether the camera 2 is in a handheld state or is fixed to a tripod or the like (see FIG. 13), and a determination signal k indicating the determination result is Output to the target position conversion unit 23 and the follow-up control unit 22.
The output of the PSD 7 is input to the target position conversion unit 23 and the tracking control unit 22 as the position signal y of the blur correction lens 4 via the amplifier, the LPF 34, and the A / D converter.
The target position conversion unit 23 determines, from the angular velocity data ω3, the signal f from the focal length encoder 32, the output D from the focus encoder 33, the parameters used for calculation from the EEPROM 31, and the position signal y of the blur correction lens 4 and the determination. In accordance with the signal k, the target position signal g of the blur correction lens 4 is calculated. Further, the target position conversion unit 23 outputs the target position signal g to the tracking control unit 22.

The follow-up control unit 22 calculates a drive signal p for the shake correction lens 4 based on the target position signal g and the position signal y and further according to the determination signal k. The drive signal p is input as a digital drive signal to the PWM driver 36 via the D / A converter.
The PWM driver 36 performs switching on the drive signal, applies a voltage to a coil portion (not shown) of the drive VCM 6 of the shake correction lens 4, and drives the VCM 6. Further, the position of the blur correction lens 4 is monitored by the PSD 7 and is fed back to the tracking control unit 22 and the target position conversion unit 23 as the position signal y through the A / D converter as described above.

Hereinafter, functions and the like of each member will be described in detail.
FIG. 2 is a block diagram showing details of the lens CPU 20. Here, with reference to the target position conversion unit 23, the HPF processing unit 24, the support state determination unit 21, and the follow-up control unit 22 are associated with each other, and various algorithms associated with blur correction control will be described in detail. The target position conversion unit 23 includes a phase advance compensation unit 25, a correction formula calculation unit 26, an integration calculation unit 27, a speed bias calculation unit 28, a position bias calculation unit 29, and the like.

(About LPF algorithm by HPF processing unit 24)
The HPF processing unit 24 includes a low-pass filter 24a. As described above, the LPF processing is performed on the angular velocity data ω1 according to the determination signal k, and the angular velocity data ω3 obtained by removing the drift component of the angular velocity sensor 3 is obtained. calculate. Specifically, the angular velocity data ω3 is obtained by subtracting the output ω2 of the low-pass filter 24a from the angular velocity data ω1.
That is, the LPF algorithm is a process for cutting a low frequency component. For example, the low frequency component ω2 is extracted by a low-pass filter 24a that is an IIR LPF, and the low frequency component ω2 is converted into an angular velocity. Subtract from the data ω1. Therefore, the calculation result ω3 has an HPF characteristic.

In the HPF processing unit 24, for example, LPF1 and LPF2 having different cutoff frequencies are selectively used according to the determination signal k from the support state determination unit 21.
The LPF 1 is a filter having a cutoff frequency for hand-held shooting, and is set to about 0.1 Hz, for example.
The LPF 2 is a filter having a cutoff frequency for photographing when the tripod is fixed, and is set to about 1.0 Hz, for example.

Here, selection of the cut-off frequency will be described.
FIG. 3 is a diagram showing the frequency distribution of drift, hand shake, and tripod shake of the angular velocity sensor 3. The horizontal axis is frequency (Hz), and the vertical axis is blur angle (deg).
As shown in FIG. 3, the frequency band of drift of the angular velocity sensor 3 is about 0 to 1.0 Hz, the frequency band of shake during hand-held shooting is about 0.1 to 10 Hz, and the frequency band when the tripod is fixed is It is about 5 to 40 Hz, and each frequency band is different. For this reason, when shooting with a tripod fixed, correction at a low frequency is not necessary as compared with hand-held shooting. Further, the drift of the angular velocity sensor 3 is that whose output drifts regardless of the shake after the angular velocity sensor 3 is turned on, and this drift takes several seconds to stabilize.

  Therefore, when the support state determination unit 21 determines that the shooting is performed when the tripod is fixed, the HPF processing unit 24 sets the cutoff frequency to about 1 Hz in order to suppress the drift of the angular velocity sensor 3. Select LPF2. Further, the HPF processing unit 24 clears the accumulated value of the LPF algorithm so that the output ω3 does not become discontinuous when the cutoff frequency is switched.

(Regarding the phase advance algorithm by the phase advance compensation unit 25)
FIG. 4 is a block diagram illustrating the operation principle of the phase lead compensation unit 25.
The phase advance algorithm by the phase advance compensation unit 25 is a process for mainly compensating for the phase lag of the angular velocity sensor 3. The reason why the phase lag occurs in the angular velocity sensor 3 is, for example, that the angular velocity sensor 3 is floatingly supported on the lens barrel 1 via the connecting portion 3a.
In the phase lead compensation unit 25, the angular velocity data ω3 from the HPF processing unit 24 is used as current information, and information before sampling 6 (1 / Z ⁶ by so-called Z conversion) is subtracted from this current information, and then a multiplication circuit 25a. Thus, by multiplying the subtraction result by a constant (a differential constant Ts that is a parameter), the signal ω is calculated, and the phase delay is compensated.
For example, a constant Ts is stored in the multiplication circuit 25a, and a determination signal k from the support state determination unit 21 is input as illustrated. Thereby, the multiplication circuit 25a can change the value of the constant Ts according to the determination signal k.

FIG. 5 is a diagram illustrating the relationship between phase advance and gain.
The phase advance algorithm is differential compensation as described above. For this reason, the phase advance algorithm advances the phase and raises the gain as shown in the figure.
This increase in gain does not affect the blur correction accuracy in the camera shake frequency band (0.1 to 10 Hz), but has a certain effect in the tripod shake frequency band (5 to 40 Hz).
Therefore, the phase advance compensation unit 25 cancels the phase advance algorithm (differential constant Ts = 0) or sets the parameter Ts according to the determination signal k when the support state determination unit 21 determines that the tripod is fixed. change.

(Regarding the correction formula calculation unit 26)
The correction formula calculation unit 26 outputs a signal ω, which is the output of the above-described phase advance algorithm, a signal f output from the focal length encoder 32, a signal D output from the subject distance (focus) encoder 32, and an output from the EEPROM 31. Based on the lens specific signal (Ds or the like) to be converted to the target speed VC of the blur correction lens 4. The lateral magnifications β and VR can be approximated as a function of f and D.
The corrected lens target speed VC is calculated by the following formula.
VC = f × ω / VR (when D ≧ Ds)
VC = β × D × ω / VR (when D <Ds)
VC: Corrected lens target speed [mm / s]
ω: Phase advance algorithm output [rad / s]
f: Focal length [mm]
β: lateral magnification D: subject distance (obtained from absolute distance encoder) [mm]
Ds: Formula switching subject distance threshold VR: VR correction coefficient (VR = (image movement amount on the image plane) / (correction lens movement amount))

(Regarding the speed bias algorithm by the speed bias calculator 28)
As shown in FIG. 2, the speed bias calculation unit 28 outputs an output signal of the PSD 7, which is a lens position calculation result y (hereinafter, current position signal y) indicating the current position of the blur correction lens 4, and a support state determination. The determination signal k from the unit 21 is input. Based on the current position signal y, the speed bias calculation unit 28 calculates a correction amount (speed bias data s1) of the target speed VC of the blur correction lens 4 according to the determination signal k.
The speed bias algorithm is a process for giving the blur correction lens 4 a centripetal force toward the bias center position. For example, the velocities are given by subtracting the speed bias data s1 from the target speed VC.

FIG. 6 is a conceptual diagram showing a speed bias algorithm.
Here, the speed bias calculation unit 28 is regarded as a multiplication circuit, and the speed bias calculation unit 28 multiplies the current position signal y from the PSD 7 by a gain of “10”, as shown in the figure, to thereby obtain speed bias data. s1 is calculated.
Multiplying the gain “10” indicates, for example, that the speed bias data s1 “10” is output when the blur correction lens 4 moves “1”. The gain “10” means an inclination at the current position signal y in a speed bias table (see FIG. 7) described later.

Next, a signal VC-s1 obtained by subtracting the speed bias data s1 from the target speed VC is output to the integral calculation unit 27. The integration calculation unit 27 calculates the position signal c before the position bias by integrating the signal VC-s1. In the figure, the process of converting the position signal c into the drive signal p for the shake correction lens 4 is omitted (see FIGS. 1 and 2).
The drive signal p is a drive signal for the shake correction lens 4 and is applied to a coil (not shown) of the drive VCM 6 for the shake correction lens 4 via the D / A converter and the PWM driver 36 as described above. Then, the position of the blur correction lens 4 driven by the VCM 6 is output from the PSD 7 as the current position signal y, and is input again to the speed bias calculation unit 28.
Here, the current position signal y from the PSD 7 is used as the input to the speed bias calculation unit 28, but the same applies even if the target position signal g of the blur correction lens 4 that is output from the position bias calculation unit 29 is used. Characteristics can be obtained.

The speed bias algorithm will be described using a transfer function.
When the input signal is the target speed VC and the output signal is the current position signal y, this transfer function is
y / VC = 1 / (S + 10)
s: indicated as an operator. This transfer function is a so-called Laplace transform transfer function, and represents a first-order LPF including the drive unit of the blur correction lens 4.

Here, the speed bias table will be described.
FIG. 7 is a diagram showing a speed bias table used in the speed bias calculator 28. The horizontal axis represents the current position signal y of the blur correction lens 4, and the vertical axis represents the speed bias output.
The speed bias table 50 shows a speed bias shape when the support state is handheld. In the speed bias table 50, as shown in FIG. 7A, the speed bias output is represented by a cubic expression of the current position signal y. The reason why the speed bias table 50 represented by a cubic function is used in the case of hand-held is that the speed bias when hand-held is intended to secure a movable range, so the bias center position is fixed at the center of the movable range. It is.
That is, in the speed bias algorithm, when the support state determination unit 21 determines that the hand is held, by using the speed bias table 50, the gradient of the cubic function becomes sharper as the blur correction lens 4 moves away from the movable range center. Therefore, the speed bias calculation unit 28 can give a large centripetal force to the shake correction lens 4.

On the other hand, when it is determined that the tripod is fixed by the determination signal k of the support state determination unit 21, the speed bias calculation unit 28 changes from the speed bias table 50 represented by the cubic equation to the velocity bias table 60 represented by the primary equation. Switch to run the speed bias algorithm.
The speed bias table 60 shows a speed bias shape when the support state is fixed to a tripod. In the speed bias table 60, as shown in FIG. 7B, the speed bias output is expressed by a linear expression of the current position signal y.

Here, the reason for using the speed bias table 60 will be described.
When the cubic velocity bias table 50 is used, the velocity bias change amount at the bias center position becomes 0 (reason: y = 0, slope = 0). Bias output is lowered.
Accordingly, when it is determined that the tripod is fixed from a hand, the velocity bias is applied even stronger in the vicinity of the bias center position by using the velocity bias table 60 which is a linear expression of the velocity bias shape (reason: Regardless of the value of y, the slope is constant).
Thereby, even when a tripod is fixed with a small amount of blur, it is possible to prevent the blur correction lens 4 from moving in the vicinity of the movable center.

In addition, the bias center position of the speed bias can be changed, and if the support determination result by the support state determination unit 21 is determined to be fixed to a tripod (see FIG. 14: step S107), the current position of the shake correction lens 4 immediately after the determination is determined. The position can be changed to the bias center position of the speed bias (see FIG. 14: Step S110).
However, the bias center position needs to have a margin with respect to the soft limit of the blur correction lens 4. This is because the blur correction effect is lost when the blur correction lens 4 reaches the soft limit. Specifically, when the bias center position is in the vicinity of the soft limit, the blur correction lens 4 operates with the bias center position as the driving center, and the possibility that the blur correction lens 4 reaches the soft limit is increased. Is considered.

FIG. 8 is a diagram illustrating frequency characteristics according to the speed bias algorithm.
The frequency characteristic according to the speed bias algorithm can be expressed, for example, by making the above-described transfer function: y / VC = 1 / (S + 10) into a Bode diagram.
Here, the gain when the speed bias gain = 10 is compared with the gain when the speed bias gain = 0. As shown in the figure, the gain when the speed bias gain = 10 is 20 dB (that is, 10 times) at 0.16 (1 / 2π: 1 rad) Hz, and a large gain suppression effect is obtained. Therefore, the effect of suppressing the low frequency gain can be obtained by strongly applying the speed bias.
That is, according to the speed bias algorithm, this gain suppression effect suppresses the drift of the angular velocity sensor 3 or the correction of a slow operation for composition determination performed by the photographer during tripod shooting. And so-called composition determination can be easily performed.

(Regarding the position bias algorithm by the position bias calculator 29)
The position bias algorithm is a process of compressing position information (position signal c) of the blur correction lens 4. The position signal c from the integration calculation unit 27 is output to the position bias calculation unit 29. The position bias calculation unit 29 calculates a target position signal g of the blur correction lens 4 based on the position signal c in accordance with the determination signal k from the support state determination unit 21 (see FIG. 2).

FIG. 9 is a diagram illustrating the shape of the position bias algorithm. The horizontal axis is the position signal c, and the vertical axis is the target position signal g which is output information of the position bias algorithm.
The shape of the position bias algorithm will be described. As shown in the figure, the target position signal g (solid line in the figure) at the time of hand-held shooting is an output value obtained by compressing an input value (for example, “gain = 0.8”) up to a threshold value Pb_s. When the input value exceeds the threshold value Pb_s, the input value is further compressed (the gain is set to an appropriate value of 1.0 or less) to obtain an output value.

In the compression processing by this position bias algorithm, for example, the blur correction lens 4 can be prevented from reaching the above-described movable range limit, but a correction error in blur correction may occur.
Further, the shake amount of the tripod shake is smaller than the shake amount of the hand shake. For this reason, it is less necessary to execute the position bias algorithm for tripod shake.
Therefore, for example, when the position bias calculation unit 29 determines that the tripod is supported by the determination signal k of the support state determination unit 21, the position bias calculation unit 29 releases the compression processing by the position bias algorithm. As a result, the relationship between the position signal c and the target position signal g is represented by the shape indicated by the dotted line in the figure, and the position signal c becomes the target position signal g as it is.
In addition, when the compression process is canceled, the position bias calculation unit 29 receives a signal from the integral calculation unit 27 when the compression process is canceled so that the target position signal g output from the position bias calculation unit 29 does not become discontinuous. The position signal c is reset (point A is changed to point B in the figure).

(Regarding the tracking control algorithm by the tracking control unit 22)
The follow-up control unit 22 calculates a drive signal p for the shake correction lens 4 based on the target position signal g and the position signal y and further according to the determination signal k (see FIG. 2).
The follow-up control unit 22 includes a first calculation unit 22a, a second calculation unit 22b, and the like. For example, the first calculation unit 22a performs PID control (feedback calculation) using a deviation between the target position signal g of the shake correction lens 4 and the current position signal y. For example, the second calculation unit 22b performs a feedforward calculation by adding the value obtained by multiplying the target position signal g by a predetermined gain (FF output) and the output value of the first calculation unit 22a.

FIG. 10 is a block diagram illustrating the operation principle of the tracking control unit 22.
The first computing unit 22a first subtracts the current position signal y from the target position signal g and multiplies the value by a proportionality constant Kp (proportional term).
Furthermore, the integral value of the difference obtained by subtracting the current position signal y from the target position signal g is multiplied by an integration constant Ki (integration term).
Further, the differential value obtained by subtracting the current position signal y from the target position signal g is multiplied by a differential constant Kd (differential term). Here, Z represents Z conversion, and 1 / Z represents information before one sampling.
The result obtained by multiplying the proportional constant Kp, the result obtained by multiplying the integral constant Ki, and the result obtained by multiplying the differential constant Kd are added together to obtain the output of the PID control unit by the first calculation unit 22a.

FIG. 11 is a diagram illustrating a state where the gain used in the second arithmetic unit 22b is set to positive or negative.
As described above, the follow-up control unit 22 uses not only the PID output from the first calculation unit 22a but also the FF output from the second calculation unit 22b. The second calculation unit 22b multiplies the target position signal g of the shake correction lens 4 by the FF output (predetermined gain).
Moreover, the 2nd calculating part 22b sets FF output to positive, when hand-held determination is made according to the determination signal k of the support state determination part 21 (FIG. 11 (a)). Thus, by setting the FF output to be positive at the time of hand-holding determination, the centripetal force in the direction of the optical axis 16 of the VCM 6 can be canceled and the control characteristics in the low frequency region can be improved (see FIG. 12). .
Moreover, the 2nd calculating part 22b sets FF output to negative, when a tripod determination is carried out according to the determination signal k of the support state determination part 21 (FIG.11 (b)). Thus, at the time of tripod determination, the gain characteristics and phase characteristics in the high frequency region can be improved by appropriately setting the FF output as negative (see FIG. 12).

FIG. 12 is a diagram illustrating frequency characteristics by the tracking control unit 22.
First, the relationship between frequency and gain will be described. For example, the gain in the vicinity of 30 Hz is about 2.5 when the holding state is determined by the support state determination unit 21, whereas it is about 0 when it is determined that the tripod is fixed (FIG. 12 (a)). )). Therefore, the tracking control unit 22 can improve the gain characteristic particularly at a frequency of 30 Hz by setting the FF output to be negative in the second calculation unit 22b when the tripod is fixed.

  Next, the relationship between frequency and phase will be described. As for the phase in the vicinity of the frequency of 30 Hz, for example, when the hand-held determination is performed by the support state determination unit 21 and the case where it is determined that the tripod is fixed, the phase when the tripod is fixed is more advanced than the phase when the hand is held. (FIG. 12B). Therefore, the tracking control unit 22 can improve the phase characteristics particularly at a frequency of 30 Hz by setting the FF output to be negative in the second calculation unit 22b when the tripod is fixed.

(About the support determination algorithm by the support state determination part 21)
FIG. 13 is a diagram illustrating in detail the processing of the support determination algorithm. This process corresponds to step S107 in FIG. 14 described later.
The support determination algorithm determines whether the camera is in a hand-held state or fixed on a tripod or the like. The blur determination by the support determination algorithm is performed on the angular velocity data ω3 from which the drift component ω2 of the angular velocity sensor 3 is removed (see FIG. 2). Further, the angular velocity data ω3 detected when the hand is held has a larger amplitude and a lower frequency than when the angular velocity data ω3 is fixed to a tripod or the like. On the other hand, the angular velocity data ω3 detected when the tripod is fixed has a small amplitude and a high frequency.

Therefore, as shown in FIG. 13, the threshold value ωk is set to the amplitude value, the amplitude value is taken in at the sampling interval st, the threshold value ωk is compared with the magnitude of the amplitude value, and the number of times larger than the threshold value is determined for a certain time. Count (n × st time).
In the figure, the circled parts are counted. When the count count k becomes larger than a certain threshold count r within a certain time (r = <k), it is determined that the count is held. Conversely, if the count is r> k, it is determined that the tripod is fixed. When the determination is completed, the count number k is reset to zero.

FIG. 14 is a flowchart (1) regarding blur correction in the blur correction apparatus 100 according to the present invention, and FIG. 15 is a flowchart (2) regarding blur correction in the blur correction apparatus 100 according to the present invention. Note that the following steps (S101 to 122) indicate the processes of the lens CPU 20 described above, and redundant descriptions are omitted as appropriate.
The lens CPU 20 recognizes the shake correction start command from the body CPU 40 synchronized with the half-press ON of the release SW 41 (S100), and then sets FLAG = tripod (S101).

In step S102, initial values are set. In the initial setting, a tripod cutoff frequency fc = 1.0 Hz, phase advance: Ts = 0, speed bias: linear expression, speed bias center: movable center, position bias: none, tracking control: FF = FF_t Is set.
Here, the gain FF used in the second calculation unit 22b of the tracking control unit 22 will be described. When the support state determination unit 21 determines that the tripod is fixed, the FF is indicated as FF_t (tripod). When it is determined that the hand is held, FF_h (hand) is indicated.

In step S103, blur correction control is started, and the process proceeds to step S104. The lens CPU 20 monitors the half-press signal based on the state of the release SW 41 (S104). If the half-press is OFF (S104: Yes), the blur correction control is stopped (S105) and the process is terminated (S106).
On the other hand, when the half-press is ON (S104: No), the support state is determined (S107, see FIG. 13). If it is determined that the hand is held, the process proceeds to step S108, and if it is determined that the tripod is fixed, the process proceeds to step S112.

  In step S108, hand-held cutoff frequency fc = 0.1 Hz, phase advance: Ts = Ts_h, speed bias: cubic equation, speed bias center: movable center, position bias: yes, follow-up control: FF = FF_h Each is set. In the next step S109, FLAG = held on hand is set, and the process proceeds to step S113.

  On the other hand, if it is determined that the tripod is fixed in the blur determination in step S107, the process proceeds to step S112, where it is determined whether FLAG = held. If FLAG = hand-held is Yes, it means that it was determined to be hand-held in the previous blur determination, so the process proceeds to the next step S110, the tripod cutoff frequency fc = 1.0 Hz, and the phase advance: Ts = 0, velocity bias: linear equation, velocity bias center: current position, position bias: none, follow-up control: FF = FF_t.

  Next, in step S111, FLAG = tripod is set, and the process proceeds to step S113. If it is determined in step S112 that FLAG = hand-held is No, it means that it was determined that the tripod was fixed in the previous blur determination, so that resetting is not performed and the process proceeds to step S113.

  Next, the lens CPU 20 monitors the full-press signal based on the state of the release SW 41 (S113). If it is fully pressed OFF (S113: NO), the process returns to the above-described step S107 again. On the other hand, in the case of full-press ON (S113: Yes), the blur correction control is temporarily stopped (S114), and the pre-exposure centering operation is performed (S115).

After the pre-exposure centering operation in step S115, the lens CPU 20 proceeds to step S116, and determines whether FLAG = hand-held. If FLAG = hand-held is Yes, the process proceeds to the next step S117, where no speed bias and no position bias are set.
Here, the reason for setting no speed bias and no position bias in spite of being determined to be hand-held is after the shake correction control is performed in steps S103 to S113, and the shake correction lens 4 is Even if it becomes somewhat unstable, it is sufficient to remove the low frequency drift during the exposure.

On the other hand, if it is determined in step S116 that FLAG = hand-held is No, the process proceeds to step S118, and the speed bias center is set as the movable center. In step S119, the blur correction control is resumed, and in step S120, it is determined whether or not the exposure is completed.
If exposure is completed, the process proceeds to the next step S121, where it is determined whether FLAG = held. If FLAG = hand-held is Yes (S121: Yes), the lens CPU 20 sets speed bias and position bias (S122), and returns to step S104 described above. If FLAG = hand-held is No (S121: No), the process returns to Step S104.

According to the present embodiment, (1) the speed bias calculating unit 28 changes the target speed VC according to the determination signal k based on the current position signal y of the blur correction lens 4 that is the output of the PSD 7. When the speed bias data s1 is determined and the support state determination unit 21 determines that the speed bias data s1 is fixed, the speed bias data s1 is increased. Therefore, the blur correction lens 4 is positioned near the movable center. Even in this case, stable blur correction can be performed.
(2) The speed bias calculation unit 28, when the support state determination unit 21 determines that it is hand-held, a first table in which the relationship between the current position signal y and the current position signal y and the speed bias output is expressed by a cubic equation. 50 is used to determine the velocity bias data s1. Similarly, when it is determined that the velocity bias data s1 is fixed, the velocity bias data s1 is determined using the second table 60 represented by the linear expression. Correction can be performed with high accuracy.
(3) The HPF processing unit 24 that extracts the low frequency component of the signal ω1 detected by the angular velocity sensor 3 changes the cutoff frequency to the high frequency side when the support state determination unit 21 determines that the signal is fixed. The drift component ω2 (noise generated when the power is turned on) of the angular velocity sensor 3 can be removed.
(4) The phase advance compensation unit 25 for compensating for the phase delay of the angular velocity sensor 3 cancels the compensation operation (Ts = 0) or changes the parameter when the support state determination unit 21 determines that it is fixed ( Ts is made small), so that an increase in gain can be prevented when the tripod is fixed.
(5) The position bias calculation unit 29 calculates the target position g of the blur correction lens 4 by compressing the gain based on the output signal c (post-integration correction lens target position) of the integration calculation unit 27, and supports it. Since the gain compression is canceled when the state determination unit 21 determines that the state is fixed, the correction error associated with the shake correction can be removed when the tripod is fixed.
(6) The follow-up control unit 22 adds the output value of the first calculation unit 22a to the first calculation unit 22a that performs feedback calculation, and a value obtained by multiplying the signal g indicating the target position by a predetermined gain FF, And a second calculation unit 22b for performing a feedforward calculation, and when the support state determination unit 21 determines that it is fixed, the gain FF is set to be negative and the blur correction is performed so as to follow the target position signal g. Since the lens 4 is controlled, gain characteristics and phase characteristics can be improved when the tripod is fixed.

(Modification)
The present invention is not limited to the embodiments described above, and various modifications and changes are possible, and these are also within the equivalent scope of the present invention.
(1) The phase advance compensation unit 25 subtracts the information 6 samples before sampling (1 / Z ⁶ in the so-called Z conversion) from the signal ω3, which is the current information, and then multiplies it by the differential constant Ts to obtain the phase advance Although compensation has been performed, the information is not limited to information before six samplings, and a suitable number of previous sampling information may be used.

(2) The HPF processing unit 24 selectively uses LPF1 (0.1 Hz) and LPF2 (1.0 Hz) having different cut-off frequencies fc, but LPF1 <LPF2 and the angular velocity sensor 3 The numerical value is not limited as long as the drift component can be removed (for example, LPF2 may be 2.0 Hz).

It is a block diagram which shows the outline | summary of the Example of the blurring correction apparatus 100 by this invention. It is a block diagram which shows the detail of lens CPU20. It is a figure which shows the frequency distribution of the drift of the angular velocity sensor 3, a hand shake, and a tripod shake. FIG. 6 is a block diagram illustrating an operation principle of a phase lead compensation unit 25. It is a figure which shows the relationship between a phase advance and a gain. It is a conceptual diagram which shows a speed bias algorithm. FIG. 6 is a diagram showing a speed bias table used in the speed bias calculation unit 28. It is a figure which shows the frequency characteristic by a speed bias algorithm. It is a figure which shows the shape of a position bias algorithm. 3 is a block diagram illustrating an operation principle of a tracking control unit 22. FIG. It is a figure which shows the state which set the gain used by the 2nd calculating part 22b to positive or negative. It is a figure which shows the frequency characteristic by the tracking control part. It is a figure which shows the process of a support determination algorithm in detail. It is a flowchart (1) regarding the blur correction in the blur correction apparatus 100 by this invention. It is a flowchart (2) regarding the blur correction in the blur correction apparatus 100 by this invention. It is a conceptual diagram which shows the camera provided with the blurring correction apparatus.

Explanation of symbols

2 Camera 3 Angular velocity sensor 4 Shake correction lens 6 VCM
7 Correction Lens Position Detection Unit 20 Lens CPU
DESCRIPTION OF SYMBOLS 21 Support state determination part 22 Follow-up control part 22a 1st calculating part 22b 2nd calculating part 23 Target position conversion part 24 HPF process part 24a Low-pass filter 25 Phase advance compensation part 26 Correction formula calculating part 27 Integration calculating part 28 Speed bias calculating part 28 29 Position Bias Calculation Unit 50 First Table 60 Second Table 100 Blur Correction Device

Claims (5)

A shake detection unit that detects shake of the imaging device;
A blur correction unit that operates to optically correct image blur according to the output of the shake detection unit;
A position detection unit for detecting the position of the blur correction unit;
A support state determination unit that determines whether the support state of the imaging apparatus is handheld or fixed based on the output of the shake detection unit;
A target speed calculation unit that calculates a target speed of the shake correction unit based on the output of the shake detection unit;
Based on the position of the blur correction unit that is calculated from the output or the target speed of the position detection unit, said etc. blur correction unit is Iho far from the movable center of the range, the so provide greater centripetal force to the blur correction unit A speed bias calculation unit that determines the amount of change in the target speed,
The speed bias calculation unit, when the supporting state determination unit determines fixed, than when it is determined that hand, the to the centripetal force rather large in the movable range around the center, the centripetal force at a position distant from the movable range center A blur correction device characterized by reducing the size of the image. A shake detection unit that detects shake of the imaging device;
A blur correction unit that operates to optically correct image blur according to the output of the shake detection unit;
A position detection unit for detecting the position of the blur correction unit;
A support state determination unit that determines whether the support state of the imaging apparatus is handheld or fixed based on the output of the shake detection unit;
A target speed calculation unit that calculates a target speed of the shake correction unit based on the output of the shake detection unit;
A speed bias calculating unit for determining an output or on the basis of the position of the blur correction unit that is calculated from the target speed, before Symbol shake correction unit change amount of the target speed to provide a centripetal force to the position detection unit, With
The speed bias calculation unit increases the amount of change of the target speed as the blur correction unit is farther from the center of the movable range, and when the support state determination unit determines that it is handheld when it is determined that it is fixed. The blur correction device is characterized in that an increase rate of the gradient of the change amount of the target speed is reduced. The blur correction device according to claim 1 or 2 ,
The speed bias calculator is
The relationship between the output of the position detection unit and the change amount of the target speed includes a first table represented by a cubic expression and a second table represented by a linear expression,
The speed bias calculation unit determines the amount of change in the target speed using the first table when the support state determination unit determines that the hand is held.
When the support state determination unit determines that the fixed state is fixed, the amount of change in the target speed is determined using the second table. In the blurring correction apparatus according to any one of claims 1 to 3 ,
A target speed calculator for calculating a target speed for optically correcting image blur using the output of the shake detector;
An amplifier for amplifying the output of the position detector;
A subtractor that subtracts the output of the amplifier from the output of the target speed calculator;
And an integration unit for integrating the output of the subtraction unit. An imaging device comprising the blur correction device according to any one of claims 1 to 4 .

Images