JP7443015B2 - Imaging device, shake correction device, and shake correction method - Google Patents

Description

The present invention relates to an imaging device, a shake correction device, and a shake correction method.

Image shake includes various shake components such as parallel shake and angular shake. Patent Document 1 discloses a technology related to translational shake correction that is installed in an imaging device.

Patent No. 5031690

However, the technique disclosed in Patent Document 1 cannot cope with correction of low-frequency and large parallel shake that accompanies photographing operations such as pressing down the shutter.

SUMMARY OF THE INVENTION An object of the present invention is to provide an imaging device capable of highly accurately correcting even low-frequency and large parallel vibrations associated with photographing operations.

In order to solve the above-mentioned problems, an imaging device of the present invention includes a calculation means for calculating a shake correction signal based on a shake detection signal including a parallel shake component outputted by a detection means for detecting shake of the imaging device; A correction means for correcting shake based on the shake correction signal, and a characteristic changing means for changing the phase characteristic of the calculation means according to the elapse of time during exposure, and the calculation means is configured to detect the shake. The characteristic changing means includes an integral filter that integrates a signal, and the characteristic changing means adjusts the degree to which the phase of the output signal of the integral filter having a frequency corresponding to the parallel shake component is delayed based on the elapsed time from the start of exposure to the initial exposure period. The phase characteristic of the arithmetic means is changed so that it is stronger at the corresponding first time than at a second time, where the elapsed time is longer than the first time.

According to the present invention, it is possible to provide an imaging device that is capable of highly accurately correcting even low-frequency and large parallel vibrations associated with photographing operations.

FIG. 1 is a diagram illustrating the configuration of an imaging device in a first embodiment. FIG. 3 is a Bode diagram showing the vibration suppression rate of the image blur correction system. FIG. 4 is a gain characteristic Bode diagram and a phase characteristic Bode diagram of the first integrating means. FIG. 6 is a diagram illustrating waveforms showing the integration operation of the first integration means. FIG. 4 is a gain characteristic Bode diagram and a phase characteristic Bode diagram of a low-pass filter and a differential filter. FIG. 3 is a gain characteristic Bode diagram and a phase characteristic Bode diagram of a phase compensation filter. FIG. 3 is a phase characteristic Bode diagram of a combination of a first integrating means and a phase compensating means. FIG. 3 is a diagram illustrating a change in characteristics of a phase compensation means. It is a figure explaining the drive distribution means in 1st Embodiment. FIG. 3 is a diagram illustrating image blur correction in the first embodiment. 7 is a flowchart showing image blur correction processing in the first embodiment. FIG. 3 is a diagram illustrating the configuration of an imaging device in a second embodiment. It is a figure explaining the change of a bandpass filter characteristic. It is a figure explaining the drive distribution means in 2nd Embodiment. 7 is a flowchart showing image blur correction processing in the second embodiment. A cross-sectional view and a block diagram of an imaging device in a third embodiment. It is a figure explaining the switching timing of the characteristic of a characteristic change means. It is a figure explaining the drive distribution means in 3rd Embodiment. 12 is a flowchart showing image blur correction processing in a third embodiment. A cross-sectional view and a block diagram of an imaging device.

(First embodiment)
FIG. 1 is a diagram illustrating the configuration of an imaging device in the first embodiment. The camera 11 is an imaging device, such as a digital single-lens reflex camera, a video camera, or a compact digital camera, that has image blur correction means. The camera 11 includes a camera body 11a and an interchangeable lens 11b that can be attached to and detached from the camera body 11a. In this embodiment, an example of an imaging device in which the interchangeable lens 11b is detachable from the camera body 11a will be described, but an imaging device in which the interchangeable lens 11b and the camera body 11a are integrated may be used.

The interchangeable lens 11b includes a photographing optical system 13 and a first driving means 13c. The photographing optical system 13 includes a plurality of lenses such as a shift lens and a zoom lens, and an aperture. A part of the lens in the photographing optical system 13 constitutes a lens shake correction means 13a, which is a first shake correction means. By driving the lens shake correction means 13a in the direction of the arrow 13b by the first driving means 13c, image shift occurring on the surface of the image sensor 14 is reduced.

The camera body 11a includes a CPU (Central Processing Unit) 12, an image sensor 14, a second drive means 14c, a vibration detection means 15, a calculation means 16, and a photographing operation means 17. In FIG. 1, the camera body 11a is provided with a CPU 12, vibration detection means 15, and calculation means 16, but they may be provided on the interchangeable lens 11b side, or may be provided on both the camera body 11a and the interchangeable lens 11b. It's okay.

The CPU 12 realizes the functions of the camera 11 by controlling the operation of each block of the camera 11 in response to a shooting instruction operation from a photographer. The image sensor 14 outputs a signal responsive to the input object light flux. A subject light beam along a photographing optical axis 10 passes through a photographing optical system 13 and enters an image sensor 14 serving as an image pickup means. The image sensor signal output from the image sensor 14 is subjected to image processing in an image processing section (not shown), and the image information after the image processing is recorded in a storage section (not shown).

Further, the image sensor 14 constitutes an image sensor shake correction means 14a, which is a second shake correction means that is driven in the direction shown by an arrow 14b. Then, the image shift generated on the surface of the image sensor 14 is reduced by driving the image sensor shake correcting means 14a. Although FIG. 1 only shows blocks that reduce image shift in the directions of arrows 13b and 14b, blocks that reduce image shift in other directions, such as a direction perpendicular to these directions, are also provided. Good too. Furthermore, although this embodiment describes an example in which image shake is optically corrected, image shake correction in the image sensor 14 may be performed electronically by image cutting out or the like.

The photographing operation means 17 includes a shutter button, and the photographer sends a photographing instruction to the camera 11 by pressing the shutter button. The vibration detection means 15 includes an angular velocity meter 15a and an accelerometer 15b. The calculation means 16 includes a first integration means 16a, a first amplitude correction means 16b, a second integration means 16c, a bandpass filter 16d, a bandpass filter 16e, a comparison means 16f, a phase compensation means 16g, a second amplitude correction means 16h, A drive distribution means 16i is provided.

Each shake detection signal detected by the angular velocity meter 15a and the accelerometer 15b is output to the calculation means 16. Specifically, the angular velocity signal, which is the shake detection signal of the angular velocity meter 15a, is input to the first integrating means 16a. The first integrating means 16a integrates the angular velocity signal of the angular velocity meter 15a and converts it into an angle signal. The gain of the converted angle signal is adjusted by the first amplitude correction means 16b depending on the sensitivity of the photographing optical system 13. The shake correction signal from the first amplitude correction means 16b is inputted to the first drive means 13c and the second drive means 14c via the drive distribution means 16i, and drives the lens shake correction means 13a and the image pickup element shake correction means 14a. This makes it possible to reduce image plane deviation caused by angular shake applied to the camera.

The shake detection signal from the accelerometer 15b is input to the second integrating means 16c. The second integrating means 16c converts the vibration detection signal of the accelerometer 15b into a velocity. The bandpass filter 16d extracts only some frequency components, such as a 2Hz frequency component, from the converted speed signal. The bandpass filter 16e extracts only some frequency components, such as a 2 Hz frequency component, from the angular velocity signal detected by the angular velocity meter 15a. The comparison means 16f calculates a rotation radius indicating the relationship between the velocity signal of the bandpass filter 16d and the angular velocity signal of the bandpass filter 16e. Note that the details of the band-pass filter 16d and the band-pass filter 16e are described in Patent Document 1, so a specific explanation will be omitted.

The signal from the first integrating means 16a is multiplied by the signal from the comparing means 16f in the second amplitude correcting means 16h via a phase compensating means 16g which is a characteristic changing means to be described later. The phase compensation means 16g is provided to enhance the effect of reducing image plane deviation caused by low-frequency parallel shake. Since the comparing means 16f calculates the angular velocity and the radius of rotation of the velocity, the second amplitude correcting means 16h multiplies the angle signal of the first integrating means 16a by the radius of rotation to convert the angle signal into a displacement signal of parallel runout. do.

The shake correction signal from the second amplitude correction means 16h is input to the first drive means 13c and the second drive means 14c via the drive distribution means 16i. The first drive means 13c and the second drive means 14c drive the lens shake correction means 13a and the image sensor shake correction means 14a according to the signal input from the drive distribution means 16i, and correct image blur caused by parallel shake. Reduce. With the above configuration, an image blur correction system that can reduce image plane shift due to parallel shake in addition to angular shake is configured.

FIG. 2 is a diagram showing the vibration suppression rate of the image blur correction system. In FIG. 2, the horizontal axis shows frequency and the vertical axis shows vibration suppression rate. On the vertical axis, the vibration damping rate increases as it goes further down the page, indicating higher vibration damping performance. In this embodiment, the angular vibration damping characteristic shown by the broken line 21 is set to a characteristic in which the damping rate becomes large near the frequency ω1 (broken line 22). The parallel vibration isolation characteristic shown by the solid line 23 is set to a characteristic in which the vibration suppression rate becomes large near the frequency ω2 (broken line 24), which is a lower frequency than the frequency ω1. The reason why the vibration suppression rate characteristics are changed between angular runout and parallel runout in this way is because it is known that parallel runout causes a larger runout in the lower frequency range than angular runout.

Next, the reason why the phase compensation means 16g is provided to increase the vibration suppression rate in the low frequency region of parallel vibration will be explained. First, the reason why the detection accuracy of low frequency angular shake and low frequency parallel shake is low in the image blur correction system will be explained using FIGS. 3 and 4. FIG. 3(A) is a Bode diagram of the gain characteristics of the first integrating means. FIG. 3(B) is a Bode diagram of the phase characteristics of the first integrating means.

In FIG. 3A, the horizontal axis represents the vibration frequency, and the vertical axis represents the output gain of the first integrating means 16a with respect to the input of the vibration detecting means 15. In FIG. 3A, a line segment 31 represents the gain characteristic of the first integrating means 16a, and a signal with a frequency higher than the folding point frequency ω3 (broken line 32), which is the integration start frequency, has the characteristic of first-order integration (frequency (Inversely proportional gain). The angular velocity signal of the vibration detection means 15 is first-order integrated by the first integration means 16a, and signals having a frequency of ω3 or higher are converted into angle signals. Here, the breaking point frequency ω3 is set to a frequency lower than the low frequency side of the blur band 33 to be detected. For example, if the vibration to be detected is distributed from 1 Hz to 10 Hz, the corner frequency ω3 is set to 0.1 Hz.

In FIG. 3(B), the horizontal axis represents the vibration frequency, and the vertical axis represents the phase difference of the first integrating means 16a with respect to the input of the vibration detecting means 15. In FIG. 3(B), a line segment 34 represents the phase characteristic of the first integrating means 16a, and the phase of a signal higher than the breaking point frequency ω3 (broken line 32) approaches -90 degrees as the frequency increases. Becomes a characteristic.

FIG. 4 is a diagram illustrating waveforms showing the integrating operation of the first integrating means 16a. In FIGS. 4(A) and 4(B), the horizontal axis represents time, and the vertical axis represents the amount of shake. FIG. 4(A) shows an example at the frequency indicated by the broken line 35 in FIG. 3(B), which is a higher frequency than the breaking point frequency ω3. On the other hand, FIG. 4(B) shows an example at the frequency of the broken line 36 in FIG. 3(B), which is near the breaking point frequency ω3. Note that the frequency indicated by the broken line 36 is also the low frequency limit in the blur band 33.

In FIG. 4A, a shake angle waveform 41 represents the angle of shake applied to the imaging device, an angular velocity waveform 42 represents the angular velocity detected by the vibration detecting means 15, and an integral signal obtained by processing the signal of the vibration detecting means 15 by the first integrating means 16a. is shown by an angle waveform 43. In FIG. 4(A), for deflection at a frequency higher than the breaking point frequency ω3 (frequency indicated by the broken line 35), the angle waveform 43 obtained by integrating the angular velocity waveform 42 matches the actual deflection angle waveform 41. Therefore, the angle waveform 43 can accurately detect the actual deflection angle.

In FIG. 4B, a shake angle waveform 44 represents the angle of shake applied to the imaging device, an angular velocity waveform 45 represents the angular velocity detected by the vibration detecting means 15, and an integral signal obtained by processing the signal of the vibration detecting means 15 by the first integrating means 16a. is shown by an angle waveform 46. In FIG. 4(B), when the deflection is near the break point frequency ω3 (the frequency indicated by the broken line 36), the angle waveform 46 obtained by integrating the deflection angular velocity waveform 45 has an ideal integration delay characteristic of −90 degrees with respect to the actual deflection angle waveform 44. A phase advance of θ1 (phase difference 37 in FIG. 3(B)) occurs. Therefore, it does not match the shake angle waveform 44 applied to the camera 11, and accurate shake angle detection cannot be performed. In this way, among the vibration waveforms applied to the camera 11, the low-frequency components are affected by the first integrating means 16a, and the detected vibration phase advances, making it impossible to obtain a large vibration suppression rate (high vibration isolation performance). The phase compensation means 16g is provided to delay the phase advance of low frequencies.

The phase compensation means 16g is composed of a known low-pass filter or differential filter. The characteristics of each filter will be explained using the Bode diagram shown in FIG. FIG. 5(A) is a Bode diagram of the gain characteristic of the low-pass filter. FIG. 5(B) is a Bode diagram of the phase characteristics of the low-pass filter. FIG. 5C is a Bode diagram of the gain characteristics of the differential filter. FIG. 5(D) is a Bode diagram of the phase characteristics of the differential filter.

A line segment 51 in FIG. 5A is the gain characteristic of the low-pass filter, and signals at frequencies higher than the bending point frequency ω4 (broken line 52) are attenuated. A line segment 53 in FIG. 5(B) is the phase characteristic of the low-pass filter, and the phase of a signal higher than the breaking point frequency ω4 (broken line 52) lags as the frequency increases.

A line segment 54 in FIG. 5(C) is the gain characteristic of the differential filter, and signals at frequencies higher than the bending point frequency ω5 (broken line 55) are amplified. A line segment 56 in FIG. 5(D) is the phase characteristic of the differential filter, and the phase of a signal higher than the breaking point frequency ω5 (broken line 55) advances as the frequency increases.

FIG. 6 is a Bode diagram of a phase compensation filter that is a combination of a low-pass filter and a differential filter. FIG. 6(A) is a Bode diagram of the gain characteristics of the phase compensation filter. FIG. 6(B) is a Bode diagram of the phase characteristics of the phase compensation filter. In FIG. 6A, a low-pass filter and a differential filter are combined to form a phase compensation filter having a gain characteristic shown by a thick solid line 61 having a center frequency ω6 (broken line 62).

A thick solid line 64 in FIG. 6(B) is the phase characteristic of the phase compensation filter shown in FIG. 6(A). In the phase characteristic of the phase compensation filter shown by the solid line 64, the phase delay is maximum at the center frequency ω6 (broken line 62). In this embodiment, a phase compensation filter that is a combination of a low-pass filter and a differential filter is used as the phase compensation means 16g in FIG.

FIG. 7 is a Bode diagram of the phase characteristics of the first integrating means 16a and the phase compensating means 16g. A line segment 71 indicates the phase characteristic of the first integrating means 16a, and a line segment 64 indicates the phase characteristic of the phase compensation means 16g. As described above, the phase characteristic (line segment 71) of the first integrating means 16a has a phase leading at low frequencies, which causes a decrease in the vibration suppression rate. Therefore, when the phase compensating means 16g is connected to the first integrating means 16a, the phase characteristic (line segment 64) of the phase compensating means 16g is added to the phase characteristic (line segment 71) of the first integrating means 16a. A line segment 72 indicates a phase characteristic obtained by adding the phase characteristic (line segment 64) of the phase compensation means 16g to the phase characteristic (line segment 71) of the first integrating means 16a. As shown by line segment 72, as the phase characteristic approaches phase zero (0deg), the shake input to the camera 11 and the shake detected by the gyro meter 15a are aligned in phase, making it possible to perform shake correction with a high vibration suppression rate. become.

Returning to the explanation of FIG. 2. The reason why the parallel vibration suppression rate indicated by the line segment 23 is high at the low frequency ω2 (broken line 24) is because the above-mentioned phase compensation means 16g is provided. However, the vibration suppression rate for parallel shake is lower than the vibration suppression rate for angular shake shown by line segment 21 at frequency ω1 (broken line 22). This is because the amplitude of the detected shake becomes smaller due to the difference G (arrow 63) in the gain characteristics of the phase compensation means 16g shown in FIG. 6(A), and it no longer matches the amplitude of the input shake. Therefore, when the phase compensating means 16g is connected to the first integrating means 16a, a high vibration suppression rate is exhibited only when low frequency vibration occurs.

Here, the characteristics of parallel runout will be explained. When a photographer presses the shutter button of the photographing operation means 17 at the start of still image exposure, a large parallel shake of low frequency occurs due to the force generated in the direction of the arrow 17a in FIG. Therefore, during the exposure period, the period during which large low-frequency parallel shake occurs is limited to a certain period from the start of exposure. Therefore, in this embodiment, the phase compensation means 16g is provided only during the period when a large low-frequency parallel shake is occurring during the exposure period, and after the large low-frequency parallel shake has subsided, the characteristics of the phase compensation means 16g are weakened. This increases the high-frequency parallel vibration suppression rate. By changing the characteristics of the phase compensation means 16g during the exposure period, a high vibration suppression rate can be obtained during the entire exposure period.

The details of changing the characteristics of the phase compensation means 16g will be explained. The characteristics of the phase compensating means 16g are changed by the CPU 12, which is a characteristics changing means. In this embodiment, the phase compensation means 16g is used with the characteristics shown in FIGS. 6(A) and 6(B) at the initial stage of exposure. Then, as time passes, the break point frequency ω5 (broken line 55) is brought closer to the break point frequency ω4 (broken line 52), and eventually the frequency ω5 and the frequency ω4 are made to match. This eliminates the characteristics of the phase compensation means 16g.

FIG. 8 is a diagram illustrating changes in the characteristics of the phase compensation means 16g. FIG. 8(A) is a diagram illustrating the timing of changing the characteristics of the phase compensation means 16g. FIGS. 8(B) to 8(D) are Bode diagrams of the phase compensating means 16g at each timing. In FIG. 8(A), the horizontal axis shows the change in exposure time, and the vertical axis shows the relationship between frequency ω4 and frequency ω5 in FIG. 6. S2 on the horizontal axis indicates the timing when the shutter button is pressed (fully pressed), that is, the exposure start time.

A line segment 81 indicates the bending point frequency ω5 of the differential filter. As shown by a line segment 81, at the initial stage of exposure, the frequency ω4 and the frequency ω5 are far apart. As shown by a line 64 in FIG. 8B, which is a boat diagram at the initial stage of exposure, the phase compensating means 16g has a characteristic of large phase delay. Then, the CPU 12 performs control to bring the frequency ω5 closer to the frequency ω4 as time passes from the initial stage of exposure. Due to the change in the characteristics of the phase compensation means 16g, the characteristics were changed to a smaller phase delay, as shown in FIG. 8(C), which is a boat diagram at the middle of exposure, and FIG. 8(D), which is a boat diagram at the end of exposure. be done. In this manner, in this embodiment, the plurality of frequency bending points (the bending point frequency ω4 of the low-pass filter and the bending point frequency ω5 of the differential filter) of the phase compensating means 16g are brought closer to each other as the exposure time passes. As a result, the phase compensation effect of the phase compensation means 16g at low frequencies becomes small, but the difference G in the gain of the phase compensation means 16g also becomes small, so that the vibration suppression rate in the high frequency band becomes high.

Next, the shake correction stroke will be explained. In order to increase the vibration suppression rate, it is necessary to increase the vibration correction stroke of the vibration correction means in addition to the frequency characteristics. Since the image blur on the imaging surface due to parallel shake increases as the photographing magnification increases, the shake correction stroke of the shake correction means in close-up photography becomes extremely large. In this embodiment, large strokes are handled by driving two shake correction means, the lens shake correction means 13a and the image sensor shake correction means 14a. Specifically, in this embodiment, parallel shake correction is performed only during exposure, and a parallel shake correction signal is sent from the start of exposure to the image sensor shake correction means 14a, which had not performed shake correction before exposure. Enter.

FIG. 9 is a diagram illustrating the drive distribution means 16i in the first embodiment. The drive distribution means 16i distributes the shake correction signal (the angular shake signal of the first amplitude correction means 16b and the parallel shake signal of the second amplitude correction means 16h) to each drive means. The drive distribution means 16i of this embodiment includes a switch 91. The angular shake signal of the first amplitude correction means 16b is directly connected to the first drive means 13c. On the other hand, the parallel shake signal of the second amplitude correction means 16h is connected to the second drive means 14c via a switch 91 included in the drive distribution means 16i.

The switch 91 is turned on and off as follows depending on the photographer's operation from the CPU 12.
- When the shooting magnification of the subject is small (when the focal length is short, when the shooting distance to the subject is long)
Always off. That is, the image sensor shake correction means 14a does not perform shake correction.
・When the magnification of the subject is high (when the focal length is long, when the shooting distance to the subject is short)
Off until the start of still image shooting (S2). On only during still image exposure.

In this way, the image sensor shake correction means 14a preserves the shake correction stroke by not performing shake correction until before still image exposure, and performs parallel shake correction only during exposure. Since the parallel shake does not become very large until the still image is exposed, there is a low possibility that the shake correction stroke will be insufficient even if the image sensor shake correction means 14a does not perform shake correction.

FIG. 10 is a diagram illustrating image blur correction in the first embodiment. FIG. 10A is a diagram illustrating image shake correction by the lens shake correction means 13a, in which the horizontal axis represents time and the vertical axis represents the lens shake correction means stroke. FIG. 10A is a diagram illustrating image blur correction by the image sensor shake correction means 14a, in which the horizontal axis represents time and the vertical axis represents the stroke of the image sensor shake correction means. Timing 1003a is the timing of half-pressing the shutter button (S1) of the photographing operation means 17, which is a photographing preparation operation. Timing 1003b is the timing at which exposure starts when the shutter button of the photographing operation means 17 is fully pressed (S2) for still image photographing. Timing 1003c is the timing at which exposure ends.

In FIG. 10(A), the lens shake correction means 13a starts correcting the angular shake shown by a waveform 1001 at timing 1003a. Regarding angular shake correction, since the correction stroke range (range 1002a to 1002b) of the lens shake correction means 13a is sufficiently secured, stable shake correction can be performed.

In FIG. 10B, it is assumed that the image sensor shake correction means 14a starts parallel shake correction shown by a broken line waveform 1004 at timing 1003a. When parallel shake correction is started from timing 1003a, the correction stroke range (between 1006a and 1006b) is exceeded between timing 1003b and timing 1003c, which are during the exposure period, and shake correction cannot be performed. This can be avoided by enlarging the correction stroke range, but if the photographing magnification of the subject is large, a large amount of translational shake correction is required, resulting in an extremely large size of the camera body 11a. Therefore, in this embodiment, as shown by a waveform 1005, parallel shake correction by the image sensor shake correction means 14a is not performed until the start of exposure, but is started at zero at the start of still image exposure (timing 1003b). This can prevent the shake correction stroke from being completely consumed during exposure.

The operation of the image blur correction system, which combines the operation of changing the characteristics of the phase compensation means 16g over time during exposure and the driving operation of the image sensor shake correction means 14a of the drive distribution means 16i, will be described using the flowchart of FIG. explain. FIG. 11 is a flowchart showing image blur correction processing in the first embodiment. The flow of FIG. 11 starts with a half-press operation (S1) of the shutter button of the photographing operation means 17, which is a photographing preparation operation in the camera 11, and the flow is circulated as long as this operation continues.

In step S1101, angular shake correction is performed by the lens shake correction means 13a. Specifically, the vibration detection signal detected by the vibration detection means 15 is processed by the calculation means 16, and the drive target value is generated by the first amplitude correction means 16b. Then, the first driving means 13c drives the lens shake correction means 13a based on the generated drive target value to perform angular shake correction.

In step S1102, the CPU 12 determines whether to start still image exposure. When the shutter button of the photographing operation means 17 is fully pressed (S2), the CPU 12 determines that still image exposure has started, and proceeds to step S1103. On the other hand, if full press of the shutter button cannot be detected, the process returns to step S1101.

In step S1103, the CPU 12 determines whether translational shake correction is necessary during the exposure period. The necessity of parallel shake correction is determined, for example, based on the magnitude of the photographing magnification obtained during the photographing operation. If the photographing magnification is larger than the predetermined magnification, it is determined that parallel shake correction is necessary, and the process advances to step S1104. On the other hand, if the parallel shake correction is less than or equal to the predetermined magnification, it is determined that parallel shake correction is not necessary, and the process advances to step S1105.

In step S1104, the image sensor shake correction means 14a is driven to perform translational shake correction. Specifically, the vibration detection signal detected by the vibration detection means 15 is processed by the calculation means 16, and the drive target value is generated by the second amplitude correction means 16h. Then, the CPU 12 turns on the switch 91 of the drive distribution means 16i, so that the second drive means 14c can acquire the drive target value generated by the second amplitude correction means 16h. The second driving means 14c drives the lens shake correction means 13a based on the acquired drive target value to perform angular shake correction.

In step S1105, the CPU 12 starts exposing the still image by an operation such as opening the shutter of the camera.
If it is determined in step S1103 that the photographing magnification is larger than the predetermined magnification and parallel shake correction is being performed, in step S1106 the CPU 12 performs phase compensation over time according to the characteristic change pattern explained in FIG. 8(A). Changing the characteristics of means 16g.

In step S1107, the CPU 12 determines whether the exposure time has ended. Steps S1106 and S1107 are repeated until the exposure time ends. On the other hand, if the exposure time has ended and the exposure is completed, the process advances to step S1108.
In step S1108, the CPU 12 stops the second drive unit 14c from driving the image sensor shake correction unit 14a, and returns to step S1102.

As described above, in this embodiment, the image sensor shake correcting means 14a is driven only during the exposure period to correct parallel shake. Therefore, stable translational shake correction can be performed without exceeding the shake correction stroke during exposure. Further, the phase compensation means 16g changes its characteristics over time during still image exposure. By changing the characteristics during the exposure period, large low-frequency shakes can be corrected with high precision in the early stage of exposure, and high-frequency shakes can be corrected with high precision in the latter half of exposure.

(Second embodiment)
FIG. 12 is a diagram illustrating the configuration of an imaging device in the second embodiment. The difference from the imaging apparatus of the first embodiment (FIG. 1) is that a signal from the CPU 12, which is a characteristic changing means, is input to a pair of bandpass filters 16d and 16e. In the first embodiment as well, the signal from the CPU 12 is input to the calculation means 16, and the CPU 12 changes the time constants of the first integration means 16a and the second integration means 16c, changes the characteristics of the phase compensation means 16g, and changes the characteristics of the first amplitude correction means. 16b and the second amplitude correction means 16h. In addition to these, the CPU 12 of the second embodiment performs control to set the extraction frequency bands of the bandpass filter 16d and the bandpass filter 16e to the low frequency side at the start of still image exposure. In this manner, in this embodiment, the extraction frequency bands of the band-pass filter 16d and the band-pass filter 16e are set to the low frequency side at the start of still image exposure, thereby increasing the accuracy of low-frequency translational shake detection. The rest of the configuration of the imaging device is the same as that of the first embodiment, so a description thereof will be omitted.

FIG. 13 is a diagram illustrating changes in extraction frequency characteristics of a bandpass filter. The bandpass filters in FIG. 13 are a pair of bandpass filters (bandpass filter 16d and bandpass filter 16e). FIG. 13(A) is a diagram illustrating the timing of changing the characteristics of the bandpass filter. FIGS. 13(B) to 13(D) are Bode diagrams of the bandpass filter at each timing.

In FIG. 13(A), the horizontal axis shows the change in exposure time, and the vertical axis shows the extraction frequency. The extraction frequency is ω7 (for example, 0.5 Hz) at the beginning of exposure, and becomes ω6 (for example, 5 Hz) at the later stage of exposure. In FIGS. 13(B) to 13(D), the horizontal axis represents frequency, and the vertical axis represents gain for each frequency. In FIG. 13(B), which is a boat diagram at the initial stage of exposure, the gain is highest at frequency ω7 (broken line 1303), and lower frequencies and higher frequencies are attenuated. Therefore, at the initial stage of exposure, only the frequency component of ω7 is extracted. In FIG. 13(D), which is a boat diagram at the end of exposure, the gain is highest at frequency ω6 (broken line 1304), and lower frequencies and higher frequencies are attenuated. Therefore, only the frequency component of ω6 is extracted at the end of exposure. In FIG. 13(C), which is a boat diagram of the middle period of exposure located between the timings of FIG. 13(B) and FIG. 13(D), the frequency intermediate between FIG. 13(B) and FIG. 13(D) is extracted. It has become a characteristic that

The pair of bandpass filters (bandpass filter 16d and bandpass filter 16e) always have the same characteristics. Therefore, in the shake detection signals of the angular velocity meter 15a and the accelerometer 15b, a low frequency (ω7) component is extracted at the start of still image exposure, and a high frequency (ω6) component is extracted at the later stage of exposure. Therefore, at the start of exposure, parallel shake detection can be performed based on the rotation center of shake determined from the low frequency component of shake, and in the later stages of exposure, parallel shake detection can be performed based on the rotation center of shake determined from the high frequency component of shake. be able to. Therefore, at the start of exposure, it is possible to detect parallel shake at a lower frequency than after that.

Next, the operation of the shake correction means will be explained. FIG. 14 is a diagram illustrating the drive distribution means 16i in the second embodiment. The drive distribution means 16i of this embodiment includes a switch 1401a and a switch 1401b. In this embodiment, the angular shake signal from the first amplitude correction means 16b is input to the first drive means 13c and the second drive means 14c. Based on the acquired angular shake signal, the first drive means 13c drives the lens shake correction means 13a, which is a first shake correction means, and the second drive means 14c drives the image sensor shake correction means, which is a second shake correction means. Drive means 14a. The lens shake correction means 13a and the image sensor shake correction means 14a each correct half of the amount of angular shake, thereby performing sufficient angular shake correction in total.

The parallel shake signal from the second amplitude correction means 16h is input to the first drive means 13c and the second drive means 14c via a switch 1401a and a switch 1401b controlled by the CPU 12. The switch 1401a is a switch provided between the second amplitude correction means 16h and the first drive means 13c, and when the switch 1401a is ON, the parallel shake signal from the second amplitude correction means 16h is transmitted to the first drive means 13c. is input. The switch 1401b is a switch provided between the second amplitude correction means 16h and the second drive means 14c, and when the switch 1401a is ON, the parallel shake signal from the second amplitude correction means 16h is transmitted to the second drive means 14c. is input.

The CPU 12 acquires the shake correction status signal from the lens shake correction means 13a and the shake correction status signal from the image sensor shake correction means 14a, and determines which shake correction means has a margin for shake correction stroke based on the shake correction status signal. Determine if there is. Then, the CPU 12 turns on the switch 1401a and the switch 1401b, which is connected to the shake correction means that has a margin for shake correction stroke at the start of still image exposure. Thereby, the parallel shake signal from the second amplitude correction means 16h is inputted to either the first drive means 13c or the second drive means 14c. In this way, by selecting a shake correction means with a margin for shake correction stroke at the start of still image exposure and performing parallel shake correction, it is possible to suppress the shortage of parallel shake correction stroke during exposure.

FIG. 15 is a flowchart showing image blur correction processing in the second embodiment. The same processing steps as in the first embodiment (FIG. 11) are given the same reference numerals, and the explanation thereof will be omitted. The flow of FIG. 15 starts with a half-press operation (S1) of the shutter button of the photographing operation means 17, which is a photographing preparation operation in the camera 11, and the flow is circulated as long as this operation continues.

In step S1501, angular shake correction is performed by the lens shake correction means 13a and the image sensor shake correction means 14a. Specifically, the vibration detection signal detected by the vibration detection means 15 is processed by the calculation means 16, and the drive target value is generated by the first amplitude correction means 16b. Then, the first drive means 13c and the second drive means 14c drive the lens shake correction means 13a and the image sensor shake correction means 14a based on the generated drive target values, respectively, to perform angular shake correction.

In step S1502, the CPU 12 determines whether there is enough margin in the shake correction stroke of the lens shake correction means 13a at the start of still image exposure. If there is room for the shake correction stroke, the process advances to step S1503. On the other hand, if there is no margin for the shake correction stroke, the process advances to step S1504.

If there is a margin in the shake correction stroke of the lens shake correction means 13a, in step S1503, the lens shake correction means 13a is driven to perform translational shake correction in addition to angular shake correction.
If there is not enough margin in the shake correction stroke of the lens shake correction means 13a, in step S1504, the image sensor shake correction means 13a is driven to perform translational shake correction in addition to angular shake correction.

In step S1505, the CPU 12 changes the characteristics of the phase compensation means 16g, bandpass filter 16d, and bandpass filter 16e as time passes. Specifically, the CPU 12 changes the characteristics of the phase compensating means 16g over time in accordance with the characteristic change pattern explained in FIG. 8(A). Further, the characteristics of the bandpass filter 16d and the bandpass filter 16e are changed over time according to the characteristic change pattern explained in FIG. 13(A). Note that, similarly to the first embodiment, the characteristics may be changed only when the imaging magnification is greater than a predetermined magnification.

In step S1506, the CPU 12 stops only the drive for parallel shake correction by the lens shake correction means 13a or the image sensor shake correction means 14a.

As described above, in this embodiment, when performing still image exposure, a shake correction means with a sufficient drive stroke is selected to perform translational shake correction. Therefore, there is no fear that the shake correction stroke will be insufficient during exposure, and stable translational shake correction is possible. Furthermore, in this embodiment, the characteristic of the extraction frequency is changed from low frequency to high frequency as time passes during still image exposure. As a result, by setting the characteristics suitable for low-frequency translational shake correction in the early stage of still image exposure, the low-frequency component of the shake is extracted and the radius of rotation of the low-frequency shake is determined, and large low-frequency parallel shake can be corrected. It can be corrected with high precision. Then, in the later stages of still image exposure, by making the characteristics suitable for high-frequency translational shake correction, the high-frequency component of the shake can be extracted, the radius of rotation of the high-frequency shake can be determined, and the high-frequency parallel shake can also be corrected.

(Third embodiment)
FIG. 16 is a diagram illustrating the configuration of an imaging device in the third embodiment. The difference between the first embodiment (FIG. 1) and the second embodiment (FIG. 12) is that the calculation means 16 includes a plurality of phase compensation means, a phase compensation selection means 16j, a plurality of bandpass filters, and a plurality of comparison means. , a comparison value selection means 16k is provided.

The plurality of phase compensation means are a first phase compensation means 16g1 and a second phase compensation means 16g2. The first phase compensation means 16g1 has a high phase compensation effect at low frequencies, and has a characteristic that improves the low frequency detection accuracy shown in FIG. 8(B). The second phase compensation means 16g2 has a high phase compensation effect at high frequencies, and has a characteristic that improves the detection accuracy of high frequencies as shown in FIG. 8(D). The phase compensation selection means 16j selects the signals of the first phase compensation means 16g1 and the second phase compensation means 16g2 according to instructions from the CPU 12.

The plurality of bandpass filters are a bandpass filter 16d1, a bandpass filter 16e1, a bandpass filter 16d2, and a bandpass filter 16e2. The bandpass filter 16d1 and the bandpass filter 16e1 (hereinafter also referred to as BPF1) are bandpass filters that extract the low frequency shown in FIG. 13(B). The bandpass filter 16d2 and the bandpass filter 16e2 (hereinafter also referred to collectively as BPF2) are bandpass filters that extract the high frequency shown in FIG. 13(D).

The plural comparison means are a first comparison means 16f1 and a second comparison means 16f2. The comparison value selection means 16k selects the signals of the first comparison means 16f1 and the second comparison means 16f2 according to instructions from the CPU 12.

The phase compensation selection means 16j, the comparison value selection means 16k, and the CPU 12 that controls them are characteristic changing means. The phase compensation selection means 16j changes the characteristics by selecting either the first phase compensation means 16g1 or the second phase compensation means 16g2 based on a signal from the CPU 12. The comparison value selection means 16k switches the characteristics by selecting either the first comparison means 16f1 or the second comparison means 16f2 based on a signal from the CPU 12.

FIG. 17 is a diagram illustrating the characteristic switching timing of the characteristic changing means. A line segment 1701 indicates the selected state of the phase compensation selection means 16j and the comparison value selection means 16k. From the still image exposure start S2 to time t1, the phase compensation selection means 16j selects the first phase compensation means 16g1, and the comparison value selection means 16k selects the first comparison means 16f1. After time t1, the phase compensation selection means 16j selects the second phase compensation means 16g2, and the comparison value selection means 16k selects the second comparison means 16f2.

By switching the above characteristics, in this embodiment as well as in the second embodiment, it is possible to increase the correction accuracy for low frequencies of parallel shake at the start of still image exposure, and thereafter increase the correction accuracy for high frequencies of parallel shake. can. Therefore, it is possible to maintain a high parallel shake correction effect over the entire exposure time. Further, unlike the first and second embodiments in which the characteristics were changed, this embodiment uses a characteristic switching method, so that the amount of calculation during still image exposure can be reduced.

Returning to the explanation of FIG. 16. The signal from the phase compensation selection means 16j is also input to the first amplitude correction means 16b. Therefore, the signal from the first phase compensation means 16g1 is input to the first amplitude correction means 16b at the start of still image exposure, and thereafter the signal from the second phase compensation means 16g2 is input. As a result, with regard to angular shake, it is possible to increase the low frequency correction accuracy at the start of still image exposure, and thereafter to increase the high frequency correction accuracy. Since not only parallel shake but also angular shake occurs when the user operates the photographing operation means 17, effective shake correction can be achieved by increasing the low frequency shake correction accuracy at the start of still image exposure. .

Next, the operation of the shake correction means will be explained. FIG. 18 is a diagram illustrating the drive distribution means 16i in the third embodiment. The drive distribution means 16i of this embodiment includes weighting means 1802a and weighting means 1802b. In this embodiment, the angular shake signal from the first amplitude correction means 16b is input to the first drive means 13c and the second drive means 14c. Based on the acquired angular shake signal, the first drive means 13c drives the lens shake correction means 13a, and the second drive means 14c drives the image sensor shake correction means 14a. The lens shake correction means 13a and the image sensor shake correction means 14a each correct half of the amount of angular shake, thereby performing sufficient angular shake correction in total.

The parallel shake signal from the second amplitude correction means 16h is input to the first drive means 13c and the second drive means 14c via weighting means 1801a and weighting means 1801b controlled by the CPU 12. The weighting means 1801a is provided between the second amplitude correction means 16h and the first driving means 13c, and the weighting means 1801b is provided between the second amplitude correction means 16h and the second driving means 14c.

The CPU 12 controls the respective driving amounts of the lens shake correction means 13a and the image sensor shake correction means 14a for correcting parallel shake by controlling the weighting means 1801a and the weighting means 1801b. Specifically, the CPU 12 calculates the ratio of the shake correction stroke margin of each shake correction means based on the shake correction state signal from the lens shake correction means 13a and the shake correction state signal from the image sensor shake correction means 14a. do. Then, the CPU 12 sets the weighting of the drive amount in each shake correction means based on the ratio of the shake correction stroke margin at the start of still image exposure. The weighting means 1802a and the weighting means 1802b perform weighting processing on the parallel shake signal from the second amplitude correction means 16h according to the set weighting. The weighted signals are input to the first driving means 13c and the second driving means 14c, respectively, from the start of still image exposure.

For example, when the ratio of the shake correction stroke allowances of the lens shake correction means 13a and the image pickup element shake correction means 14a is 6:4, the weighting in the weighting means 1802a is set to A=0.6. The weighting by the weighting means 1802b is 1-A=0.4. As a result, at the start of still image exposure, 60% of the drive amount for parallel shake correction is input to the lens shake correction means 13a, and the remaining 40% is input to the image sensor shake correction means 14a. In this way, in this embodiment, parallel shake correction is performed during exposure by distributing the parallel shake correction amount to each shake correction means based on the shake correction stroke allowance ratio of the plurality of shake correction means at the start of still image exposure. It is possible to prevent the stroke from being insufficient.

FIG. 19 is a flowchart showing image blur correction processing in the third embodiment. The same processing steps as in the first embodiment (FIG. 11) and the second embodiment (FIG. 15) are given the same reference numerals, and their explanations will be omitted. The flow in FIG. 19 starts with a half-press operation (S1) of the shutter button of the photographing operation means 17, which is a preparatory operation for photographing in the camera 11, and the flow continues as long as this operation continues.

In step S1901, the CPU 12 controls the drive distribution means 16i to distribute the parallel shake drive amount input to the first drive means 13c and the second drive means 14c, and the lens shake correction means 13a and the image sensor shake correction means. 14a performs parallel shake correction. Specifically, as explained in FIG. 18, the ratio of the shake correction stroke margin of the lens shake correction means 13a and the image sensor shake correction means is detected at the start of still image exposure, and parallel shake correction is performed according to the ratio. The drive amount for this purpose is distributed to each shake correction means. Each shake correction means performs parallel shake correction according to the distribution of the parallel shake driving amount.

In step S1902, the CPU 12 switches the characteristics of the phase compensation means and the bypass filter as time passes. The phase compensation selection means 16j selects either the first phase compensation means 16g1 or the second phase compensation means 16g2 based on an instruction from the CPU 12, and switches the characteristics. The comparison value selection means 16k selects either BPF1 or BPF2 based on an instruction from the CPU 12 and switches the characteristics. The signal selected by the phase compensation selection means 16j is input to the first amplitude correction means 16b and the second amplitude correction means 16h. The signal selected by the comparison value selection means 16k is input to the second amplitude correction means 16h. The CPU 12 switches the characteristics so that the precision of translational shake correction is high for low frequency at the initial stage of exposure and for high frequency thereafter.

As described above, in this embodiment, when performing still image exposure, the translational shake correction amount is distributed between the lens shake correction means 13a and the image sensor shake correction means 14a according to the ratio of the drive stroke margin, and parallel shake correction is performed. . As a result, it is possible to suppress insufficient shake correction strokes during exposure and perform stable translational shake correction.

Furthermore, in this embodiment, the characteristics of the angular shake signal and the parallel shake signal are switched as time passes during still image exposure. Unlike the first and second embodiments in which the characteristics were changed, this embodiment adopts a characteristic switching method, so that calculations during still image exposure can be lightened. Further, unlike the first and second embodiments in which only the parallel shake characteristics were changed, in this embodiment, the angular shake characteristics are also changed during exposure. Thereby, it is possible to accurately correct low-frequency angular shake that occurs in the first half of still image exposure, and it is also possible to suppress deterioration in shake correction accuracy in the second half of still image exposure.

In the first to third embodiments, two shake correction means, the lens shake correction means 13a and the image sensor shake correction means 14a, are provided, but it is possible to change the characteristics of the image shake correction system during still image exposure. Regarding this, it is not necessary to have a plurality of shake correction means. For example, the present invention is applicable to a case where only the image sensor shake correction means 14a is provided as the shake correction means. FIG. 20 is a diagram illustrating the configuration of an imaging device in a modification of the third embodiment.

In FIG. 20, only the image sensor shake correction means 14a is provided as the shake correction means. The image sensor shake correction means 14a acquires the angular shake signal from the first amplitude correction means 16b via the drive distribution means 16i, and performs angular shake correction from the time of photographing preparation operation. Further, the image pickup element shake correction means 14a acquires a parallel shake signal from the second amplitude correction means 16h via the drive distribution means 16i, and performs parallel shake correction from the start of still image exposure. That is, during the start of still image exposure, the image sensor shake correction means 14a corrects parallel shake in addition to angular shake. Note that the present invention is also applicable to an image stabilization system in which only the lens shake correction means 13a is provided as a shake correction means.

Furthermore, the contents described in the first to third embodiments can be combined as appropriate. For example, changing the frequency corner of the phase compensation means described in the first embodiment may be combined with the drive distribution means described in the second embodiment.

(Other embodiments)
The present invention provides a system or device with a program that implements one or more of the functions of the embodiments described above via a network or a storage medium, and one or more processors in the computer of the system or device reads and executes the program. This can also be achieved by processing. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.

Although preferred embodiments of the present invention have been described above, the present invention is not limited to these embodiments, and various modifications and changes can be made within the scope of the gist thereof.

12 CPUs
13a Lens shake correction means 14a Image sensor shake correction means
15 vibration detection means 16 calculation means 16g phase compensation means 16i drive distribution means

Claims (9)

a calculation means for calculating a shake correction signal based on a shake detection signal including a parallel shake component outputted by a detection means for detecting shake of the imaging device;
a correction means for correcting shake based on the shake correction signal;
Characteristic changing means for changing the phase characteristic of the calculating means according to the passage of time during exposure,
The calculation means includes an integral filter that integrates the shake detection signal,
The characteristic changing means controls the degree to which the phase of the output signal of the integral filter having a frequency corresponding to the parallel shake component is delayed when the elapsed time from the start of exposure is a first time corresponding to an initial period of exposure. An imaging device characterized in that the phase characteristic of the calculating means is changed so as to be stronger than when the elapsed time is a second time longer than the first time. The calculation means has a phase compensation means including a filter having a plurality of frequency bending points,
2. The imaging apparatus according to claim 1, wherein the characteristic changing means changes the characteristic of the calculating means by bringing the plurality of frequency bending points closer together as time passes during exposure. The imaging device according to claim 2, wherein the phase compensation means includes a low-pass filter and a differential filter. The calculation means includes a bandpass filter that extracts a predetermined frequency component from the shake detection signal,
Imaging according to any one of claims 1 to 3, wherein the characteristic changing means changes the extraction frequency of the bandpass filter from a low frequency to a high frequency in accordance with the passage of time during exposure. Device. The calculation means includes a plurality of phase compensation means and a plurality of bandpass filters,
The characteristic changing means changes the phase of the output signal of the integral filter having a frequency corresponding to the parallel shake component from the first phase compensating means to the first phase compensating means according to the passage of time during exposure. 2. The imaging apparatus according to claim 1, wherein the second phase compensation means is switched to a second phase compensation means which has a weaker degree of delay, and the bandpass filter for extracting low frequencies is switched to the bandpass filter for extracting high frequencies. The shake correction signal is a parallel shake signal corresponding to the parallel shake component and an angular shake signal corresponding to the angular shake component,
6. The imaging apparatus according to claim 1, wherein the correction means performs shake correction based on the parallel shake signal only during exposure. The correction means includes a first shake correction means and a second shake correction means,
6. The correcting means is characterized in that the first shake correcting means performs shake correction based on the angular shake signal, and the second shake correcting means performs shake correction based on the parallel shake signal. The imaging device described in . a calculation means for calculating a shake correction signal based on a shake detection signal including a component of parallel shake output by the detection means for detecting shake;
Characteristic changing means for changing the phase characteristic of the calculating means according to the passage of time during exposure,
The calculation means includes an integral filter that integrates the shake detection signal,
The characteristic changing means controls the degree to which the phase of the output signal of the integral filter having a frequency corresponding to the parallel shake component is delayed when the elapsed time from the start of exposure is a first time corresponding to an initial period of exposure. A shake correction device characterized in that the phase characteristic of the calculating means is changed so as to be stronger than when the elapsed time is a second time longer than the first time. A shake correction method,
a detection step of detecting shake of the imaging device and outputting a shake detection signal;
a calculation step of calculating a shake correction signal based on an output signal obtained by integrating a shake detection signal including a component of parallel shake output in the detection step with an integral filter;
The degree to which the phase of the output signal of the integral filter of the frequency corresponding to the parallel shake component is delayed is determined by determining the degree of delay of the phase of the output signal of the integral filter having a frequency corresponding to the parallel shake component . Characteristic changing means for changing the phase characteristic in the calculation step so as to make it stronger than during a second time longer than the first time;
A shake correction method comprising the step of correcting shake based on the shake correction signal.

Images