JP2016206656A - Vibration-proof control device, optical instrument, and vibration-proof control program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform excellent vibration-proof control, while using a vibration sensor, during imaging by which a motion vector is not obtained.SOLUTION: A vibration-proof control device 30 performs vibration-proof control by using a vibration detection signal and a motion vector detection signal. The device includes vibration extraction means 66 for extracting a first vibration signal which is a frequency component higher than a predetermined frequency from the vibration detection signal, prediction means 68 for generating a prediction vibration signal indicating the prediction value of a vibration by using the motion vector detection signal, synthesis means 69 for synthesizing the first vibration signal and the prediction vibration signal and generating a second vibration signal, and control means 71 for performing the vibration-proof control by using the second vibration signal. The vibration-proof control using the second vibration signal is performed in the imaging of a still image and the prediction vibration signal is generated by using a specific motion vector signal.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an image stabilization control device that reduces image blur in an optical apparatus such as a digital camera or an interchangeable lens.

  Shifting lenses and image sensors (CMOS sensors, etc.) relative to the photographic optical axis as an anti-vibration technology that corrects (reduces) image blur caused by camera shake caused by user shake when taking still images with a digital camera There is optical anti-vibration to make. The camera shake uses a shake sensor such as an angular velocity sensor (gyro) or calculates a motion vector between frame images constituting a moving image as disclosed in Patent Document 1. Is detected.

Japanese Patent Laid-Open No. 05-122590

  However, when a shake sensor is used, the camera shake detection accuracy is lowered by superimposing low-frequency noise (error component) called drift on the sensor output corresponding to the original camera shake. There is a risk that vibration control cannot be performed. On the other hand, when detecting a motion vector, camera shake can be detected with high accuracy, but the motion vector cannot be detected during the exposure time of the image sensor for capturing a still image. It is impossible to perform the image stabilization control using.

  The present invention provides an image stabilization control apparatus and an optical apparatus equipped with the image stabilization control apparatus that can perform satisfactory image stabilization control while using a shake sensor during still image capturing, in which a motion vector cannot be obtained.

  An image stabilization control apparatus according to one aspect of the present invention is a motion vector detection signal indicating a motion vector detected in a shake detection signal obtained by detecting shake of an optical device and a video signal generated by imaging using the optical device. Are used to perform image stabilization control for reducing image blur due to the shake of the optical device. The anti-vibration control device uses a shake extraction unit that extracts a first shake signal that is a frequency component higher than a predetermined frequency in the shake detection signal, and a predicted shake signal that indicates a predicted value of the shake using the motion vector detection signal. Predicting means for generating the first shake signal and the predicted shake signal to synthesize the first shake signal and generating the second shake signal, and the control means for performing the image stabilization control using the second shake signal. Have. Further, the control means performs image stabilization control using the second shake signal in capturing a still image, and the prediction means generates a predicted shake signal using the specific motion vector signal.

  According to another aspect of the present invention, an image stabilization control apparatus includes a motion detection signal obtained by detecting a shake of an optical device and a motion detected in a video signal generated by imaging using the optical device and an image sensor. Using the motion vector detection signal indicating the vector, the image stabilization control for reducing the image blur due to the shake of the optical device is performed. The anti-vibration control device uses a shake extraction unit that extracts a first shake signal that is a frequency component higher than a predetermined frequency in the shake detection signal, and a predicted shake signal that indicates a predicted value of the shake using the motion vector detection signal. Predicting means for generating the first shake signal and the predicted shake signal to synthesize the first shake signal to generate the second shake signal, and the control means for performing the image stabilization control using the second shake signal. Have And it has the frequency change means which changes a predetermined frequency according to progress of the exposure time of the image pick-up element in imaging, or according to the length of exposure time, It is characterized by the above-mentioned.

  In addition, the optical apparatus which has the said image stabilization control apparatus comprises the other one side of this invention.

  Further, an image stabilization control program as a computer program that causes a computer to operate as the image stabilization control apparatus constitutes another aspect of the present invention.

  According to the present invention, the second shake signal obtained by synthesizing the first shake signal containing the error component and not including the low frequency component and the predicted shake signal predicted using the motion vector detection signal is used. Thus, good image stabilization control can be performed even during still image capturing in which a motion vector cannot be obtained.

1 is a block diagram illustrating a configuration of an image stabilization control apparatus that is Embodiment 1 of the present invention. FIG. 2 is a block diagram illustrating a configuration of an imaging apparatus equipped with the image stabilization control apparatus according to the first embodiment. FIG. 3 is a diagram illustrating a shake signal according to the first embodiment. FIG. 3 is a frequency characteristic diagram showing a filter cutoff value in the first embodiment. FIG. 3 is a block diagram illustrating a configuration of a shake prediction unit according to the first embodiment. FIG. 3 is a block diagram illustrating a prediction process in a shake prediction unit using an adaptive algorithm according to the first embodiment. FIG. 6 is a diagram illustrating a modification of the first embodiment. The block diagram which shows the structure of the image stabilization control apparatus which is Example 2 of this invention. The block diagram which shows the structure of the image stabilization control apparatus which is Example 3 of this invention. 5 is a flowchart illustrating an image stabilization control process for executing the process of the first embodiment. 6 is a flowchart showing another image stabilization control process according to the first embodiment. 6 is a flowchart showing still another image stabilization control process according to the first embodiment.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 2 shows a configuration of a digital camera 1 as an image pickup apparatus (optical apparatus) including an image stabilization control apparatus that is Embodiment 1 of the present invention. The camera 1 uses an imaging element 11 that photoelectrically converts (images) an optical image formed by the imaging lens 2, the imaging optical system 20 in the imaging lens 2, and an imaging signal output from the imaging element 11. And an image processing unit 12 that generates a still image signal. In addition, the camera 1 includes a memory unit 13 described later, a motion vector detection unit (motion vector detection unit) 14 that detects a motion vector from the video signal, an operation unit 15 that recognizes an operation of a user (photographer), an image And an image display unit 16 for displaying signals. Further, the camera 1 includes a camera control unit 10 that controls the entire camera 1 including the photographing lens 2.

  The motion vector detection unit 14 detects a motion vector indicating the distance (displacement) and direction between corresponding points between two consecutive frame images constituting the video signal, and outputs a motion vector detection signal indicating the detection result To do.

  The photographic optical system 20 in the photographic lens 2 includes a lens group 21 composed of a variable power lens and other lenses, a focus lens 22 that adjusts the focus, and an iris diaphragm 23 that adjusts the amount of light. Further, the photographic optical system 20 moves (shifts) in the direction orthogonal to the optical axis of the photographic optical system 20 to correct (reduce) image blur caused by camera shake, which will be described later. A shift lens 24 as a vibration element is included. The shift lens 24 shown in the figure includes a shift lens, a shift actuator that drives the shift lens in a direction orthogonal to the optical axis, and a driver circuit that drives the shift actuator.

  Further, the photographing lens 2 performs operations of a gyro 26 as a shake sensor (a shake detection unit) that detects a shake of the camera 1 caused by a shake of the camera 1, a focus lens 22, an iris diaphragm 23, and a shift lens 24. And a lens control unit 25 for controlling. The gyro 26 outputs an angular velocity detection signal that is a signal indicating the angular velocity of camera shake.

  The lens control unit 25 controls the operations (AF and AE) of the focus lens 22 and the iris diaphragm 23 while communicating with the camera control unit 10. Further, the lens control unit 25 uses the motion vector detection signal from the motion vector detection unit 14 and the angular velocity detection signal from the gyro 26 to control shift drive of the shift lens 24, that is, control of shift actuator drive. An anti-vibration control unit that performs vibration control is included. This image stabilization control unit corresponds to an image stabilization control device.

  An imaging signal (analog signal) from the imaging device 11 obtained by photoelectrically converting the optical image formed by the photographing optical system 20 is converted into a digital signal by an A / D converter (not shown) in the image processing unit 12. The image processing unit 12 performs various image processing such as white balance processing, gamma correction processing, pixel interpolation processing, and the like on the digital imaging signal in response to a command from the camera control unit 10, and generates an image signal (video signal and still image). Signal). The image signal generated by the image processing unit 12 is recorded in a recording medium such as a semiconductor memory (not shown) in the memory unit 13.

  The camera control unit 10 is configured as a computer including a CPU, an internal memory, and the like. The camera control unit 10 outputs timing signals and commands to the respective units in the camera 1 and the lens control unit 25 in the photographing lens 2 at the time of imaging. In addition, the camera control unit 10 performs an imaging preparation operation (such as AF and AE) according to the first release signal SW1 as an imaging preparation instruction generated when the release switch included in the operation unit 15 is half-pressed by the user. ) Is controlled. Further, a still image capturing operation (still image generation / recording) is controlled in accordance with the second release signal SW2 as a still image capturing instruction generated when the release switch is fully pressed.

  Next, the outline of the image stabilization control at the time of still image capturing will be described with reference to FIG. In the graph shown in the upper part of FIG. 3, the horizontal axis indicates time, and the vertical axis indicates camera shake amount. A thick solid line 31 indicates a motion vector detection signal from the motion vector detection unit 14 and corresponds to an actual camera shake amount regardless of the focal length of the photographing optical system 20 (the size of the subject image on the image sensor 11) or the like. It is shown as a signal corrected so as to have a signal value. A thin solid line 32 is a shake detection signal (hereinafter referred to as a gyro shake detection signal) generated by integrating the angular velocity signal from the gyro 26 and has the same camera shake amount as the camera shake amount indicated by the motion vector detection signal. It shows as a signal produced | generated with respect to it.

  An alternate long and short dash line 33 indicates an error component (hereinafter also referred to as a drift component) corresponding to a drift (noise) generated in the gyro 26, and is included in the gyro shake detection signal 32. That is, the gyro shake detection signal 32 is a signal in which the drift component is superimposed on the shake detection signal corresponding to the original camera shake amount. A signal indicated by a broken line 34 is a predicted shake signal (predicted value of a motion vector) calculated from the motion vector detection signal 31 by a prediction process described later. The predicted shake signal 34 is used together with the gyro shake detection signal in image stabilization control during exposure of the image sensor 11 for still image capturing (hereinafter referred to as still image exposure).

  A timing chart 35 shown at the bottom of FIG. 3 shows the exposure timing of the image sensor 11 in still image capturing. H is during still image exposure of the image sensor 11, and L is during non-exposure of the image sensor 11. Show. The period T1 is an imaging preparation period before still image exposure. In this period, the short-time exposure for generating the frame image described above is repeated a plurality of times at a constant cycle. A motion vector is sequentially detected by the motion vector detection unit 14 between two temporally adjacent frame images of these frame images, and prediction filters used for generating the predicted shake signal 34 are sequentially generated based on the detection result. (Updated). This prediction filter will be described later.

  Note that the short exposure period in the period T1 is set to 120 FPS, 240 FPS, or the like. Further, in the period T1, an imaging preparation operation is performed corresponding to the first release signal SW1 generated by the user operating the release switch of the operation unit 15.

  In the period T2 that starts from the timing 36 in response to the user generating the second release signal SW2 through the operation unit 15, the still image exposure of the image sensor 11 is performed. In the period T2, since one exposure is continuously performed, a motion vector cannot be acquired.

  The period T3 is a period after the end of the still image exposure, and a short period of periodic exposure of the image sensor 11 is performed as in the period T1.

  In FIG. 3, since the motion vector detection signal 31 calculates a motion vector from a video signal generated by imaging, the magnitude of the motion vector substantially matches the actual camera shake amount. On the other hand, since the gyro shake detection signal 32 includes the drift component 33 as described above, there is a deviation from the motion vector detection signal 31.

  Therefore, in this embodiment, the image stabilization control unit of the lens control unit 25 uses the low-frequency component (a specific motion vector signal described later) in the motion vector detection signal in the period T1 to predict a subsequent motion vector. Generate (update) the prediction filter. As described above, since the magnitude of the motion vector substantially matches the actual camera shake amount, the motion vector prediction value can be regarded as the camera shake prediction value. For this reason, also in the present embodiment, a signal indicating a predicted value of a motion vector is referred to as a predicted shake signal.

  Then, when the second release signal SW2 is generated at the timing 36, the image stabilization control unit performs a prediction process for calculating the predicted shake signal 34 in the period T2 in which the still image exposure is performed using the prediction filter.

Thereafter, in period T2, the image stabilization control unit synthesizes the predicted shake signal (low frequency component) and the high frequency component (first shake signal) of the gyro shake detection signal, specifically, adds them to prevent the shake. A shake signal used for shake control (second shake signal: hereinafter referred to as a synthesized shake signal) is generated. The image stabilization control unit that has obtained the combined shake signal in this way performs image stabilization control using the combined shake signal.
As described above, during the still image exposure, the image stabilization control is performed using the combined shake signal from which the low-frequency component including the drift component generated in the gyro 26 is removed, so that it is favorable without being affected by the drift component ( Highly accurate anti-vibration control can be performed.

  Next, the configuration and operation of the above-described image stabilization control unit 30 provided in the lens control unit 25 will be described with reference to FIG. In FIG. 1, the components shown in FIG. 2 are given the same reference numerals as in FIG.

  In the image stabilization control unit 30, 61 is an analog / digital (A / D) converter, which quantizes the analog angular velocity signal from the gyro 26. The sampling frequency of the A / D converter 61 is set to the same value as the frame rate of the video signal used in the motion vector detection unit 14 (that is, the sampling frequency for the image sensor 11). An integrator 62 integrates the digital angular velocity signal from the A / D converter 61 to generate a gyro shake detection signal that is an angular displacement signal.

  Reference numerals 63a and 63b denote band pass filters (BPFs), which extract signals in predetermined frequency bands from the gyro shake detection signal from the integrator 62 and the motion vector detection signal from the motion vector detection unit 14, respectively. A gain comparison unit 64 compares the gain (magnitude) of the gyro shake detection signal that has passed through the BPF 63a and the motion vector detection signal from the motion vector detection unit 14 that has passed through the BPF 63b. A gain correction unit 65 corrects the gain of the gyro shake detection signal from the integrator 62 in accordance with the gain comparison result in the gain comparison unit 64.

  The gyro shake detection signal is a signal indicating angular displacement, whereas the motion vector detection signal is a signal indicating displacement of the subject image on the image sensor 11. For this reason, even if the camera 1 shakes in the same way, the camera shake amount indicated by the gyro shake detection signal and the motion vector detection signal differs depending on the focal length and subject distance of the photographing optical system 20. Therefore, the gain comparison unit 64 has a gyro shake detection signal and a motion vector detection signal in which extremely low frequency signal components containing a large amount of high-frequency shake components and drift components that are not so much included in camera shake by the BPFs 63a and 63b are cut. Compare Then, the gain correction unit 65 corrects the gain of the gyro shake detection signal so as to be the same level as the gain of the motion vector detection signal in accordance with the gain comparison result.

  Reference numeral 66 denotes a high-pass filter (HFP) as shake extraction means. The HFP 66 passes a frequency component (first shake signal: hereinafter referred to as a high-frequency gyro shake signal) higher than a cutoff frequency as a predetermined frequency among the gyro shake detection signals whose gains have been corrected by the gain correction unit 65. Extract this. 67 is a low pass filter (LPF) as a motion vector extracting means. The LPF 67 extracts a motion vector detection signal from the motion vector detection unit 14 by passing a specific motion vector signal (hereinafter referred to as a low frequency motion vector signal) that is a frequency component lower than the cut-off frequency. .

  Reference numeral 68 denotes a shake prediction unit (prediction means). The shake prediction unit 68 generates (updates) a prediction filter used for prediction processing using the low-frequency motion vector signal that has passed through the LPF 67 before the still image exposure. Specifically, an adaptive process (adaptive operation) for sequentially updating the filter coefficients of the prediction filter is performed. Then, during still image exposure, a prediction process for calculating a predicted shake signal using a prediction filter is performed.

  An adder (synthesizing unit) 69 adds the high-frequency gyro shake signal that has passed through the HFP 66 and the predicted shake signal (low frequency component) from the shake prediction unit 68 to generate a synthesized shake signal. Reference numeral 70 denotes an output changeover switch which is turned off before still image exposure and turned on during still image exposure, and sends a combined shake signal from the adder 69 to the shift control unit 71. Execution / non-execution of prediction processing of a predicted shake signal in the shake prediction unit 68 and on / off switching of the output changeover switch 70 are performed according to a command from the camera control unit 10 shown in FIG.

  The shift control unit (control means) 71 shift-drives the shift lens 24 to the target shift position (by the target shift drive amount) according to the combined shake signal input via the output changeover switch 70. The shift position of the shift lens 24 is detected by the position detection unit 27. The detection result of the shift position by the position detection unit 27 (hereinafter referred to as detection shift position) is fed back to the shift control unit 71, and the shift control unit 71 performs feedback control of shift driving so that the detected shift position matches the target shift position. Do.

  In this embodiment, the cutoff frequencies of the HPF 66 and the LPF 67 are set to the same frequency. FIG. 4 shows cut-off frequencies of HPF 66 and LPF 67 and their frequency characteristics. FIG. 4 also shows the frequency of the drift component included in the gyro shake detection signal. In FIG. 4, the horizontal axis indicates the frequency, and the vertical axis indicates the gain of each frequency component.

  A thick solid line 80 is the frequency characteristic of the HPF 66, and a broken line 81 is the frequency characteristic of the LPF 67. A frequency 82 is a cutoff frequency fc of the HPF 66 and the LPF 67. A hatched range 83 indicates the frequency range of the drift component. The gyro shake detection signal includes a drift component in the frequency range 83, but most of the drift component is blocked by the HPF 66 in which a cutoff frequency 82 higher than the upper limit frequency of the frequency range 83 is set. In other words, the high frequency gyro shake signal that has passed through the HPF 66 contains almost no drift component.

  On the other hand, a frequency component higher than the cut-off frequency 82 of the LPF 67 in the predicted shake signal generated using the motion vector detection signal from the motion vector detection unit 14 is blocked by the LPF 67.

  The combined shake signal obtained by adding the high-frequency gyro shake signal that has passed through the HPF 66 and the LPF 67 and the predicted shake signal, which is a low-frequency component, has the drift component removed and has a flat gain characteristic with respect to the frequency. . By using such a combined shake signal, it is possible to perform the image stabilization control with good controllability without being affected by the drift component.

  Next, the configuration and operation of the shake prediction unit 68 will be described with reference to FIG. In FIG. 5, 51 is a linear prediction unit as a prediction filter. 52 is an adaptation unit, and 53 is a subtractor. Reference numeral 42 denotes two unit delay devices, 43a denotes a prediction input changeover switch, and 43b and 43c denote contacts of the prediction input changeover switch 43a. 45 is an adaptive processing switch, 46a is an output changeover switch, and 46b and 46c are contacts of the output changeover switch 46a. The prediction input changeover switch 43a, the output changeover switch 46a, and the adaptive processing switch 45 are switched in response to a command from the camera control unit 10.

  41 is an input terminal to which the motion vector detection signal from the motion vector detection unit 14 shown in FIG. 1 is input to the shake prediction unit 68, and 48 is an output terminal that outputs a predicted shake signal from the shake prediction unit 68. Yes.

  u (n) represents a motion vector detection signal (hereinafter also referred to as an observation value) input from the motion vector detection unit 14 to the input terminal 41, and y (n) represents a predicted shake signal (hereinafter referred to as an observation value) from the linear prediction unit 51. Also referred to as a predicted value). e (n) indicates an error that occurs in the linear prediction processing in the linear prediction unit 51. This error e (n) is an error of the linear prediction process itself different from an error such as a drift component generated in the gyro 26, and is a difference value between the observed value u (n) and the predicted value y (n). is there. In the following description, this error e (n) is referred to as a prediction processing error. N in the parentheses indicates the n-th sunling value.

  First, the operation of the shake prediction unit 68 before still image exposure will be described. Before the still image exposure, the prediction input switch 43a is set to the contact 43b side, the adaptive processing switch 45 is set to ON, and the output switch 46a is set to the contact 46c side.

  At this time, the output of the input changeover switch 43a becomes u (n-1) obtained by delaying the observed value u (n) from the input terminal 41 by one unit delay unit 42, and this u (n-1) is Input to the linear prediction unit 51. The linear prediction unit 51 outputs a predicted value y (n) according to the input u (n−1). That is, the linear prediction unit 51 generates the current predicted value y (n) based on the observed value u (n−1) at the time of sampling one unit before the nth (current). In the present embodiment, the process of obtaining the current predicted value from the observed value at the time of sampling one unit or more before is called the prediction process. Details of the operation of the linear prediction unit 51 will be described later.

  The subtractor 53 calculates a prediction processing error e (n) = u (n) −y (n) that is a difference between the current observed value u (n) and the predicted value y (n). The adaptation unit 52 uses the prediction processing error e (n) to update the filter coefficient of the linear prediction unit 51 (prediction filter) using an appropriate adaptation algorithm. In the present embodiment, the operation for appropriate updating of the linear prediction unit 51 performed by the adaptation unit 52 is referred to as adaptation processing. Details of the operation of the adaptation unit 52 will be described later. The observed value u (n) is output to the output terminal 48 via the output changeover switch 46a.

  As described above, in a state in which a motion vector detection signal is obtained from the motion vector detection unit 14 before still image exposure, the shake prediction unit 68 outputs the motion vector detection signal input to the input terminal 41 to the output terminal 48 as it is. To do. On the other hand, the adaptation unit 52 is caused to perform an adaptation process for the linear prediction unit 51.

  On the other hand, the operation of the shake predicting unit 68 in a state in which a motion vector detection signal cannot be obtained during still image exposure, that is, an observed value u (n) cannot be obtained will be described. During still image exposure, the prediction unit input changeover switch 43a is set to the contact 43c side, the adaptive processing switch 45 is set to OFF, and the output changeover switch 46a is set to the contact 46b side.

  At this time, the output from the prediction unit input changeover switch 43a becomes the previous prediction value y (n-1) from the other unit delay unit 42, and this y (n-1) is the linear prediction unit 51. Is input. The linear prediction unit 51 outputs the current predicted value y (n) according to the previous predicted value y (n−1) that has been input. Since the adaptive processing switch 45 is off, the adaptive unit 52 and the subtractor 53 are stopped. The predicted value y (n) is output to the output terminal 48 via the output changeover switch 46a.

In this way, during still image exposure, the shake prediction unit 68 outputs the predicted value y (n) generated by the linear prediction unit 51 to the output terminal 48 and stops the operation of the adaptation unit 52.
In the present embodiment, the lens control unit 25 needs to know in advance whether or not still image exposure is in progress. Therefore, the camera control unit 10 transmits a signal (notification) indicating that still image exposure is being performed to the lens control unit 25 in response to the input of the second release signal SW2 from the operation unit 15.

  Next, detailed configurations and operations of the linear prediction unit 51 and the adaptation unit 52 will be described with reference to FIG. FIG. 6A shows these operations before still image exposure, and FIG. 6B shows these operations during still image exposure. 6A and 6B illustrate only the operation of the linear prediction unit 51 and its surroundings, and the switches 43a and 46a shown in FIG. 5 and the like that are not related to the description here are illustrated. Is omitted. In these drawings, the components shown in FIG. 5 are denoted by the same reference numerals as those in FIG. 44 is a filter coefficient and 47 is an adder.

As shown in FIG. 6A, the linear prediction unit 51 is configured by a so-called transversal filter. However, as the linear prediction unit 51, another filter (for example, a lattice filter) that can use an appropriate adaptive algorithm may be used.
First, the adaptation process will be described. When the adaptive process as shown in FIG. 6A is performed, as is clear from the figure, the current predicted value y (n) is obtained using the following equation (1).

However, M is the order of the filter, and is appropriately set according to the sampling frequency of the signal to be subjected to the prediction process and the adaptive process, the time for performing the prediction process, and the like.

Various adaptive algorithms, which are algorithms for adaptive processing, have been proposed. Here, the LMS (Least Mean Square) algorithm will be described. This LMS algorithm is derived from the gradient method, and updates the filter coefficient h _{n according} to the following equation (2). However, the subscript n at the lower right of h indicates the filter coefficient of the nth sample.

μ is a positive coefficient called a step size parameter.

The LMS algorithm is an algorithm that uses the steepest descent method. According to this, the filter coefficient h _n approaches the minimum error value from the initial value. If the prediction processing error e (n) is sufficiently small, that is, if the predicted value y (n) is an approximate value close to the observed value u (n), the update amount by the adaptive processing is small.

  Next, the prediction process will be described. When the prediction process as shown in FIG. 6B is performed, the predicted value y (n) is used instead of the observed value u (n). In the example of FIG. 6B, y (n-1) is used instead of u (n-1). On the other hand, since the observed value is used as the input value in the previous prediction, FIG. 6B shows the case where the observed value as the input value in one prediction cannot be obtained appropriately.

  If the prediction processing error e (n) is sufficiently small by the adaptive processing described above, u (n-1) ≈y (n-1), and therefore, the input value in one prediction is predicted from the observed value. It is expected that the predicted value y (n) obtained again by replacing with the value y (n−1) is also a good approximate value. In the next prediction, y (n) is used instead of u (n) as an input value. By repeating this sequentially, prediction can be performed not only once but multiple times.

  As described above, in this embodiment, the sampling frequency of the A / D converter 61 for A / D converting the angular velocity signal from the gyro 26 is set to be the same as the sampling frequency of the image sensor 11, but other frequencies are used. It may be set to.

  In general, the sampling frequency of the image sensor is lower than that of a sensor such as a gyro. For example, a sensor such as a gyro can set a sampling frequency of 1 to 50 kHz with respect to a sampling frequency of the imaging element of 30 to 240 Hz. Therefore, in this embodiment, the sampling frequency of the A / D converter 61 is matched with the sampling frequency of the image sensor 11. However, the sampling frequency of the A / D converter 61 may be set to a higher frequency if the sampling rates match at the time of addition in the adder 69.

  For example, a downsampling device that converts the sampling frequency low is inserted between the A / D converter 61 and the integrator 62. As a result, the angular velocity signal sampled by the A / D converter 61 at a high sampling frequency is sampled at the sampling frequency of the image sensor 11 to perform downsampling. Then, an upsampling unit that converts the sampling frequency to a high level is inserted after the adder 69, and the combined shake signal from the adder 69 is sampled at the sampling frequency of the original A / D converter 61 to perform upsampling. .

  However, regardless of the sampling frequency of the A / D converter 61, the cutoff frequency of the HPF 66 and LPF 67 is equal to or lower than the Nyquist frequency (frame rate 1/2 or lower), which is half the sampling frequency (frame rate) for the image sensor 11. Is set.

  Further, down-sampling may be performed before the shake prediction unit 68 so that the sampling frequency of the motion vector detection signal is lower than the sampling frequency (predetermined sampling frequency) for the image sensor 11. The prediction filter of the shake prediction unit 68 is configured by an FIR filter, and a certain number of filter taps are required to predict a low-frequency component of a motion vector. For example, in order to predict a signal of frequency F1 [Hz], the product of the filter tap number N and the reciprocal of the sampling frequency fs is set so as to include one cycle of the frequency F1, as shown in Equation (3). It is desirable that

That is, the number of taps of the filter is required to predict the low frequency component rather than the prediction of the high frequency component, and as a result, the scale of the arithmetic circuit increases. Therefore, by lowering the sampling frequency, it is possible to predict low frequency components while suppressing an increase in the number of taps. Normally, when the sampling frequency of a signal is lowered by downsampling, the high frequency component contained in the signal is lost. However, in this embodiment, only the low frequency component is generated by the prediction process. The scale can be reduced.

  Then, an upsampling device that increases the sampling frequency of the predicted shake signal is inserted in the subsequent stage of the shake prediction unit 68 so that the sampling frequency of the predicted shake signal matches the sampling frequency (predetermined sampling frequency) for the image sensor 11. That's fine.

  Alternatively, the sampling frequency (frame rate) for the image pickup device 11 may be originally reduced, and an upsampler may be inserted only in the subsequent stage of the shake prediction unit 68. Then, the sampling frequency of the predicted shake signal may be matched with the sampling frequency of the A / D converter 61 on the gyro 26 side by this upsampling device.

  Further, the cut-off frequencies of the HPF 66 and the LPF 67 may be changed according to the state of the imaging device and the imaging conditions. For example, a cutoff frequency changing unit (frequency changing means) that changes the cutoff frequency of the HPF 66 and the LPF 67 is provided in the image stabilization control unit 30. Then, the camera control unit 10 gives an instruction to the cutoff frequency changing unit to change to the cutoff frequency according to the state of the imaging device and the imaging conditions. The state of the imaging device includes the state of the gyro 26, that is, the state of occurrence of drift in the gyro 26 (which frequency has a drift component generated). Further, the photographing condition includes an exposure time for still image exposure.

  First, a case where the cutoff frequency is changed according to the frequency of the drift component will be described. Before the still image exposure, both the gyro shake detection signal and the motion vector detection signal can be obtained in the entire frequency band. If the drift component of the gyro 26 is not included in the gyro shake detection signal, the camera shake amounts respectively indicated by the gyro shake detection signal and the motion vector detection signal substantially coincide with each other. That is, if the difference value between these two detection signals is obtained, the drift component of the gyro 26 can be extracted. The frequency component generated as the drift component does not change greatly in a short time from the framing before the still image exposure to the end of the still image exposure.

  Therefore, as shown in FIG. 7A, a differentiator (error extracting means) 72, a frequency analyzing section (analyzing means) 73, and a cutoff frequency changing section 74 are added to the configuration shown in FIG. The subtractor 72 extracts a drift component (error component) by calculating a difference between the gyro shake detection signal from the gain correction unit 65 and the motion vector detection signal from the motion vector detection unit 14 before the still image exposure. .

  The frequency analysis unit 73 performs frequency analysis such as fast Fourier transform (FFT) on the drift component extracted by the differentiator 72, and calculates the frequency of the drift component that is the analysis result. The cut-off frequency changing unit 74 changes the cut-off frequency of the HPF 66 and the LPF 67 to a frequency suitable for removing the drift component of the frequency calculated by the frequency analyzing unit 73 (higher than the frequency of the drift component). Since the cutoff frequency is preferably a lower frequency if the drift component can be removed, changing the cutoff frequency according to the calculated frequency of the drift component is effective for good anti-vibration control.

  Next, a case where the cutoff frequency is changed according to the exposure time will be described. For example, in this embodiment, a high-frequency gyro shake signal and a predicted shake signal (low-frequency component) are simply added during still image exposure, but the elapsed time from the start of still image exposure (length). The cut-off frequency may be changed accordingly.

  For example, since the prediction accuracy of the predicted shake signal decreases as the exposure time increases, the cutoff frequency changing unit 74 shown in FIG. 7A changes the cutoff frequency as time elapses from the start of still image exposure. Decrease gradually. For example, the cutoff frequency is kept at the initial setting value until 1/4 [sec] as a predetermined time from the start of still image exposure, and when it exceeds 1/4 [sec], the cutoff frequency is gradually lowered. When the cut-off frequency finally reaches a predetermined lower limit frequency, substantially only the high-frequency gyro shake signal is used for the image stabilization control. That is, image stabilization control is performed using a combined shake signal generated using a predicted shake signal until a predetermined time in the exposure time.

  However, when the predetermined time is exceeded, the utilization rate of the predicted shake signal is decreased and the utilization rate of the high-frequency gyro shake signal is increased. In other words, the weighting when combining both signals is changed. By controlling the cut-off frequency in this way, it is possible to perform the image stabilization control with high accuracy even at the time of long exposure.

  Also, an appropriate exposure time for still image exposure is calculated (set) according to the exposure amount detected by an exposure amount detection unit (not shown) before still image exposure, and cut according to the length of the set exposure time. The off frequency may be changed. As described above, when the exposure time is long, the prediction accuracy of the predicted shake signal decreases. Therefore, when the determined exposure time is long, it is cut before the still image exposure so that the utilization rate of the gyro shake detection signal is increased. The off frequency may be lowered from the initial setting value.

  Since the motion vector detection unit 14 detects the displacement of the subject image on the image sensor 11, it can also detect a shift shake that is a shake parallel to the imaging surface of the image sensor 11 of the camera 1. Shift shake tends to affect shooting with a short subject distance and a large image magnification. On the other hand, since the gyro 26 is an angular velocity sensor, shift shake cannot be detected. That is, during still image exposure, only the angular shake can be detected by the gyro 26, but the shift shake is also reduced by performing the anti-shake control using the predicted shake signal which is a low frequency component. As described above, this embodiment is effective for photographing with a large image magnification and a large influence of shift shake in the low frequency band.

  In the present embodiment, the lens-integrated camera 1 including the photographing lens 2 and the image sensor 11 has been described. However, the camera having the image sensor and an interchangeable lens (optical) that is detachably attached to the camera. It can also be applied to the case of using a device. In this case, either the camera control unit 10 provided in the camera or the lens control unit 25 provided in the interchangeable lens may have the image stabilization control unit 30. When the image stabilization control unit 30 is provided in the lens control unit 25, it is only necessary to obtain a motion vector detection signal and information indicating the start of still image exposure from the camera. On the other hand, when the image stabilization control unit 30 is provided in the camera control unit 10, the gyro shake detection signal is received from the interchangeable lens side, and the combined shake signal generated by the image stabilization control unit 30 is transmitted to the lens control unit 25. That's fine. Further, the gyro 26 may be provided in the camera.

  In the present embodiment, the case where the shift lens 24 serving as an optical image stabilizing element is shifted in the direction orthogonal to the photographing optical axis has been described. However, the imaging element 11 is shifted in the direction orthogonal to the photographing optical axis, The entire lens 2 (that is, the photographing optical axis) may be tilted. In this case, the entire image sensor 11 and the photographing lens 2 correspond to the optical image stabilizer.

  Furthermore, in the present embodiment, the case where a prediction filter using an adaptive algorithm is used in motion vector prediction processing has been described, but other prediction processing may be used. For example, if the exposure time is about 1/30 [sec], the low-frequency component of about 0.1 to 1 [Hz] is less than ¼ period, so that the motion vector before still image exposure can be more easily Even when linear linear prediction using the slope of the change is used, a certain degree of prediction accuracy can be obtained.

  The first-order linear prediction of the motion vector will be described with reference to FIG. FIG. 3 shows motion vectors in the entire frequency band, while FIG. 7B shows only the low frequency components. A thick solid line 91 is a low frequency motion vector signal after being detected by the motion vector detecting unit 14 and passing through the LPF 67. A thin solid line 92 is a low-frequency motion vector signal during still image exposure, and a broken line 93 is a predicted shake signal during still image exposure. The predicted shake signal 93 is obtained by first-order linear prediction in which the slope of the period 94 immediately before the start of still image exposure is extended as it is. As shown by the thin solid line 92 in FIG. 7B, if the exposure time is short, the low frequency component can be predicted with a certain degree of accuracy even in the first-order linear prediction, and good anti-vibration control is possible.

  In this embodiment, the case where the low-frequency component is extracted from the motion vector detection signal using the LPF 67 has been described. However, the sampling frequency of the image sensor 11 is set to the cutoff frequency corresponding to the low-frequency component to be extracted. It is also possible to make it equal to 1/2. When the sampling frequency is set in this way, the low frequency component of the desired motion vector detection signal can be extracted from the sampling theorem without the LPF 67, and the circuit can be simplified.

  The processing described in the present embodiment may be executed by the image stabilization control unit 30 configured by a computer according to an image stabilization control program as a computer program. FIG. 10 shows a flow of processing executed by the image stabilization control unit 30 that is a computer according to the image stabilization control program.

  In step S101, the image stabilization control unit 30 determines whether or not the camera control unit 10 has notified that the second release signal SW2 has been input. When the input of the second release signal SW2 is not notified, the process proceeds to step S102, and when notified (the still image is being exposed), the process proceeds to step S104.

  In step S102, the image stabilization control unit 30 acquires a motion vector detection signal from the motion vector detection unit 14, and acquires a low frequency component of the motion vector detection signal by passing the motion vector detection signal through the LPF. .

  In step S103, the image stabilization control unit 30 performs the above-described prediction filter adaptation processing using the low-frequency component of the motion vector detection signal.

  Thereafter, the image stabilization control unit 30 returns to step S101 and determines again whether or not the second release signal SW2 has been notified. If not notified, the processing of step S102 and step S103 is repeated.

  On the other hand, in step S <b> 104, the image stabilization control unit 30 performs A / D conversion and integration on the angular velocity signal from the gyro 26 to generate a gyro shake detection signal.

  Subsequently, in step S105, the image stabilization control unit 30 acquires the high frequency component of the gyro shake detection signal by passing the gyro shake detection signal through the HPF.

  In step S106, the image stabilization control unit 30 performs a prediction process for calculating the predicted shake signal. In step S107, the high-frequency component of the gyro shake detection signal and the predicted shake signal are added (synthesized) to generate a synthesized shake. Generate a signal.

  Next, in step S108, the image stabilization control unit 30 performs shift driving (image stabilization control) of the shift lens 24 using the combined shake signal.

  Finally, in step S109, the image stabilization control unit 30 determines whether still image exposure is in progress. If still image exposure is in progress, the process returns to step S104 and repeats the processing from steps S104 to S108. On the other hand, when the still image exposure is completed, the process returns to step S101.

  The image stabilization control of the present embodiment has been described to be effective for photographing with a large image magnification and a large influence of shift shake in the low frequency band, but the cutoff frequency is changed from the initial setting value according to the image magnification. May be raised. This will be described with reference to FIG. FIGS. 11A and 11B are flowcharts showing processing executed by the image stabilization control unit 30 when performing control to increase the cutoff frequency from the initial setting value to a higher frequency in accordance with the image magnification.

  In step S <b> 201, the image stabilization control unit 30 determines whether the camera control unit 10 has been notified that the second release signal SW <b> 2 has been input. When the input of the second release signal SW2 has not been notified, the process proceeds to step S202, and when notified, the process proceeds to step S205.

  In step S202, the image stabilization control unit 30 performs a cut-off frequency changing process as a subroutine process shown in the flowchart of FIG.

  In the cut-off frequency changing process, in step S301, the image stabilization control unit 30 detects the lens state of the photographing lens 2 including the position of the focus lens 22 and the focal length through the lens control unit 25. In step 302, the image stabilization control unit 30 serving as the image magnification detection unit calculates the image magnification of the subject based on the detected lens state.

  In step S303, the image stabilization control unit 30 determines whether or not the calculated image magnification is larger than a preset image magnification threshold. If the image magnification is larger than the preset image magnification threshold, the image stabilization control unit 30 proceeds to step S304 because the influence of the shift shake is large. Otherwise, the present processing is terminated and the processing shown in FIG. Returning to the flowchart, the process proceeds to step S203.

  In step S304, the image stabilization control unit 30 serving as the frequency changing unit sets the cutoff frequency to a higher frequency. Then, the image stabilization control unit 30 ends this processing and proceeds to step S203 in FIG.

  The processing from step S203 to S210 is the same as the processing from step S102 to S109 in FIG.

  In this way, when the image magnification is larger than the threshold value, by controlling to change the cutoff frequency, even in the frequency band on the higher frequency side than the frequency band of the shift shake that has been cut off at the conventional cutoff frequency. Anti-vibration control can be performed by the prediction process.

  The value of the cut-off frequency is desirably determined in consideration of the prediction accuracy of the shake prediction unit 68 using the signal of the motion vector detection unit 14 and the estimated amount of shift shake that is not detected by the gyro 26.

  When imaging is performed with such a large image magnification (the influence of shift shake is large), since a prediction signal is used for angular shake in a frequency band lower than the raised cutoff frequency, high-precision prevention using a gyro shake detection signal is used. There is a risk that the accuracy may decrease with respect to vibration control. However, since the shift shake frequency band that has been excluded from the correction target can be corrected by prediction, a still image with reduced image shake can be obtained as a result of the conventional image stabilization control.

  Further, the operation of the shake prediction unit 68 may be controlled as follows using the angular velocity signal output from the gyro 26.

  The shake prediction unit 68 of the present embodiment updates the prediction filter by adaptive processing, but when the sudden disturbance is received, the filter coefficient cannot be updated and the prediction filter is stabilized again. Need time. For example, when the user holds the camera 1 and performs large panning with the release switch (SW1) pressed down, the motion vector detection unit 14 detects a large shake different from the user's hand shake. Then, when the shake prediction unit 68 updates the prediction filter using the motion vector detection signal obtained at that time, it takes time until the filter coefficient changes greatly and the prediction filter is stabilized again.

  On the other hand, since the motion vector detection signal used in the shake prediction unit 68 is detected by comparing two images that are continuously exposed, a slight delay occurs with respect to the shake detection by the gyro 26, although it depends on the frame rate. Occurs. For example, if the frame rate is 60 fps, the motion vector detection unit 14 has a sampling rate of 60 Hz, whereas the sampling rate of the gyro 26 is 500 Hz to several kHz. However, since the motion vector detection signal is a signal in a low frequency band, there is little correction error due to the effect of delay even if it is used as it is for anti-vibration control.

  Therefore, as described below, when a large movement such as panning is detected by the gyro 26, the update of the filter coefficient by the adaptive processing may be temporarily stopped.

  The flowchart of FIG. 12 shows processing when the above-described control is performed. In step S401, the image stabilization control unit 30 determines whether or not the camera control unit 10 has notified that the second release signal SW2 has been input. When the input of the second release signal SW2 is not notified, the process proceeds to step S402, and when notified, the process proceeds to step S405.

  In step S402, the image stabilization control unit 30 acquires a motion vector detection signal from the motion vector detection unit 14, and acquires a low-frequency component that has passed through the LPF from the motion vector detection signal. In step S <b> 403, the image stabilization control unit 30 starts the prediction filter adaptation process described above using the low-frequency component of the motion vector detection signal.

  In step S404, the image stabilization control unit 30 performs a panning detection process as a subroutine process shown in the flowchart of FIG. In step S501 of FIG. 12B, it is determined whether the camera shake amount (hereinafter referred to as shake detection value) indicated by the gyro shake detection signal from the gyro 26 is equal to or greater than a predetermined value. When the shake detection value equal to or greater than the predetermined value is detected by the gyro 26, the image stabilization control unit 30 proceeds to step S502, and when the shake detection value less than the predetermined value is detected, this process is terminated.

  In step S501 of FIG. 12B, the image stabilization control unit 30 determines whether or not the shake detection value by the gyro 26 is equal to or greater than a predetermined value. If camera shake of a predetermined value or more is detected by the gyro 26, the process proceeds to step S502, and if the camera shake is less than the predetermined value, this process ends.

  In step S502, the image stabilization control unit 30 determines that the user has moved greatly different from camera shake such as panning because the camera shake detection value is greater than or equal to the predetermined value in step S502, and updates the filter coefficient by adaptive processing. Is temporarily stopped.

  In step S503, the image stabilization control unit 30 waits for processing for a predetermined time after stopping the update of the filter coefficient. The predetermined time is set to about 0.2 to 0.5 sec assuming a panning time, for example, and this is an update interval of filter coefficients updated at 60 Hz which is the sampling rate of the motion vector detection unit 14. Long enough compared to. While the update of the filter coefficient by the adaptive processing is stopped, the prediction shake signal may be used as the motion vector detection signal by performing the prediction processing in the same manner as the operation during still image exposure.

  In step S504, the image stabilization control unit 30 determines whether or not the shake detection value from the gyro 26 has dropped below a predetermined value. If it is less than the predetermined value, the process proceeds to step S505. If it is equal to or greater than the predetermined value, the process returns to step S503, and again waits for a predetermined time.

  In step S505, the image stabilization control unit 30 restarts the update of the filter coefficient by the adaptive process, and ends this process. After the end of this process, the image stabilization control unit 30 returns to step S401 shown in FIG. The subsequent processes in steps S405 to S410 are the same as the processes in steps S104 to S109 in FIG. 10 and steps S205 to S210 in FIG.

  As described above, when a large movement different from the user's hand movement such as panning occurs, the filter coefficient update (adaptive processing) is temporarily stopped until the movement is settled, and the adaptation is performed when the large movement falls. Resume processing. By controlling in this way, it is possible to prevent the filter coefficient of the prediction filter from being disturbed by a large movement different from camera shake due to panning or the like, and as a result, it is possible to shorten the time required for stabilization of the prediction filter. .

  Next, a digital camera provided with an image stabilization control apparatus (image stabilization control unit 30 ') according to the second embodiment of the present invention will be described with reference to FIG. In the first embodiment, the case where the image stabilization control is started in response to the second release signal SW2 instructing still image exposure has been described. However, in this embodiment, the image display section 16 has less image blur even before still image exposure. Anti-vibration control is performed so that live view images can be displayed. Specifically, the image stabilization control unit 30 ′ starts the image stabilization control in response to the output of the first release signal SW1 instructing the imaging preparation operation from the operation unit 15. At this time, image stabilization control using a motion vector detection signal is performed instead of the combined shake signal described in the first embodiment until the second release signal SW2 is output from the operation unit 15. Thereafter, when the second release signal SW2 is output from the operation unit 15, the image stabilization control using the combined shake signal is performed as in the first embodiment.

  The configuration of the camera in this embodiment is the same as that of the digital camera 1 shown in FIG. Further, in FIG. 8, the same reference numerals as those in the first embodiment are attached to the same components as those in the image stabilization control unit 30 described in the first embodiment with reference to FIG.

  In FIG. 8, reference numeral 75 denotes an adder, which is a signal obtained by adding a motion vector detection signal from the motion vector detection unit 14 and a signal indicating a shift drive amount (control amount) given from the shift control unit 71 to the shift lens 24. Output to BPF 63b and LPF 67. Reference numeral 70 'denotes an output changeover switch. The terminal 70a connected to the shift controller 71 is connected to the terminal 70b to which the motion vector detection signal is supplied, and the terminal 70c to which the combined shake signal from the adder 69 is supplied. It is possible to switch to a state connected to. The output changeover switch 70 'is switched in response to a command from the camera control unit 10 shown in FIG.

  Before the first release signal SW1 is output, the terminal 70a of the output changeover switch 70 'is in an off state in which it is not connected to any of the terminals 70b and 70c. In this state, the shift driving of the shift lens 24, that is, the image stabilization control is not performed.

  When the first release signal SW1 is output (when the imaging preparation period is reached), the terminal 70a of the output changeover switch 70 'is connected to the terminal 70b. Thereby, the motion vector detection signal from the motion vector detection unit 14 is input to the shift control unit 71, and the image stabilization control using the motion vector detection signal is performed.

  At this time, the motion vector detection signal after the start of the image stabilization control corresponds to an image shake (hereinafter referred to as an afterimage shake) on the image sensor 11 that cannot be removed by the shift driving of the shift lens 24. That is, the sum of the shift driving amount of the shift lens 24 and the motion vector corresponding to the afterimage shake becomes the motion vector corresponding to the actual camera shake amount. Therefore, in this embodiment, a low-frequency component of a signal indicating a motion vector corresponding to an actual camera shake amount (hereinafter referred to as an addition motion vector detection signal) is extracted, and the extracted addition motion vector detection signal (low The prediction filter is updated by the shake prediction unit 68 using the frequency component.

  Specifically, the shift control unit 71 outputs a signal indicating a shift driving amount of the shift lens 24 (a signal indicating a converted value to a motion vector) to the adder 75, and the adder 75 adds the signal and the motion vector detection unit. 14 and the motion vector detection signal from 14 are added. Then, the LPF 67 extracts a low frequency component of the addition motion vector detection signal generated by this addition. The shake prediction unit 68 performs prediction filter adaptation processing using the low-frequency component of the addition motion vector detection signal from the LPF 67.

  When the second release signal SW2 is output (during still image exposure), the terminal 70a of the output changeover switch 70 'is connected to the terminal 70c. The shake prediction unit 68 outputs to the adder 69 the predicted shake signal in the low frequency band calculated by the prediction filter that has been subjected to the adaptive processing so far. The added motion vector detection signal from the adder 75 is input to the gain comparison unit 64 through the BPF 63b. The gain correction unit 65 corrects the gain of the gyro shake detection signal in accordance with the gain comparison result between the gyro shake detection signal and the added motion vector detection signal in the gain comparison unit 64. Then, the HPF 66 extracts a high frequency component of the gyro shake detection signal whose gain has been corrected, and outputs the high frequency component to the adder 69. The adder 69 adds the gyro shake detection signal (high frequency component) and the predicted shake signal (low frequency component) to generate a synthesized shake signal, which is input to the shift control unit 71 via the output changeover switch 70 '. To do. In this way, image stabilization control using the combined shake signal is performed.

  When the still image exposure is completed, the processing of the shake prediction unit 68 and the output changeover switch 70 ′ are returned to the state before the first release signal SW1 is output.

  According to the present embodiment, since the image stabilization control using the motion vector detection signal is performed from the imaging preparation period before the still image exposure, the user performs the framing while watching the live view image in which the image blur is well corrected. Can do. During still image exposure, as in the first embodiment, the image stabilization control is performed using the combined shake signal from which the low-frequency component including the drift component generated by the gyro 26 is removed, so that it is affected by the drift component. Therefore, good vibration control can be performed.

  Next, a digital camera provided with an image stabilization control device (image stabilization control unit 30 ″) that is Embodiment 3 of the present invention will be described with reference to FIG. The image stabilization control is performed even during the preparation period, but differs from the second embodiment in that the image stabilization control is performed using a gyro shake detection signal instead of a motion vector detection signal. 2 is the same as the digital camera 1 shown in Fig. 2. The description is omitted in Fig. 9. In Fig. 9, the anti-vibration control unit 30 described in the first embodiment with reference to Fig. 1 and the second embodiment in Fig. Constituent elements common to the constituent elements of the image stabilization control unit 30 ′ described above are denoted by the same reference numerals, and are not described.

  In FIG. 9, reference numeral 76 denotes a differentiator, which outputs a difference between the gyro shake detection signal after gain correction from the gain correction unit 65 and the motion vector detection signal from the motion vector detection unit 14 to the LPF 67. Reference numeral 70 ″ denotes an output changeover switch. The terminal 70 a connected to the shift control unit 71 is connected to the terminal 70 b to which the gyro shake detection signal is supplied from the gain correction unit 65, and the synthesized shake signal is supplied from the adder 69. The output changeover switch 70 ″ can be switched according to a command from the camera control unit 10 shown in FIG.

  Before the first release signal SW1 is output, the terminal 70a of the output changeover switch 70 ″ is in an off state in which it is not connected to any of the terminals 70b and 70c. Vibration control is not performed.

  When the first release signal SW1 is output (when the imaging preparation period is reached), the terminal 70a of the output changeover switch 70 ″ is connected to the terminal 70b. Thereby, the gyro shake detection after gain correction from the gain correction unit 65 is performed. The signal is input to the shift control unit 71, and the image stabilization control using the gyro shake detection signal is performed instead of the synthesized shake signal.

  At this time, the motion vector detection signal obtained from the motion vector detection unit 14 is a signal corresponding to the drift component of the gyro 26. Therefore, the subtractor 76 subtracts the motion vector detection signal corresponding to the drift component from the gyro shake detection signal from the gain correction unit 65 to remove the drift component (hereinafter referred to as a drift removal gyro shake signal). Is generated. Then, the LPF 67 extracts the low frequency component of the drift removal gyro shake signal from the difference unit 76, and the shake prediction unit 68 performs adaptive processing of the prediction filter using the low frequency component of the drift removal gyro shake signal.

  When the second release signal SW2 is output (when still image exposure is in progress), the terminal 70a of the output changeover switch 70 ″ is connected to the terminal 70c. The shake prediction unit 68 uses the prediction filter that has been subjected to adaptive processing so far. The calculated low-frequency band predicted shake signal is output to the adder 69. In this embodiment, the drift-removed gyro shake signal is not the predicted shake signal as the motion vector prediction value as in the first and second embodiments. A predicted shake signal as a predicted value is calculated, but the same as in the first and second embodiments in that the motion vector detection signal is used in the adaptive processing of the prediction filter as described above.

  The addition motion vector detection signal from the adder 75 is input to the gain comparison unit 64 through the BPF 63b.

  The gain correction unit 65 corrects the gain of the gyro shake detection signal in accordance with the gain comparison result between the gyro shake detection signal and the added motion vector detection signal in the gain comparison unit 64. Then, the HPF 66 extracts a high frequency component of the gyro shake detection signal whose gain has been corrected, and outputs the high frequency component to the adder 69. The adder 69 adds the gyro shake detection signal (high frequency component) and the predicted shake signal (low frequency component) to generate a synthesized shake signal, which is input to the shift control unit 71 via the output changeover switch 70 ″. In this way, image stabilization control using the combined shake signal is performed.

  When the still image exposure is completed, the processing of the shake prediction unit 68 and the output changeover switch 70 ″ are returned to the state before the first release signal SW1 is output.

According to the present embodiment, since the image stabilization control using the gyro shake detection signal from which the drift component has been removed from the imaging preparation period before the still image exposure is performed, the user can view the live view image in which the image shake is well corrected. Framing can be performed while watching. During still image exposure, as in the first embodiment, the image stabilization control is performed using the combined shake signal from which the low-frequency component including the drift component generated by the gyro 26 is removed, so that it is affected by the drift component. Therefore, good vibration control can be performed.
(Other examples)
The present invention supplies a program that realizes one or more functions of the above-described embodiments to a system or apparatus via a network or a storage medium, and one or more processors in a computer of the system or apparatus read and execute the program This process can be realized. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.
Each embodiment described above is only a representative example, and various modifications and changes can be made to each embodiment in carrying out the present invention.

DESCRIPTION OF SYMBOLS 1 Camera 2 Shooting lens 11 Image pick-up element 14 Motion vector detection part 24 Shift lens 25 Lens control part 26 Gyro 30,30 ', 30 "Anti-vibration control part 68 Shake prediction part 71 Shift control part

Claims (13)

An image due to the shake of the optical device using a shake detection signal obtained by detecting the shake of the optical device and a motion vector detection signal indicating a motion vector detected in a video signal generated by imaging using the optical device. An anti-vibration control device that performs anti-vibration control to reduce shake,
A shake extracting means for extracting a first shake signal which is a frequency component higher than a predetermined frequency in the shake detection signal;
Predicting means for generating a predicted shake signal indicating a predicted value of the shake using the motion vector detection signal;
Combining means for combining the first shake signal and the predicted shake signal to generate a second shake signal;
Control means for performing the image stabilization control using the second shake signal;
Motion vector extraction means for extracting a specific motion vector signal that is a frequency component lower than the predetermined frequency in the motion vector detection signal;
The control means performs the image stabilization control using the second shake signal in capturing a still image,
The image stabilization control apparatus, wherein the prediction unit generates the predicted shake signal using the specific motion vector signal. Using a shake detection signal obtained by shake detection of an optical device and a motion vector detection signal indicating a motion vector detected in a video signal generated by imaging using the optical device and an imaging device, the optical device An image stabilization control device that performs image stabilization control to reduce image blur due to image blur,
A shake extracting means for extracting a first shake signal which is a frequency component higher than a predetermined frequency in the shake detection signal;
Predicting means for generating a predicted shake signal indicating a predicted value of the shake using the motion vector detection signal;
Combining means for combining the first shake signal and the predicted shake signal to generate a second shake signal;
Control means for performing the image stabilization control using the second shake signal,
An anti-vibration control device comprising: a frequency changing unit that changes the predetermined frequency in accordance with a lapse of an exposure time of the image sensor in the imaging or in accordance with a length of the exposure time. The prediction means includes
Generating the predicted shake signal using a prediction filter;
The image stabilization control apparatus according to claim 1, wherein an adaptive process of updating the prediction filter using the motion vector detection signal is performed before the prediction shake signal is generated.   The image stabilization control apparatus according to any one of claims 1 to 3, wherein the predetermined frequency is set to ½ or less of a sampling frequency for an image sensor used for the imaging. Error extraction means for extracting an error component included in the shake detection signal from a difference between the shake detection signal and the motion vector detection signal;
Analyzing means for analyzing the frequency of the error component;
5. The image stabilization control device according to claim 1, further comprising: a frequency changing unit that changes the predetermined frequency in accordance with an analysis result of the frequency.   Sampling means for down-sampling the specific motion vector signal generated at a predetermined sampling frequency and up-sampling the predicted shake signal generated using the down-sampled specific motion vector signal at the predetermined sampling frequency; The anti-vibration control device according to any one of claims 1 to 5. When the control means performs the image stabilization control using the motion vector detection signal instead of the second shake signal,
The said prediction means produces | generates the said predicted shake signal using the signal obtained by adding the control amount and the said motion vector detection signal in this anti-shake control, The any one of Claim 1 to 6 characterized by the above-mentioned. An anti-vibration control device according to claim 1. When the control means performs the image stabilization control using the shake detection signal instead of the second shake signal,
The image stabilization control according to any one of claims 1 to 6, wherein the prediction unit generates the predicted shake signal using a difference between the shake detection signal and the motion vector detection signal. apparatus. An image magnification detection means for detecting an image magnification formed on the image sensor from a photographing state of the optical device;
9. The image stabilization control apparatus according to claim 1, further comprising frequency changing means for increasing the predetermined frequency when the detected image magnification is larger than an image magnification threshold. If the first shake signal indicates the shake greater than a predetermined value,
The prediction means temporarily stops the adaptation process,
The image stabilization control apparatus according to claim 3, wherein the predicted shake signal generated using the prediction filter in which the adaptive processing is stopped is used as the motion vector detection signal. The image stabilization control device according to any one of claims 1 to 10,
An optical apparatus comprising: an optical image stabilization element driven by the image stabilization control device. Using a shake detection signal obtained by detecting a shake of the optical device and a motion vector detection signal indicating a motion vector detected in a video signal generated by imaging using the optical device, on a computer, A computer program that performs image stabilization control to reduce image blur due to image blur,
In the computer,
A first shake signal that is a frequency component higher than a predetermined frequency is extracted from the shake detection signal;
Using the motion vector detection signal, generating a predicted shake signal indicating a predicted value of the shake,
Combining the first shake signal and the predicted shake signal to generate a second shake signal;
The image stabilization control is performed using the second shake signal,
A specific motion vector signal that is a frequency component lower than the predetermined frequency is extracted from the motion vector detection signal;
further,
The image stabilization control using the second shake signal is performed in capturing a still image,
An anti-vibration control program that generates the predicted shake signal using the specific motion vector signal. Using a shake detection signal obtained by detecting a shake of an optical device and a motion vector detection signal indicating a motion vector detected in a video signal generated by imaging using the optical device and an imaging device, on the computer, A computer program for performing image stabilization control to reduce image blur due to optical device shake,
In the computer,
A first shake signal that is a frequency component higher than a predetermined frequency is extracted from the shake detection signal;
Using the motion vector detection signal, generating a predicted shake signal indicating a predicted value of the shake,
Combining the first shake signal and the predicted shake signal to generate a second shake signal;
The image stabilization control is performed using the second shake signal,
An anti-vibration control program that changes the predetermined frequency in accordance with the lapse of the exposure time of the image sensor in the imaging or in accordance with the length of the exposure time.