JP6590018B2 - Blur correction device and camera - Google Patents

Description

  The present invention relates to a shake correction apparatus and a camera.

There is a technique for performing optical image stabilization based on a signal from an angular velocity sensor (gyro). In this technique, it is necessary to accurately obtain the reference value (gyro output value at 0 deg / s) of the angular velocity sensor, but the reference value of the angular velocity sensor changes depending on drift characteristics immediately after startup, temperature characteristics, and the like. It is difficult to find accurately.
Therefore, a technique has been proposed in which the output signal of the angular velocity sensor is calculated by LPF (low-pass filter) processing, and the reference value is corrected using motion vector information (see Patent Document 1).
Here, the “motion vector” is not a motion of the subject but a “motion vector corresponding to camera shake”.

Japanese Patent No. 4419466

However, when the reference value is corrected by using the motion vector information, the calculation interval of the reference value is usually shorter than the calculation interval of the motion vector. Therefore, there is a problem that the reference value is updated even when the motion vector is not calculated, and the accuracy of the reference value is deteriorated.
An object of the present invention is to provide a camera shake correction apparatus and a camera capable of correcting a reference value with high accuracy.

The present invention captures an image of a subject formed by an optical system and outputs a first signal; a detection unit that detects a movement of an optical device having the image sensor and outputs a second signal; said detector shake correction unit which is driven to correct the motion detected by said from the detection by the detection unit is initiated until the first hour has elapsed, the second of said first signal A motion vector is calculated from an image generated from the first signal output at a time interval, and after the first time has elapsed since the detection by the detection unit was started, the first signal out of the first signal Based on the motion vector and the second signal, a motion vector calculation unit that calculates the motion vector from an image generated from the first signal output at an interval of a third time longer than the second time Calculate the drive amount of the image stabilization unit A driving amount calculation unit relates blur correction apparatus comprising a.
The present invention also relates to a camera having the blur correction device.

  ADVANTAGE OF THE INVENTION According to this invention, the blurring correction apparatus and camera which can perform an accurate reference value correction can be provided.

It is sectional drawing which shows typically a camera provided with the blurring correction apparatus of 1st Embodiment. It is a block diagram of the blurring correction apparatus of 1st Embodiment. It is the flowchart which showed the flow of operation | movement of the blurring correction apparatus of 1st Embodiment. It is a figure explaining a reference value calculating part, (a) is a reference value calculating part, (b) is a figure showing LPF of a reference value calculating part. It is a flowchart which shows a reference value correction calculation. (A) is a graph which shows the reference value after correction | amendment, (b) is a graph which shows the direction of a motion vector. It is the graph which compared the case where correction | amendment of this embodiment is performed (solid line), and the case where correction | amendment is not performed (dotted line). It is a figure corresponding to FIG. 6 which shows 2nd Embodiment of this invention.

(First embodiment)
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings and the like.
FIG. 1 is a cross-sectional view schematically showing a camera 1 including the shake correction apparatus of the first embodiment.
The camera 1 is a digital single-lens reflex camera, and includes a camera housing 1A and a lens barrel (optical member) 1B that is detachably attached to the camera housing 1A.

  The camera housing 1A includes a camera CPU 2A, an image sensor 3, a recording medium 13, an EEPROM 14, a signal processing circuit 15, an AF sensor 16, a release switch 17, a rear liquid crystal 18, a mirror 19, a sub mirror 19a, and a shutter 20.

The camera CPU 2A is a central processing unit that performs overall control of the camera 1, and constitutes a part of the shake correction device of the present embodiment.
The imaging element 3 is an element that captures a subject image formed by the photographing lenses (4, 5, 6), exposes the subject light to convert it into an electrical image signal, and outputs it to the signal processing circuit 15. The image pickup device 3 is configured by an element such as a CCD or a CMOS.

The recording medium 13 is a medium for recording captured image data, and an SD card, a CF card, or the like is used.
The EEPROM 14 is a memory that stores adjustment value information such as the gain value of the angular velocity sensor 12, information unique to the lens barrel, and the like, and outputs the memory to the CPU 2.
The signal processing circuit 15 is a circuit that receives an output from the image sensor 3 and performs processing such as noise processing and A / D conversion.

The AF sensor 16 is a sensor for performing AF (automatic focus adjustment), and a CCD or the like can be used.
The release switch 17 is a member that performs a photographing operation of the camera 1 and is a switch that operates a shutter driving timing and the like.

  The rear liquid crystal 18 is provided on the rear surface of the camera housing 1A of the camera 1 and displays a subject image (reproduced image, live view image) photographed by the image sensor 3 and information (menu) related to operation. It is.

  The shutter 20 is disposed behind the mirror 19. Subject light is incident on the shutter 20 when the mirror 19 rotates upward and becomes ready for photographing. The shutter 20 travels through a shutter curtain in response to a shooting instruction from the release switch 17 or the like, and controls subject light incident on the image sensor 3. A sub-mirror 19 a is provided behind the mirror 19 and guides subject light to the AF sensor 16.

  The lens barrel 1B includes a zoom lens group 4, a focus lens group 5, a shake correction lens group (optical element) 6, a zoom lens group drive mechanism 7, a focus lens group drive mechanism 8, and a shake correction lens group drive mechanism (optical element drive). Part) 9, an aperture 10, an aperture drive mechanism 11, an angular velocity sensor 12, and a lens CPU 2B.

  The lens CPU 2B calculates the movement amount of the lens groups such as the zoom lens group 4, the focus lens group 5, and the blur correction lens group 6. The zoom lens group drive mechanism 7, the focus lens group drive mechanism 8, the blur correction lens group drive mechanism 9 and the aperture drive mechanism 11 are instructed to move the zoom lens group 4, the focus lens group 5, and the blur correction lens group 6. Move.

The zoom lens group 4 is a lens group that is driven by the zoom lens group drive mechanism 7 and moves along the optical axis direction to continuously change the magnification of the image.
The focus lens group 5 is a lens group that is driven by the focus lens group drive mechanism 8 and moves in the optical axis direction to focus.
The shake correction lens group 6 (optical element) is a lens group that is optically shake-corrected and driven on a plane perpendicular to the optical axis by a shake correction lens group drive mechanism 9 such as a VCM.

The diaphragm 10 is a mechanism that is driven by the diaphragm drive mechanism 11 and controls the amount of subject light passing through the photographing lenses (4, 5, 6).
The angular velocity sensor 12 is a sensor that detects an angular velocity of a shake that occurs in each sensor unit.

FIG. 2 is a block diagram of the shake correction apparatus 100 of the present embodiment.
The shake correction apparatus 100 includes a camera CPU 2A, a lens CPU 2B, an angular velocity sensor 12, a shake correction lens group drive mechanism (lens drive unit) 9, a lens position detection unit 21, and a shake correction lens group 6.

The camera CPU 2 </ b> A includes an image sensor 3, a signal processing unit 40, and a motion vector calculation unit 41.
The lens CPU 2B includes an amplification unit 31, a first A / D conversion unit 32, a second A / D conversion unit 33, a reference value calculation unit 34, a reference value correction amount calculation unit 35, and a target position calculation unit 36 including an integration unit therein. A center bias calculator 37, a center bias remover 38, and a drive amount calculator 39.
The angular velocity sensor 12 includes the lens barrel 1B around the X axis (Pitch), around the Y axis (Yaw) (where the X axis is the horizontal axis of the light receiving surface of the image sensor 3 at the normal position of the camera, This is a sensor such as a vibration gyro that detects the angular velocity of the vertical axis.

The amplifying unit 31 amplifies the output of the angular velocity sensor 12.
The first A / D converter 32 performs A / D conversion on the output of the amplifier 31.
The reference value calculation unit 34 calculates a reference value (corrected angular velocity reference value) of the vibration detection signal (output of the first A / D conversion unit 32) obtained from the angular velocity sensor 12.
Then, the reference value calculated by the reference value calculation unit 34 is subtracted from the output of the first A / D conversion unit 32 by the subtraction unit 43.
The target position calculation unit 36 calculates the target position of the blur correction lens group 6 based on the output of the angular velocity sensor 12 after the reference value is subtracted by the subtraction unit 43.

The center bias calculation unit 37 biases the centripetal force for moving the image blur correction lens 6 toward the center of the movable range based on the target position of the image blur correction lens 6 calculated by the target position calculation unit 36. Calculate as a quantity.
Then, the control position of the image blur correction lens 6 is calculated by subtracting the calculated bias amount from the target position of the image blur correction lens 6.
By performing the centering bias processing in this way, it is possible to effectively prevent the image blur correction lens 6 from colliding with the hard limit, and it is possible to improve the appearance of the captured image.

  The drive amount calculation unit 39 is a blur correction lens group obtained from the target position from the target position calculation unit 36 and the value detected by the lens position detection unit 21 and A / D converted by the second A / D conversion unit 33. 6 is used to calculate the driving amount of the blur correction lens group driving mechanism 9.

The image sensor 3 is provided on the planned focal plane of the photographing optical system. The image pickup device 3 is composed of a device such as a CCD or a CMOS, and generates an analog image signal by photoelectrically converting an input subject image.
The signal processing unit 40 performs predetermined processing on the image signal generated by the image sensor 3. Actually, in order to obtain a motion vector with the configuration of FIG. 1, the mirror is raised, a through image (live view) is obtained, and calculation is performed in a state where light (image) reaches the image sensor 3.

The motion vector calculation unit 41 calculates a motion vector (first motion vector) indicating the motion of the image (motion direction, amount of motion) from the captured image processed by the signal processing unit 40.
Specifically, the motion vector calculation unit 41 detects the motion direction and the motion amount of the image by comparing the luminance information included in two consecutive frame image data captured by the image sensor 3. Calculate the motion vector. Actually, a motion vector equivalent to camera shake is calculated from the motion of the image.

The center bias removing unit 38 is a bias amount calculated by the center bias calculating unit 37 (subtracted from the target position of the image blur correction lens 6) from the first motion vector (MV1) that is the output of the motion vector calculating unit 41. Is subtracted.
The reference value correction amount calculation unit (reference value correction unit) 35 calculates a reference value correction amount based on the second motion vector (MV2) from which the center bias amount has been removed by the center bias removal unit 35.
Then, the subtraction unit (reference value correction unit) 42 subtracts the reference value correction amount obtained by the reference value correction amount calculation unit 35 from the output of the reference value calculation unit 34.

Next, an operation flow of the shake correction apparatus 100 according to the present embodiment will be described.
FIG. 3 is a flowchart showing an operation flow of the shake correction apparatus 100 of the present embodiment.

  After the camera 1 is turned on, the shake correction apparatus 100 starts a calculation for optical image stabilization. Depending on the camera, when the half-push switch 45 is pressed, the shake correction apparatus 100 starts calculation for optical image stabilization (step S001).

After the output of the angular velocity sensor 12 is amplified by the amplification unit 31, it is A / D converted by the first A / D conversion unit 32 (step S002).
The reference value calculation unit 34 calculates a reference value (a value corresponding to zero deg / s) of the calculated angular velocity based on the signal after A / D conversion of the output of the angular velocity sensor 12. Since the reference value of the angular velocity changes depending on the temperature characteristic, the drift characteristic immediately after startup, etc., for example, the stationary output of the angular velocity sensor 12 at the time of factory shipment cannot be used as the reference value.

  As a method for calculating the reference value, a method of calculating a moving average for a predetermined time and a method of calculating by LPF processing are known. In the present embodiment, reference value calculation by LPF processing is used.

  4A and 4B are diagrams for explaining the reference value calculation unit 34. FIG. 4A shows the reference value calculation unit 34 (HPF), and FIG. 4B shows the LPF 34a of the reference value calculation unit 34. .

  The cut-off frequency fc of the LPF 34 is generally set to a low frequency of about 0.1 [Hz]. This is because camera shake has a dominant frequency of about 1 to 10 [Hz]. If the fc is 0.1 [Hz], there is little influence on the camera shake component, and good blur correction can be performed.

  However, during actual shooting, low-frequency motion such as fine adjustment of the composition (a level at which panning cannot be detected) is added, which may cause an error in the ω0 calculation result. In addition, since fc is low (time constant is large), there is a problem that it takes time to converge to a true value when the error is once increased. In the present embodiment, this ω0 error is corrected.

Returning to FIG. 3, when the information of the first motion vector MV1 is updated (S004, YES), the process proceeds to S005, and when it is not updated (S004, NO), the process proceeds to S006. The description of S004 to S005 will be described later in detail.
In addition, since information on the first motion vector MV1 cannot be obtained during exposure, this step is performed until immediately before the exposure.
The outputs of the angular velocity sensor 12 after subtracting the reference value are integrated, and the target position of the shake correction lens group 6 is calculated based on the focal length, subject distance, shooting magnification, and shake correction lens characteristic information (S006).
Center bias processing is performed to prevent the blur correction lens group 6 from reaching the movable end (S007).

  There are various methods for the center bias processing, such as a method for setting a bias amount in accordance with the target position information, an HPF processing (in S006), and the method is not limited here.

The lens driving amount is calculated from the difference between the target position information including the center bias component and the blur correction lens position information (S008).
The blur correction lens group 6 is driven to the target position (S009), and the process returns to S002.

Next, reference value correction (S004 to S005, S011 to S016) will be described. FIG. 5 is a flowchart showing the reference value correction calculation.
As described above, when the information of the first motion vector MV1 is updated (S004), the process proceeds to S005. If it has not been updated, the process proceeds to S006.
Since the optical blur correction control cycle is sufficiently early with respect to the MV1 update cycle, the same arithmetic processing as normal image stabilization is performed until the MV1 is updated. Here, it is assumed that the control period of optical blur correction is 1 [ms] and the update period of the first motion vector MV1 is 33 [ms] (= 30 [fps]). A well-known technique is used for the calculation method of the first motion vector MV1.

All the received first motion vectors MV1 are summed (S011).
The center bias component calculated in S007 is converted to the same scale as the first motion vector MV1 (S012).
The conversion method is calculated based on the focal length, the subject distance, the shooting magnification, and the resolution information of the first motion vector MV1.
Bias_MV = Bias_θ * f (1 + β) / MV_pitch

Bias_MV: Center bias component (same motion vector scale)
Bias_θ: Center bias component (angle)
f: Focal length β: Image magnification MV_pitch: MV1 pitch size

  In addition, since a delay time occurs until the first motion vector MV1 is detected, it is preferable that the center bias component also has a delay time equivalent to that of the first motion vector MV1. For example, if there is a delay time of 3 frames at 30 [fps], the delay is about 100 [ms]. Therefore, the center bias component included in the first motion vector MV1 can be calculated more accurately by using the bias information before 100 [ms].

The center bias component calculated in S012 is subtracted from the first motion vector MV1 (S013). Thereby, the information of the second motion vector MV2 due to the reference value error can be acquired.
A difference: MV_diff between the latest second motion vector MV2 (n) and the second motion vector MV2 (n-1) one frame before is acquired (S014).
An amount for correcting the reference value is set based on MV_diff. As the reference value, a correction amount is set based on the following idea (S015).

MV_diff> 0: ω0_comp = −ω0_comp_def
MV_diff <0: ω0_comp = + ω0_comp_def
MV_diff = 0: ω0_comp = 0

ω0_comp: reference value correction amount ω0_comp_def: reference value correction constant

  Ω0_comp calculated in S015 is subtracted from the reference value calculated in S003 (S016). Specifically, the value of V4 ′ in FIG. 4B is corrected.

FIG. 6A is a graph showing the relationship between the reference value obtained by correcting the value of V4 ′ in FIG. 4B and the elapsed time. A dotted line in the figure indicates a reference value when correction is not performed according to the present embodiment, and a solid line in the figure indicates a reference value when correction is performed according to the present embodiment.
FIG. 6B is a graph showing the relationship between MV_diff and elapsed time.
For example, when MV_diff is in the plus direction as shown at time t1 in FIG. 6B, the reference value is corrected to minus as shown in (a).

  From the position of time 0, there is a motion vector output (positive direction) in the second frame (T = 2), and similar outputs are also in the third, fourth, and fifth frames. Further, there is no motion vector output in the sixth and seventh frames, and there is a reverse MV_diff output (one pixel in the negative direction) in the eighth frame.

  The result of calculating the corrected reference value using these MV_diff information is the graph of FIG. Since the MV_diff information is not obtained from the time 0 position until the second frame, the reference value of the angular velocity is output as it is.

Next, the reference value of the angular velocity is corrected from the information of MV_diff in the second frame (since MV_diff is in the positive direction, a fixed value is subtracted from the reference value), and the corrected reference value is output. This value is held as a constant value until the next (third frame) MV_diff information is acquired.
Note that the fixed value to be subtracted is not limited to this, and this value can be obtained through experiments or the like.

  Similarly, after the next frame, the reference value is corrected by the information of MV_diff. That is, if MV_diff is in the positive direction, the fixed value is subtracted. If MV_diff is in the negative direction, the fixed value is added. If MV_diff is 0, the value is continued. The corrected angular velocity reference value is held during each frame.

  As described above, in the present embodiment, the reference value is corrected in accordance with the cycle in which MV_diff is obtained, that is, the motion vector detection cycle. The reference value is held as a constant value between frames. For this reason, the lens CPU 2A is not subjected to the calculation load of the reference value in the reference value calculation unit 34, and can quickly perform calculations for removing the center bias component in the center bias removal unit 38. As a result, the accuracy of the calculation result of the reference value in the next frame is improved.

Note that when the motion vector detection accuracy is good, the calculation of the reference value in the reference value calculation unit 34 may be stopped. Then, it becomes possible to calculate the corrected angular velocity reference value of the current frame only by using the reference value at the previous frame, the motion vector of the current frame, and the center bias amount of the current frame, thereby reducing the calculation load on the lens CPU 2B. it can.
If the motion vector detection accuracy is poor, the corrected angular velocity reference value may be calculated again using the angular velocity reference value calculation and the motion vector again after a certain period of time without stopping the angular velocity reference value calculation.

  FIG. 7 is a graph comparing the calculation errors when the correction for the reference value according to the present embodiment is performed (solid line) and when the correction for the reference value is not performed (dotted line). As shown in the figure, according to the present embodiment, the calculation error for the true value is reduced.

The reference value is corrected using the motion vector only during the exposure preparation period (during half-push vibration isolation and during moving image shooting), and the reference value is set to a constant value during exposure.
As shown in the figure, according to the present embodiment, the reference value correction is performed only during the exposure preparation period, but the exposure can be started in a state where the reference value immediately before the exposure is closer to the true value. In this case, it is possible to perform good blur correction.

(Second Embodiment)
FIG. 8 is a view corresponding to FIG. 6 showing a second embodiment of the present invention. In the second embodiment, the motion vector calculation is increased from 1 frame interval to 2 frame intervals and 3 frame intervals as time passes. Since other processes are the same as those in the first embodiment, the description thereof is omitted.

When the accuracy of the reference value is increased, the size of the motion vector is reduced, and there is a possibility that the motion vector detection resolution or lower (1 pixel in the embodiment) is obtained. In this case, since it becomes impossible to detect a motion vector, the number of frames serving as a motion vector calculation interval is increased in order to secure a motion vector size greater than the detection resolution.
When the reference value approaches the true value and the center bias component is small, the motion vector is “reference value error + center bias component”, and thus the size of the motion vector decreases.
A motion vector between one frame is used until the reference value is stabilized, and a motion vector between separated frames is used after stabilization. This is because the convergence of the reference value can be accelerated and the accuracy of the stabilized reference value can be ensured.

(Deformation)
The present invention is not limited to the above-described embodiment, and various modifications and changes as described below are possible, and these are also within the scope of the present invention.
(1) In the present embodiment, the mode of performing the blur correction by moving the blur correction lens group has been described, but the present invention is not limited to this. For example, blur correction may be performed by moving the image sensor, or blur correction may be performed by moving the blur correction lens group and the image sensor.
(2) Further, for example, the reference value correction may not be performed under the following conditions.
When the blur correction lens exceeds the movable end (calculation limit value) This is because the reference value cannot be corrected because a camera shake component is added to the motion vector information.
When it is determined that subject blur is included in the motion vector information, this is because the reference value error and subject blur cannot be separated.
-When it is determined that the reliability of the motion vector information is low Note that the reliability determination is omitted because it is a known technique.
・ During composition change determination / tripod determination In addition, although embodiment and a deformation | transformation form can also be used in combination suitably, detailed description is abbreviate | omitted. Further, the present invention is not limited to the embodiment described above.

  1: camera, 1A: camera housing, 1B: lens barrel, 2A: camera CPU, 2B: lens CPU, 3: image sensor, 4: zoom lens group, 5: focus lens group, 6: blur correction lens group, 7: Zoom lens group drive mechanism, 8: Focus lens group drive mechanism, 9: Blur correction lens group drive mechanism, 12: Angular velocity sensor, 21: Lens position detection unit, 31: Amplification unit, 32: First A / D conversion unit 33: second A / D conversion unit, 34: reference value calculation unit, 35: center bias removal unit, 35: reference value correction amount calculation unit, 36: target position calculation unit, 37: center bias calculation unit, 38: center Bias removal unit, 39: drive amount calculation unit, 40: signal processing unit, 41: vector calculation unit, 42: reference value correction unit, 43: subtraction unit, 100, 200: blur correction device

Claims (5)

An image sensor that captures an image of a subject formed by the optical system and outputs a first signal;
A detection unit that detects the movement of the optical device having the image sensor and outputs a second signal;
A shake correction unit driven to correct the motion detected by the detection unit;
Wherein the detection by the detection unit is initiated until the first hour has elapsed, it calculates the motion vector from the image generated from the first signal output at intervals of second time of the first signal Then, after the first time has elapsed since the detection by the detection unit started, the first signal is generated from the first signal output at a third time interval longer than the second time. A motion vector computing unit for computing the motion vector from the image to be processed;
A drive amount calculation unit that calculates a drive amount of the shake correction unit based on the motion vector and the second signal;
An image stabilization apparatus comprising: The blur correction device according to claim 1,
The drive amount calculation unit corrects the reference value of the second signal calculated from the second signal based on the motion vector, and the correction amount of the shake correction unit is calculated from the corrected reference value and the second signal. Blur correction device that calculates drive amount. The blur correction device according to claim 2,
The drive amount calculation unit is a shake correction apparatus that holds the reference value corrected until the motion vector is updated after the reference value is corrected based on the motion vector. The blur correction device according to claim 3,
The drive amount calculation unit is a shake correction apparatus that corrects the reference value and then stops calculating the reference value based on the second signal.   The camera which has the blurring correction apparatus as described in any one of Claim 1- Claim 4.

Images