JP6420888B2 - Image blur correction apparatus, control method therefor, program, and storage medium - Google Patents

Description

  The present invention relates to a technique for correcting an image shake of a captured image caused by a shake of an image pickup apparatus using a method of deforming an image in the image pickup apparatus.

  In recent years, with the advancement of technology for correcting shakes occurring in an imaging apparatus, not only the image shake of a captured image caused by hand shake when the photographer is stationary, but also the photographer performs shooting while walking. An image blur correction function for correcting image blur of a captured image that occurs sometimes has become widespread. When the photographer takes a picture while walking, not only the image shake in the horizontal and vertical directions of the picked-up image but also the image shake as shown below occurs on the picked-up image. For example, image shake in which a captured image rotates as the imaging apparatus rotates around the optical axis, or image blur in which the captured image is distorted in a trapezoidal shape when the imaging apparatus is tilted with respect to a subject.

  As a method of correcting various image shakes that occur in a captured image during shooting while walking, there is a known method of calculating the image deformation amount of the captured image and deforming the image so as to cancel the image deformation amount. (See Patent Document 1). Further, as a method for calculating various image deformation amounts of a captured image, a method of calculating an image deformation amount using a motion vector obtained from the captured image in addition to a sensor such as an angular velocity sensor is known (Patent Literature). 2).

JP 2011-146260 A JP 2009-124597 A

  In the technique described in Patent Document 1, the amount of deformation generated in a captured image due to shake applied to the image capturing apparatus is decomposed into deformation components of translation (horizontal / vertical), tilt (horizontal / vertical), enlargement / reduction, rotation, and shear. To do. And filtering processing etc. are performed about each and the projection transformation matrix (homography matrix) is calculated. However, since a filtering process or the like is performed for each shake component added to the imaging apparatus, the amount of calculation is very large and the process is complicated.

  In the technique described in Patent Document 2, a correction amount for rotation is calculated using a sensor such as an angular velocity sensor, and a correction amount for translation is calculated using a motion vector. However, under shooting conditions in which motion vectors cannot be detected correctly, the accuracy of detection of image blur caused by translational motion is deteriorated, resulting in degradation of video quality. For example, depending on the subject, the contrast of the captured image may be extremely low, and in such a case, the detection accuracy of the motion vector may deteriorate.

  The present invention has been made in view of the above-described problems, and an object of the present invention is to provide an image pickup apparatus that can obtain a good image blur correction effect with a simple configuration.

Image blur correction device according to the present invention, a motion vector detecting means for detecting a motion vector between two images among the images captured by an imaging means, of the prior SL two images, the image of the previous frame a was imaged until an image of the next frame, based on the output of the angular velocity detecting means for detecting a shake, shake calculates the shake amount of the imaging device between to shoot the two images calculated A result obtained by adding the output of the image blur conversion unit, the output of the image blur conversion unit, the output of the motion vector detection unit, and the output of the image blur conversion unit at a predetermined ratio. based on a correction amount calculating means for calculating a correction amount for correcting the image blur amount of the image electronically, based on the output of the correction amount calculating means, for electronically correcting the image blur of the image control to control the correction means And means, and the moving image shooting, the correction amount calculation unit, when the imaging apparatus is determined to be a panning or tilting state, the imaging device is determined not to be a panning or tilting state The ratio of adding the output of the motion vector detecting means is made smaller than that in the case of the above .

  According to the present invention, it is possible to provide an imaging apparatus that can obtain a good image blur correction effect with a simple configuration.

It is a block diagram which shows the structural example of the video camera as an example of the imaging device which concerns on the 1st Embodiment of this invention. It is a figure explaining a pinhole camera model. It is a block diagram which shows the structure of the image deformation amount calculating part in 1st Embodiment. It is a figure for demonstrating the motion vector detection part in 1st Embodiment. It is a block diagram which shows the structure of the shake calculation part between frames in 1st Embodiment. It is a block diagram which shows the structure of the translation synthetic | combination part in 1st Embodiment. It is a figure for demonstrating operation | movement of the translation synthetic | combination part in 1st Embodiment. It is a block diagram which shows the structural example of the video camera as an example of the imaging device which concerns on the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the image deformation amount calculating part in 2nd Embodiment. It is a figure for demonstrating operation | movement of the translation synthetic | combination part in the 3rd Embodiment of this invention. It is a figure for demonstrating the definition of "shake" in embodiment of this invention. It is a figure for demonstrating the definition of "image blur" in embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  First, definitions of terms used in the embodiment of the present invention will be described. In the description of the embodiment of the present invention, the vibration applied to the imaging device is referred to as “shake”, and the deformation of the captured image caused by the shake applied to the imaging device is referred to as “image shake”. In the embodiment of the present invention, as shown in FIGS. 11 (a) and 11 (b), the “runout” includes three types of “angle runout” in the YAW, PITCH, and ROLL directions, and axes orthogonal to each other (horizontal axis). , Vertical axis, optical axis), which is a general term for three types of “parallel shake” (shift shake) in the horizontal, vertical, and optical axis directions. The YAW direction is a direction around the vertical axis. PITCH is the direction around the horizontal axis. The ROLL direction is a direction around the optical axis. “Image blur” is a generic name for deformation components of translation (horizontal / vertical), rotation, tilt (horizontal / vertical), enlargement / reduction, and shear as shown in FIGS. 12 (a) to 12 (f). Means.

(First embodiment)
FIG. 1 is a block diagram showing a configuration of a video camera as an example of an imaging apparatus according to the first embodiment of the present invention. Hereinafter, each component of the imaging apparatus 100 in FIG. 1 and an example of the operation thereof will be described in detail.

  The angular velocity sensor 102 detects a shake applied to the imaging apparatus 100 as an angular velocity signal, and supplies the angular velocity signal to the A / D converter 103. The A / D converter 103 digitizes the angular velocity signal from the angular velocity sensor 102 and supplies it as angular velocity data to the inter-frame shake calculation unit 300 (image blur conversion unit) in the μCOM 101. The inter-frame shake calculation unit 300 calculates the amount of angular shake applied to the imaging apparatus between frames of moving images that are sequentially captured, and calculates the amount of deformation that electronically corrects the image shake of the image as angle data. The amount is supplied to the amount calculation unit 200 (correction amount calculation unit). Details of the processing of the inter-frame shake calculation unit 300 will be described later.

  The imaging optical system 120 performs an operation such as zooming by moving the zoom lens and focusing by moving the focus lens, and forms an object image on the image sensor 123. A zoom encoder 119 serving as a focal length detection unit detects the position (zoom position) of the variable magnification optical system 121 in the imaging optical system 120 and outputs it to the image deformation amount calculation unit 200 inside the μCOM 101.

  The imaging element 123 converts the subject image formed by the imaging optical system 120 into an electrical signal as a captured image signal, and supplies the electrical signal to the signal processing unit 124. The signal processing unit 124 generates a video signal (video signal) based on, for example, the NTSC format from the signal obtained by the image sensor 123 and supplies the video signal to the image memory 125. The motion vector detection unit 126 is stored in the image memory 125 and two captured images that are temporally preceding and following frames, specifically, the luminance signal included in the current video signal generated by the signal processing unit 124. The motion vector of the image is detected based on the luminance signal included in the video signal one frame before. The motion vector data detected by the motion vector detection unit 126 is supplied to the image deformation amount calculation unit 200.

  The image deformation amount calculation unit 200 uses the angle data, the motion vector data, and the output of the zoom encoder 119 described above to calculate the image deformation amount for correcting the image shake of the captured image, and the image deformation unit 127 calculates the image deformation amount. Sets the amount of image deformation. Details of the processing of the image deformation amount calculation unit 200 will be described later.

  The image deforming unit 127 corrects image shake of the captured image by deforming the image stored in the image memory 125 based on the image deformation amount calculated by the image deformation amount calculating unit 200, and the recording control unit 128 and The data is output to the display control unit 130. The display control unit 130 outputs the video signal supplied from the image transformation unit 127 and causes the display device 131 to display an image. The display control unit 130 drives the display device 131, and the display device 131 displays an image by a liquid crystal display element (LCD or viewfinder) or the like.

  In addition, the recording control unit 128 outputs the video signal supplied from the image transformation unit 127 to the recording medium 129 when the recording of the video signal is instructed by an operation unit (not shown) used for recording start and end instructions. Let me record. The recording medium 129 is an information recording medium such as a semiconductor memory, or a magnetic recording medium such as a hard disk or a magnetic tape.

  Next, image deformation performed by the image deformation unit 127 will be described in detail.

  The image transformation unit 127 performs image transformation using geometric transformation such as projective transformation. Specifically, the pixel coordinates in the untransformed image (image stored in the image memory 125) are (X0, Y0) (however, the center of the captured image corresponding to the optical axis of the imaging optical system 120 is the origin). ), And the pixel coordinates in the transformed image (the output image of the image transformation unit 127) are represented by (X1, Y1) and expressed in homogeneous coordinates, it can be expressed as (Equation 1).

  The left side and the right side of (Formula 1) show an equivalence relationship (the meaning does not change even if an arbitrary magnification is applied to the left side or the right side), and in the normal equal sign, they are (Formula 2) and (Formula 3).

  In (Expression 1), a 3 × 3 matrix is generally called a projective transformation matrix, and the elements h1 to h8 of the matrix are set by the image deformation amount calculation unit 200. In the following description, the image transformation of the image transformation unit 127 is described as using projective transformation, but any transformation method such as affine transformation may be used.

  Next, details of processing performed by the image deformation amount calculation unit 200 will be described. The image deformation amount calculation unit 200 uses the shake angle of the imaging device calculated from the output of the angular velocity sensor 102 and the focal length of the imaging optical system 120 calculated from the zoom encoder 119 to perform image deformation of the image deformation unit 127. Calculate the amount. Specifically, the projective transformation matrix of (Expression 1) is calculated.

  Here, a calculation method of the projective transformation matrix using the shake angle and the focal length of the imaging optical system 120 will be described below.

  FIG. 2A illustrates the projection of the subject on the imaging surface by the imaging apparatus using a pinhole camera model. In FIG. 2A, the origin (0, 0, 0) of the XYZ space coordinates is the pinhole position in the pinhole camera model. If the imaging surface is arranged behind the pinhole position, the image projected on the imaging surface is inverted, so that the image is virtually pinned in FIG. The imaging surface I is arranged before the hole position.

  The distance in the Z-axis direction between the origin (0, 0, 0) of the XYZ space coordinates and the imaging surface I is the focal length f. Coordinates on the imaging plane I are defined as uv plane coordinates, and the origin (0, 0) of the uv plane coordinates is assumed to coincide with (0, 0, f) in the XYZ space coordinates. The coordinates P (u, v) on the uv plane coordinates are coordinates when the subject A (X, Y, Z) on the XYZ space coordinates is projected onto the imaging plane I. At this time, the coordinate P can be expressed by (Formula 4).

  (Expression 4) can be expressed by (Expression 5) when homogeneous coordinates are used.

  Since the element in the fourth column of the 3 × 4 matrix of (Expression 5) remains 0 in the description of this embodiment, (Expression 5) is the same as (Expression 6).

  FIG. 2B shows the pinhole camera model shown in FIG. 2A when the image is rotated R (rotation shake R is generated in the image pickup apparatus, and parallel shake that is parallel movement of the image pickup apparatus is not generated). is there. In FIG. 2B, coordinates obtained by rotating the XYZ space coordinates in FIG. 2A by R are defined as X′Y′Z ′ space coordinates. It is assumed that the origin (0, 0, 0) of the X′Y′Z ′ space coordinates coincides with the XYZ space coordinates. That is, FIG. 2B is a simplified representation of a state in which a rotational shake R occurs in the imaging apparatus and no parallel shake, which is a parallel movement of the imaging apparatus, using a pinhole camera model.

  In the pinhole camera model of FIG. 2B, the imaging plane I ′ is arranged at a focal distance f at a distance from the origin (0, 0, 0), as in FIG. 2A. The coordinates on the imaging plane I ′ are defined as u′v ′ plane coordinates, and the origin (0,0) of the u′v ′ plane coordinates is (0,0, f) in the X′Y′Z ′ space coordinates. It shall be consistent with The coordinates P ′ (u ′, v ′) on the u′v ′ plane coordinates are the same as the subject A ′ (X ′, Y ′, Z ′) on the X′Y′Z ′ space coordinates on the imaging plane I ′. The coordinates when projected. Note that the position of the subject A in FIG. 2A and the subject A ′ in FIG. 2B in the world coordinate system is the same position (that is, the subject has not moved). At this time, the coordinate P ′ can be expressed by (Expression 7) similarly to (Expression 6) when homogeneous coordinates are used.

  Further, since the positions of the subject A and the subject A ′ in the world coordinate system are the same, the relationship between the coordinates of both can be expressed by (Equation 8).

  Furthermore, when (Expression 6) and (Expression 7) are modified and substituted into (Expression 8), (Expression 9) can be derived.

  (Equation 9) is the position of the subject image on the imaging surface before and after the pinhole camera makes R rotation (rotation shake R occurs in the imaging apparatus, and parallel translation that is parallel movement of the imaging apparatus does not occur). The correspondence relationship is shown. That is, when an R rotation shake is applied to the imaging apparatus, the equation indicates where the pixel on the imaging surface moves from where. Therefore, in order to correct the image blur, it is only necessary to perform conversion to restore the pixel movement amount when the image pickup apparatus is shaken. That is, according to (Equation 10), it is only necessary to perform conversion to restore the pixel position that has moved due to the addition of the R rotation shake to the imaging apparatus.

  Therefore, assuming that the shake applied to the image pickup apparatus 100 in FIG. 1 is R, the focal length of the image pickup optical system 120 is f, and the projective transformation matrix for performing image shake correction is H, H becomes (Expression 11).

  Note that the angular shake amount in the YAW direction, which is a shake around the first direction on the plane perpendicular to the optical axis applied to the imaging device, is θy, in a direction orthogonal to the first direction on the plane perpendicular to the optical axis. If the angular shake amount in the PITCH direction, which is a shake around a second direction, is θp, and the angular shake amount in the ROLL direction is θr, R can be expressed by (Equation 12).

  H in (Expression 11) is decomposed into deformation components of translation t →, scaling s (constant), rotation r (matrix), shear k (matrix), and tilt v → by using (Expression 13). be able to.

here,
tx… horizontal translation amount ty… vertical translation amount θ… rotation angle vx… horizontal tilt amount vy… vertical tilt amount α… shear anisotropic angle φ… shear direction angle.

  From (Expression 11), (Expression 12), and (Expression 13), solving the equations for each deformation component yields (Expression 14) to (Expression 21).

  Here, in the present embodiment, the processing of the image deformation amount calculation unit 200 simplifies the calculation as described below. The angular shake applied to the imaging apparatus does not become so great except for a movement such as panning in which the user intentionally operates the imaging apparatus. Therefore, when the shake angle applied to the imaging device is γ, it can be approximated as cos γ = 1, sin γ tan γ = 0, sin γ sin γ = 0, and (Expression 14) to (Expression 21) are expressed by (Expression 22) to (Expression 29). Can be expressed as

  At this time, the enlargement / reduction is 1 time from (Expression 27), and the anisotropic magnification of shear is 1 time and the direction angle is 0 degree from (Expression 28) and (Expression 29). That is, it can be seen that the deformation component generated in the captured image due to the angular shake applied to the image capturing apparatus is smaller than that of the other deformation components.

  Therefore, in the processing of the image deformation amount calculation unit 200 in the present embodiment, correction of each deformation component of enlargement / reduction and shear is not performed, but only translation, rotation, and tilt correction are performed.

  Hereinafter, the components of the image deformation amount calculation unit 200 and the inter-frame shake calculation unit 300 and the operation of an example thereof will be described in detail with reference to the block diagrams of FIGS. 3 and 4.

  First, a block for calculating a motion vector of a captured moving image will be described. In the block diagram of FIG. 3, the motion vector detection unit 126 detects a local motion vector between frame images. At this time, as the detection position of the local motion vector, for example, as shown in FIG. 4, the detection blocks 401 are arranged in a grid of 6 blocks vertically and 10 blocks horizontally. Then, a local motion vector is calculated for each detection block 401. Since the motion vector can be calculated using a known method such as template matching or gradient method, a detailed description thereof will be omitted.

  The local motion vector detected for each block is supplied to the motion vector processing unit 201 to determine a representative vector which is a motion vector value of the entire screen. Specifically, the median value or average value of local motion vector values is used as the representative vector value of the entire image.

  At this time, the motion vector processing unit 201 obtains the reliability (scale of accuracy) of the representative motion vector value. Specifically, the motion vector processing unit 201 obtains a variance value of the local motion vector value, and determines the reliability of the representative motion vector value based on the variance value. When the variance value is small (that is, when the local motion vector is not dispersed), the motion vector processing unit 201 determines that the reliability of the representative motion vector value is high. Conversely, when the variance value is large (that is, when the local motion vector is dispersed), the motion vector processing unit 201 determines that the reliability of the representative motion vector value is low. This is based on the fact that when the motion direction of the video signal is uniform, the local motion vector value in units of blocks obtained from the image is a uniform vector amount. Note that, in the motion vector processing unit 201, a reference value (reference value of a variance value of local motion vectors) serving as a criterion for determining reliability is determined in advance. The determination result of the reliability by the motion vector processing unit 201 and the representative motion vector value are supplied to the translation synthesis unit 203. In addition, although the method of calculating the reliability of the motion vector based on the variance value of the local motion vector has been described as an example, any method may be used as long as the reliability of the motion vector can be evaluated.

  Next, a block for processing an output signal of the angular velocity sensor will be described. Note that the angular deflection in the YAW direction causes horizontal translation and horizontal tilt from (Equation 22) and (Equation 25), and the vertical deflection from (Equation 23) and (Equation 26) due to the angular deflection in the PITCH direction. Translation, vertical tilt occurs. However, since both processes are the same, only the control of either one will be described in the following description.

  FIG. 5 is a block diagram illustrating the configuration of the interframe shake calculation unit 300. In the block diagram of FIG. 5A, the angular velocity data supplied from the A / D converter 103 is integrated by the integrator 302. The integrator 302 outputs angle data that is an integration result of the angular velocity data. The timing generation unit 301 outputs a signal synchronized with the timing at which the frame images constituting the moving image are captured to the difference calculation unit 303. The difference calculation unit 303 receives the signal from the timing generation unit 301 and holds the output data of the integrator 302. Then, the difference calculation unit 303 outputs the difference from the previously held value, that is, the amount of angular shake of the imaging device from the time when the previous frame image was captured to the time when the current frame image was captured.

  As another example of the inter-frame shake calculation unit 300, it may be as shown in FIG. That is, a configuration may be adopted in which the signal output from the timing generation unit 301 is directly supplied to the integrator 302 and the amount of angular shake between frames is acquired by resetting the integration value.

  In this way, the inter-frame shake calculation unit 300 calculates the angular shake amount of the imaging apparatus from the time when the previous frame image was captured to the time when the current frame image was captured, and supplies the calculated amount to the image deformation amount calculation unit 200. To do.

  In the block diagram of FIG. 3, the amount of angular shake between frames, which is the output of the interframe shake calculation unit 300, is supplied to the translation conversion unit 202. The translation conversion unit 202 converts the angular shake amount between frames into a translational image shake amount on the image. Specifically, the translation conversion unit 202 multiplies the angular shake amount between one frame, which is the output of the inter-frame shake calculation unit 300, by the focal length f of the imaging optical system 120 supplied from the focal length calculation unit 204. This is converted into a translational image blur amount on an image generated during one frame. The calculation method of multiplying the focal length f is based on the translation calculation formulas of (Formula 22) and (Formula 23). The focal length f can be calculated by the focal length calculation unit 204 from the output of the zoom encoder 119 based on the output of the zoom encoder 119.

  By the way, the representative motion vector that is the output of the motion vector processing unit 201 is the translational image blur amount itself on the screen between frames. Therefore, by calculating the amount of angular shake between frames from the output of the angular velocity sensor as described above, the translational image shake amount obtained from the angular velocity sensor and the motion vector supplied from the motion vector detecting unit 126 are the same. Units can be handled as time series data on the same time axis. Hereinafter, in order to distinguish between the translational image blur amount obtained from the angular velocity sensor and the translational image blur amount obtained from the motion vector, the former is referred to as sensor translational image blur and the latter is referred to as vector translational image blur.

  The translation synthesis unit 203 is supplied with the sensor translational image blur as an output of the translation conversion unit 202, the representative motion vector (vector translational image blur) as the output of the motion vector processing unit 201, and its reliability. Finally, translational image blur data used for calculating the correction amount is obtained.

  FIG. 6 is a block diagram for explaining a specific operation of the translation synthesis unit 203. The vector translational image blur data supplied from the motion vector processing unit 201 is supplied to the multiplier 242 and multiplied by a predetermined magnification. The sensor translational image blur data supplied from the translation conversion unit 202 is supplied to the multiplier 243 and multiplied by a predetermined magnification. The output of the multiplier 242 and the output of the multiplier 243 are added at a ratio based on the above magnification by the adder 244, and finally output as translational image blur data for use in calculating the correction amount. The control unit 241 is supplied with the reliability of the representative motion vector, which is the output of the motion vector processing unit 201, and sets the multipliers of the multiplier 242 and the multiplier 243 according to the reliability. Assuming that the magnification set in the multiplier 242 is K1, and the magnification set in the multiplier 243 is K2, K1 and K2 are set to values that satisfy the following expressions.

0 ≦ K1 ≦ 1 (Formula 30)
0 ≦ K2 ≦ 1 (Formula 31)
(K1 + K2) = 1 (Expression 32)
FIG. 7 is a graph showing the characteristics of K1 and K2 determined by the control unit 241. When the reliability is smaller than the threshold value 1, a magnification is set such that K1 <K2. When the reliability is between the threshold value 1 and the threshold value 2, it is set so that K1 increases (K2 decreases) as the reliability increases. When the reliability is larger than the threshold value 2, a magnification that satisfies K1> K2 is set. Thus, when the reliability of the motion vector is high, the ratio of the translational image shake amount obtained from the motion vector is increased. On the contrary, when the reliability of the motion vector is low, the ratio of the translational image blur data obtained from the motion vector is reduced, and instead, the ratio of the translational image blur data obtained from the angular velocity sensor is increased.

  Returning to FIG. 3 again, the blocks 211 to 214 for calculating the translational image blur correction amount will be described. The HPF 211 is supplied with the translational image blur data in the YAW direction or the PITCH direction among the outputs from the translation synthesis unit 203 described above. The HPF 211 has a function of changing its characteristics in an arbitrary frequency band, and outputs a signal in a high frequency band by cutting off a low frequency component included in the translational image blur. The HPF 211 is not an essential component in the present embodiment, and the output from the translation synthesis unit 203 may be directly supplied to the centering unit 212.

  The centering unit 212 performs a process of adding an input value (hereinafter referred to as a centering amount) that returns the correction amount to zero with respect to the output of the HPF 211 in the YAW direction or the PITCH direction of the imaging apparatus 100. The processing of the centering unit 212 is performed when panning or a large angular shake that cannot be corrected occurs. The centering unit 212 is not essential in the present embodiment, and the output from the HPF 211 may be directly supplied to the integrator 213.

  The integrator 213 has a function capable of changing the characteristics in an arbitrary frequency band, integrates the output from the centering unit 212, and supplies it to the saturation prevention control unit 214. The saturation prevention control unit 214 performs control to limit the output from the integrator 213 so that the output from the integrator 213 does not exceed a predetermined value (hereinafter referred to as a limit value). Further, when the output from the integrator 213 approaches the limit value, the saturation prevention control unit 214 changes the cutoff frequency of the HPF 211 to the high frequency side, shortens the time constant of the integrator 213, Control such as increasing the centering amount of the unit 212 is performed. By the processing of the saturation prevention control unit 214, it is possible to block a low frequency band motion component generated by panning, and to prevent a trouble that hinders the panning operation of the photographer. Specifically, by making the correction means (the image deforming unit 127 and the correction optical system 122) easily move toward the center, the degree of image shake correction with respect to the detected shake is reduced (followability of image shake correction is reduced). Become. The output of the saturation prevention control unit 214 becomes the final translation correction amount and is supplied to the image deformation amount combining unit 205.

  Next, the blocks 220 to 224 for calculating the correction amount of the tilt image blur will be described. The blocks 220 to 224 are the same as the blocks 211 to 214 except for the processing of the tilt conversion unit 220. Therefore, only the tilt conversion unit 220 will be described, and other blocks will be described. Description is omitted.

  Of the outputs from the above-described inter-frame shake calculation unit 300, the tilt conversion unit 220 is supplied with angular shake data in the YAW direction or the PITCH direction. The tilt conversion unit 220 divides the angular shake data by the focal length f calculated by the focal length calculation unit 204 and supplies the divided data to the HPF 221. The division by the focal length f is based on the tilt calculation formula of (Formula 25) and (Formula 26). The output of the saturation prevention control unit 224 becomes the final tilt correction amount and is supplied to the image deformation amount combining unit 205.

  Next, blocks 231 to 234 for calculating the correction amount of the rotational image blur around the optical axis will be described.

  The HPF 231 is supplied with the angular shake data in the ROLL direction among the outputs from the inter-frame shake calculation unit 300 described above. The HPF 231 has a function of changing its characteristics in an arbitrary frequency band, and outputs a signal in a high frequency band by cutting off a low frequency component included in the angular shake data. The HPF 231 is not an essential component in the present embodiment, and the output from the inter-frame shake calculation unit 300 may be directly supplied to the centering unit 232.

  The centering unit 232 performs a process of adding the centering amount to the output of the HPF 231 in the same way as the centering units 212 and 222 when a large angular shake that cannot be corrected occurs in the ROLL direction of the imaging apparatus 100. Note that the centering unit 232 is not an essential component in the present embodiment, and the output from the HPF 231 or the inter-frame shake calculation unit 300 may be directly supplied to the integrator 233.

  The integrator 233 has a function of changing its characteristics in an arbitrary frequency band, integrates the output from the centering unit 232, and supplies it to the saturation prevention control unit 234. The saturation prevention control unit 234 performs control to limit the output from the integrator 233 so that the output from the integrator 233 does not exceed a predetermined limit value. Further, when the output from the integrator 233 approaches the limit value, the saturation prevention control unit 234 changes the cutoff frequency of the HPF 231 to the high frequency side, shortens the time constant of the integrator 233, Control such as increasing the centering amount of the unit 232 is performed. The output of the saturation prevention control unit 234 becomes the final rotation correction amount and is supplied to the image deformation amount combining unit 205.

  The image deformation amount combining unit 205 performs a calculation for combining the translation, tilt, and rotation correction amounts output from the saturation prevention control units 214, 224, and 234. Specifically, the projective transformation matrix of (Expression 1) is calculated according to (Expression 13). At this time, the deformation amounts of enlargement / reduction and shearing are the values shown in (Expression 27), (Expression 28), and (Expression 29). The image deformation amount synthesis unit 205 outputs the calculated values of the elements of the projective transformation matrix to the image deformation unit 127. The image deformation unit 127 performs image blur correction by image deformation based on the output from the image deformation amount combining unit 205.

  As described above, in the first embodiment of the present invention, the translation composition unit 203 that adds the translational image shake amount calculated from the angular shake amount generated between frames and the motion vector at a predetermined ratio is provided. And a configuration for correcting the translational image blur based on the result. As a result, the translation correction amount obtained from the angular velocity sensor and the translation correction amount obtained from the motion vector can be shared, and good image blur correction performance can be realized with a simple configuration.

  Moreover, the translation composition unit 203 is configured to change the addition ratio according to the reliability of the motion vector. As a result, when the reliability of the motion vector is low, it is possible to increase the ratio of the translational image shake amount obtained from the angular velocity sensor, and the image quality is deteriorated even in a shooting situation where the motion vector is not correctly detected. Can be minimized.

(Second Embodiment)
FIG. 8 is a block diagram showing a configuration of a video camera as an example of an imaging apparatus according to the second embodiment of the present invention. In FIG. 8, the same components as those in FIG. 8 includes a correction optical system 122 that optically corrects image shake of a captured image by moving in a direction perpendicular to the optical axis, and a block for controlling the correction optical system 122 in addition to the configuration of FIG. It has been configured.

  The output of the A / D converter 103 is supplied to the inter-frame shake calculation unit 300 and is also supplied to the HPF 104 in the configuration of FIG. The HPF 104 has a function of changing the characteristics in an arbitrary frequency band, and outputs a signal in the high frequency band by cutting off the low frequency component included in the angular velocity data from the A / D converter 103. When a large angular shake such as panning occurs in the imaging apparatus 100, the cutoff frequency of the HPF 104 is set to a higher value than when the angular shake is small.

  The sensitivity calculation unit 105 calculates a coefficient for calculating the optimum driving amount of the correction optical system 122 at each zoom position according to the output of the zoom encoder 119, and multiplies the output of the HPF 104 to the integrator 106. And supply. The integrator 106 has a function capable of changing its characteristics in an arbitrary frequency band, integrates the output from the sensitivity calculation unit 105, and supplies it to the saturation prevention control unit 107. When a large angular shake such as panning occurs in the imaging apparatus 100, the time constant of the integrator 106 is set to a shorter value than when the angular shake is small.

  When the output of the integrator 106 becomes larger than a predetermined limit value, the saturation prevention control unit 107 limits the output of the integrator 106 so as to be suppressed within the limit value, and supplies the limited output to the subtractor 108. To do.

  The position detection unit 117 includes a magnet and a hall sensor provided at a position facing the magnet, and detects the amount of movement of the correction optical system 122 in a direction perpendicular to the optical axis. The A / D converter 118 converts the position of the correction optical system 122 detected by the position detection unit 117 from an analog signal to a digital signal (A / D conversion). The subtracter 108 subtracts the digitized position data of the correction optical system 122 from the output of the saturation prevention control unit 107, and supplies the resulting deviation data to the control filter 109. Here, the terminal B114 and the terminal B116 indicate that each is electrically connected.

  The control filter 109 includes an amplifier that amplifies input data with a predetermined gain, and a phase compensation filter. After the signal processing by the amplifier and the phase compensation filter is performed in the control filter 109, the deviation data supplied from the subtractor 108 is output to the pulse width modulation unit 110.

  The pulse width modulation unit 110 modulates the data supplied through the control filter 109 into a waveform (that is, a PWM waveform) that changes the duty ratio of the pulse wave, and supplies the modulated data to the motor driving unit 111. The motor 112 is a voice coil type motor for driving the correction optical system 122, and is driven by the motor drive unit 111 to move the correction optical system 122 in a direction perpendicular to the optical axis. Here, the terminal A113 and the terminal A115 indicate that each is electrically connected.

  . The above block constitutes a feedback control system that subtracts the output of the position detection unit 117 from the output of the saturation prevention control unit 107 and follows the amount of movement of the correction optical system 122 in the direction perpendicular to the optical axis. doing. The output of the A / D converter 118 is supplied to the image deformation amount calculation unit in addition to the subtractor 108.

  The correction optical system 122 is a shift lens, for example, and is an optical system capable of shake correction that deflects the optical axis by being moved in a direction perpendicular to the optical axis. As a result of the movement of the correction optical system 122, an image in which the image shake in the translation direction of the subject on the image pickup surface caused by the shake of the image pickup apparatus 100 is corrected is formed on the image sensor 123.

  Hereinafter, the components of the image deformation amount calculation unit 200 and the operation of an example thereof will be described in detail with reference to the block diagram of FIG. In FIG. 9, the same components as those in FIG. 3 are denoted by the same reference numerals, and description thereof is omitted. The configuration of the image deformation amount calculation unit 200 in FIG. 9 is obtained by adding a translation correction amount conversion unit 206 and a subtracter 207 to FIG.

  The translation correction amount conversion unit 206 calculates the translation correction amount on the imaging surface corrected by the correction optical system 122 by using the output of the A / D converter 118. Image blur image blur correction by the correction optical system 122 can optically correct the image blur in the translational direction among the deformation components of the image blur image blur on the imaging surface. Therefore, if the translation correction amount is calculated using all the output data from the angular velocity sensor as described with reference to FIG. 9, the translation correction amount conversion unit 206 calculates the translation correction amount of the correction optical system 122, and the subtractor 207 subtracts the translation correction amount of the correction optical system 122 from the output of the translation conversion unit 202. Thus, only the image blur other than the translational image blur and the translation correction amount that could not be corrected by the correction optical system 122 are corrected by the image deformation unit 127.

  As described above, in the second embodiment of the present invention, in addition to the configuration of the first embodiment, the correction optical system 122 is used to optically correct translational image blur. This makes it possible to further reduce the number of surplus pixels necessary for deformation in the translation direction, keep the number of pixels that can be output by the image deformation unit 127 as large as possible, and minimize deterioration in image quality. .

  In the second embodiment, the correction optical system 122 has been described as an example of the optical image blur correction unit. However, the present invention is not limited to this. For example, various image blur correction means such as a method of driving the image sensor 123 and a method of using a prism can be used.

(Third embodiment)
Next, a second method of processing executed by the translation synthesis unit 203 in FIG. 3 will be described in detail as a third embodiment with reference to the drawings. In the first embodiment, the ratio of adding the vector translational image blur and the sensor translational image blur is changed according to the reliability of the motion vector, but the reliability of the motion vector is estimated by analogy without directly evaluating the motion vector. It is possible to do. For example, when the zoom operation is performed, the local motion vector calculated by the motion vector detection unit is detected in the radial direction from the center of the screen. The representative motion vector calculated using these local motion vectors may be considered as having low reliability. In addition, in the case of shooting while panning and tilting the image pickup device greatly, the amount of movement on the image between frames increases, and the motion vector detection range may be exceeded. Even in such a case, the reliability of the motion vector may be low. In the third embodiment, a method of changing the ratio of adding vector translational image blur and sensor translational image blur using information other than motion vectors will be described.

  The configuration of the imaging apparatus in the present embodiment can be realized with the same configuration as that of FIGS. 1, 3, 5, and 6. The control unit 241 in the translation synthesis unit 203 in FIG. 6 receives the reliability of the motion vector calculated by the motion vector processing unit 201, receives the output of the A / D converter 103 and the output of the zoom encoder 119, and receives these information. Based on the above, the ratio of adding the vector translational image blur and the sensor translational image blur is changed.

  The flowchart of FIG. 10 shows an example of processing of the control unit 241. The process of the flowchart of FIG. 10 is repeatedly executed at a predetermined cycle such as 60 Hz.

  In step S100, if the variable SENSOR_OUT1 is a signal generated based on the output of the A / D converter 103 in the YAW direction or the PITCH direction, it is determined whether or not SENSOR_OUT1 is larger than the threshold AMP_TH1. The signal generated based on the output of the A / D converter 103 in the YAW direction or PITCH direction may be any signal as long as the signal is generated based on the output of the A / D converter 103. Good. For example, the output itself of the A / D converter 103 may be used, or a signal obtained by performing various filtering processes on the output of the A / D converter 103 or applying a gain or the like. If it is determined in step S100 that SENSOR_OUT1 is greater than AMP_TH1, the process proceeds to step S101.

  In step S101, the value of the counter COUNTER1 is incremented, and the process proceeds to step S102. In step S102, it is determined whether the counter COUNTER1 is greater than the threshold TIME_TH1. If it is determined in step S102 that COUNTER1 is equal to or less than TIME_TH1, the process proceeds to step S106. If it is determined in step S102 that COUNTER1 is greater than TIME_TH1, the process proceeds to step S103 and 1 is written to PAN_FLAG. The process of step S103 is performed when a shake of a predetermined time (TIME_TH1) or more and a predetermined magnitude (AMP_TH1) or more is applied in the YAW direction or the PITCH direction, that is, the process of step S103 is performed. This is when the image capturing apparatus 100 is in a panning or tilting operation (panning or tilting state).

  If it is determined in step S100 that SENSOR_OUT1 is equal to or less than AMP_TH1, the process proceeds to step S104. In step S104, the value of the counter COUNTER1 incremented in step S101 is cleared, and the process proceeds to step S105. In step S105, PAN_FLAG is cleared, and the process proceeds to step S106.

  In step S106, how much the output value of the zoom encoder 119 has changed since the last processing is obtained, stored in the variable ZOOM_SPEED, and the process proceeds to step S107. That is, in step S106, the zoom magnification changing speed when the zoom operation is performed is obtained and stored. If it is determined in step S107 that ZOOM_SPEED is greater than ZOOM_TH1, the process proceeds to step S108. In step S108, 1 is written in ZOOM_FLAG, and the process proceeds to step S110. If it is determined in step S107 that ZOOM_SPEED is equal to or less than ZOOM_TH1, the process proceeds to step S109. In step S109, ZOOM_FLAG is cleared, and the process proceeds to step S110.

  In step S110, it is determined whether PAN_FLAG is 1 or ZOOM_FLAG is 1. That is, in step S110, it is determined whether a panning or tilting operation has been performed on the imaging apparatus 100 or a zoom operation has been performed. When either one is 1, it progresses to step S111. In step S111, the magnification K1 of the multiplication unit 242 is set to 0, and the magnification K2 of the multiplication unit 243 is set to 1. That is, it means that the sensor translational image blur data itself is used for calculating the correction amount without using the vector translational image blur data. If it is determined in step S110 that neither PAN_FLAG nor ZOOM_FLAG is 1, the process proceeds to step S112. In step S112, the magnifications of the multiplication units 242 and 243 are set according to the reliability of the motion vector calculated by the motion vector processing unit 201, and the processing in FIG. Note that the processing executed in step S112 is the same as that in the first embodiment, and a description thereof will be omitted.

  As described above, in the third embodiment of the present invention, in addition to the reliability of the motion vector, the vector translation image shake in the translation synthesis unit 203 is determined according to the output of the angular velocity sensor and the output of the zoom encoder. The ratio of adding the sensor translational image blur is changed. As a result, when panning, tilting operation, or zooming operation is performed, by performing image blur correction using translational image blur data obtained from the angular velocity sensor, in a shooting situation where motion vectors are not detected correctly However, it is possible to minimize degradation of image quality.

  Although the present invention has been described in detail based on preferred embodiments thereof, the present invention is not limited to these specific embodiments, and various forms within the scope of the present invention are also included in the present invention. included. A part of the above-described embodiments may be appropriately combined.

(Other embodiments)
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, etc.) of the system or apparatus reads the program. It is a process to be executed.

100: imaging device, 101: μCOM, 102: angular velocity sensor, 103: A / D converter, 120: imaging optical system, 123: imaging device, 126: motion vector detection unit

Claims (9)

Motion vector detection means for detecting a motion vector between two images of the image captured by an imaging means,
During one of the previous SL two images, the image of the previous frame from the imaging until capturing an image of the next frame, based on the output of the angular velocity detecting means for detecting a shake, to photograph the two images a shake calculation means for calculating a shake amount of an imaging device in,
Image blur conversion means for converting the output of the shake calculation means into an image shake amount of an image;
Correction amount calculation for calculating a correction amount for electronically correcting the image shake amount of the image based on a result obtained by adding the output of the motion vector detection means and the output of the image shake conversion means at a predetermined ratio. Means,
On the basis of the output of the correction amount calculating means, and control means for controlling the correction means for electronically correcting image blur of the image,
At the time of moving image shooting, the correction amount calculation means, when it is determined that the imaging device is in a panning or tilting state, rather than when it is determined that the imaging device is not in a panning or tilting state, An image blur correction apparatus characterized in that a ratio of adding outputs of motion vector detection means is reduced . The correction amount calculation unit adds the output of the motion vector detection unit and the output of the image blur conversion unit at a predetermined ratio, and then blocks a low frequency component included in the added image blur data to generate a high frequency. The image blur correction apparatus according to claim 1, further comprising a high-pass filter that allows the component to pass therethrough. The correction amount calculation means calculates the correction amount using the output of the image blur conversion means without using the output of the motion vector detection means when the reliability of the output of the motion vector detection means is a predetermined threshold value or less. And when the reliability of the output of the motion vector detection means is higher than the predetermined threshold, the correction amount is calculated using the output of the motion vector detection means and the output of the image blur conversion means. The image blur correction apparatus according to claim 1. When the reliability of the output of the motion vector detection unit is higher than the predetermined threshold, the correction amount calculation unit increases the reliability of the output of the motion vector detection unit. The image blur correction apparatus according to claim 3, wherein a ratio of adding outputs is increased. The shake calculation unit has an integration unit that integrates the output of the angular velocity detection unit with a cycle shorter than the cycle sequentially imaged by the imaging unit,
Of the two images, the difference between the output of the integration unit when the image of the previous frame is captured and the output of the integration unit when the image of the next frame is captured is calculated. it allows image blur correction apparatus according to any one of claims 1 to 4, characterized in that to calculate the shake amount of the imaging device between the two images. The shake calculation unit has an integration unit that integrates the output of the angular velocity detection unit with a cycle shorter than the cycle sequentially imaged by the imaging unit,
The amount of shake of the imaging device between the two images is calculated by acquiring the output of the integration unit when an image constituting the moving image is captured and resetting the integration value of the integration unit an image blur correction device according to any one of claims 1 to 4, characterized in that. A motion vector detection step of detecting a motion vector between two images of the image captured by an imaging means,
During one of the previous SL two images, the image of the previous frame from the imaging until capturing an image of the next frame, based on the output of the angular velocity detecting means for detecting a shake, to photograph the two images a shake calculation step calculates the shake amount of the imaging device in,
An image blur conversion step of converting the output of the blur calculation step into an image blur amount of an image;
Correction amount calculation for calculating a correction amount for electronically correcting the image shake amount of the image based on a result obtained by adding the output of the motion vector detection step and the output of the image shake conversion step at a predetermined ratio. Process,
Based on the output of the correction amount calculating step, have a, a correction step of electronically correcting image blur of the image,
At the time of moving image shooting, in the correction amount calculating step, when it is determined that the imaging device is in a panning or tilting state, the imaging device is more than in the case where it is determined that the imaging device is not in a panning or tilting state. A control method for an image blur correction apparatus , characterized in that a ratio of adding outputs of motion vector detection means is reduced . A program for causing a computer to execute the control method according to claim 7 . A computer-readable storage medium storing a program for causing a computer to execute the control method according to claim 7 .

Images