JP5029573B2 - Imaging apparatus and imaging method - Google Patents

Description

The present invention relates to an imaging apparatus and an imaging method .

  In recent years, video cameras are equipped with a camera shake correction function for reducing camera shake during shooting. The normal camera shake correction function has a problem in that when the subject in the captured image has a motion, the motion of the subject is erroneously determined as a camera shake, and an unnatural camera shake correction is performed.

Japanese Patent Application Laid-Open No. 2004-151561 discloses a camera shake correction technique that solves the problem. Patent Document 1 divides a captured image into a plurality of regions, determines whether or not a person's face as a subject is included in each region, and determines the region determined to include a person's face. A method of detecting the amount of camera shake by excluding from the region for detecting camera shake is described.
JP 2008-124787 A

  However, with the above-described camera shake correction method, if the subject moves even in a region that does not include a human face, an appropriate amount of camera shake cannot be detected, and an unnatural image is displayed. was there.

The present invention has been made in view of such problems, and an imaging apparatus and an imaging method capable of detecting or detecting an appropriate amount of camera shake by determining or predicting the movement of a subject even in an area that does not include a human face. The purpose is to provide.

In order to solve the above-described problems of the conventional technology, the present invention provides a calculation unit (23) that calculates a motion vector of each area obtained by dividing an image acquired by the imaging unit (11), and a moving object from the image. Moving object region detection for detecting whether or not there is a moving object that can be detected, and detecting the moving object region as a first moving object region from among the regions obtained by dividing the image Unit (14), an extraction unit (15) for extracting the motion vector calculated by the calculation unit (23) as a reference motion vector for the first moving object region detected by the moving object region detection unit (14), It is determined whether or not the motion vector calculated by the calculation unit (23) for the region excluding the first moving body region from the image is similar to the reference motion vector, and the motion vector determined to be similar The specifying unit (15) that specifies the region for which the value is calculated as the second moving object region, and the calculating unit (23) for the region excluding the first moving object region and the second moving object region from the image An image pickup apparatus (1) is provided that includes a camera shake amount calculation unit (15) that calculates a camera shake amount of the entire image based on the motion vector .
Here, the imaging apparatus (1) further includes a predicted motion vector calculation unit (62) that calculates a predicted motion vector for predicting the motion of the moving target based on the reference motion vector and the amount of camera shake of the entire image. The specifying unit (15) further determines, based on the predicted motion vector, an area predicted to include the moving object as a third moving body area for an image temporally later than the image. In particular, the camera shake amount calculation unit (15) further calculates the camera shake amount of the entire subsequent image based on a motion vector of an area excluding at least the third moving body area from the subsequent image. May be.
Also, the moving object region detection unit (14) discriminates the existence of the face in each area, an area is determined that a face is present may be a first moving object area where the moving object exists.

In order to solve the above-described problems of the conventional technology, the present invention calculates a motion vector of each area obtained by dividing an image acquired from the imaging unit (11), and a motion that can be a moving subject from the image. A moving object region detecting step for determining whether or not a target exists, and detecting a moving object region determined to be present as a first moving object region from each of the divided regions, and the moving object region An extraction step for extracting the motion vector calculated in the calculation step as a reference motion vector for the first moving body region detected in the detection step, and the calculation step for a region obtained by removing the first moving body region from the image It is determined whether or not the motion vector calculated in step 3 is similar to the reference motion vector. A specifying step of specifying the calculated region as a second moving body region, and a region obtained by removing the first moving body region and the second moving body region from the image based on the motion vector calculated in the calculating step , to provide an imaging method comprising the shake amount calculation step of calculating the shake amount of the whole image.
Here, the imaging method further includes a predicted motion vector calculating step of calculating a predicted motion vector for predicting the motion of the moving target based on the reference motion vector and the amount of camera shake of the entire image, and the specifying step further includes Based on the predicted motion vector, a region that is predicted to include the moving object is specified as a third moving body region for an image temporally later than the image, and the amount of camera shake calculation step includes: Furthermore, the amount of camera shake of the entire subsequent image may be calculated based on a motion vector of an area obtained by removing at least the third moving body area from the subsequent image.
In the moving object region detecting step, the presence or absence of a face may be determined for each region , and the region determined to have a face may be set as the first moving object region in which the moving object exists .

According to the present invention, it is possible to detect the appropriate blur amount according to the situation of the subject of the shadow image shooting.

<First Embodiment>
Hereinafter, a camera shake correction apparatus and a camera shake correction method according to a first embodiment will be described with reference to the accompanying drawings. FIG. 1 is a block diagram showing an example of a camera shake correction apparatus according to the present invention.

  The camera shake correction apparatus 1 includes an imaging unit 11, a signal processing unit 12, a motion vector detection unit 13, a moving object region detection unit 14, a camera shake amount detection unit 15, an address control unit 16, and a field memory 17.

  The imaging unit 11 has an optical system (not shown), an aperture, an imaging element such as a CCD (Charge Coupled Devices) and a CMOS (Complementary Metal Oxide Semiconductor), and a driver for controlling the optical system and the aperture. The imaging unit 11 receives an optical image from a subject by an imaging device via an optical system and a diaphragm, and obtains an optical signal. Then, the imaging unit 21 photoelectrically converts an optical signal obtained by receiving light into an analog image signal and supplies the analog image signal to the signal processing unit 12.

 The signal processing unit 12 converts the analog image signal supplied from the imaging unit 11 into a digital image signal, and separates the digital image signal into color signals to generate RGB signals. Further, the signal processing unit 12 generates a luminance signal Y and color difference signals Cb and Cr from the obtained RGB signals. The luminance signal Y is supplied to the representative point storage unit 21, the correlation value data calculation unit 22, and the moving object region detection unit 14 in the motion vector detection unit 13. Further, the luminance signal Y and the color difference signals Cb and Cr are supplied to the field memory 17. The RGB signal, the luminance signal Y, and the color difference signals Cb and Cr are subjected to gamma correction processing as necessary.

  The motion vector detection unit 13 detects a motion vector for each region obtained by dividing a captured image for one field into a vertical number and a horizontal direction by using, for example, a representative point matching method. FIG. 2 shows an example in which a captured image is divided into a plurality of regions. The representative point storage unit 21 divides the captured image into regions P1 to P16. Each region is composed of 32 pixels in the horizontal direction and 20 pixels in the vertical direction. The entire region including the regions P1 to P16 in the first embodiment is a region used to calculate a camera shake amount by cutting out a part of the captured image.

  The representative point storage unit 21 determines a designated pixel as a representative point in each region. FIG. 3 is a diagram for explaining representative points of each region, and pixels indicated by black circles are representative points. The representative point storage unit 21 uses 4 × 8 = 32 pixels in total as a representative point for every four lines in the horizontal direction and every five lines in the vertical direction. Then, the representative point storage unit 21 stores the luminance signal Y corresponding to each representative point determined in all regions for one field period, and supplies the luminance signal Y data stored for one field period to the correlation value data calculation unit 22. To do.

  For each representative point, the correlation value data calculation unit 22 supplies the luminance value (luminance signal value) of the representative point one field before supplied from the representative point storage unit 21 and all the values included in the predetermined area of the current field. The difference in luminance value from the pixel is calculated. The predetermined area of the current field is, for example, an area of ± 16 pixels in the vertical and horizontal directions from the same coordinate position as each representative point in the previous field. Since the entire region including the regions P1 to P16 in the first embodiment cuts out a part of the captured image, it is possible to secure ± 16 pixels at each representative point.

  Further, the correlation value data calculation unit 22 calculates the minimum value of the luminance difference and the pixel position of the current field that minimizes the luminance difference from the calculation result of the luminance value difference (luminance difference) for each representative point. Ask. Then, for each representative point, the correlation value data calculation unit 22 obtains a correlation value indicating a relationship between the minimum value of the luminance difference and the pixel position that minimizes the luminance difference and the representative point. Pixel position data that minimizes the luminance difference between the representative points is supplied to the motion vector calculation unit 23, and correlation value data of each representative point is supplied to the correlation value data analysis unit 24.

  The motion vector calculation unit 23 obtains a motion vector between each representative point and a pixel position corresponding to the representative point, and averages the motion vectors of all the representative points included in one region, thereby obtaining one region. On the other hand, one motion vector is calculated. Then, the motion vector calculation unit 23 obtains motion vectors for all the regions by the above-described method. The motion vector data obtained for each region is supplied to the camera shake amount detection unit 15.

  The correlation value data analysis unit 24 determines the validity of each region based on the correlation value data supplied from the correlation value data calculation unit 22. The validity of the area indicates that the variation in the magnitude of the motion vector of each representative point in the area is equal to or greater than a predetermined value, and the direction of the motion vector is equal to or greater than a predetermined ratio. By determining the effectiveness of each region, a region with low reliability such as a large variation in the size of the motion vector of each representative point in the region or a variation in the direction of the motion vector will be described later. It can be excluded from the area for calculating the amount of camera shake on the entire screen.

  For example, the correlation value data analysis unit 24 obtains the effectiveness of each region by the following method. First, the correlation values of representative points included in the region are summed, whether the sum of correlation values is larger than a predetermined threshold, and a value obtained by dividing the average of each correlation value by the minimum value is smaller than the predetermined threshold. Determine whether it is larger. When the sum of correlation values is larger than a predetermined threshold value and the value obtained by dividing the average of correlation values by the minimum value is also larger than the predetermined threshold value, the area is determined to be valid. Otherwise, it is determined that the area is invalid. Data indicating the effectiveness of each region is supplied to the camera shake amount detection unit 15.

  The moving body region detection unit 14 determines whether or not there is a target to be a moving subject (hereinafter referred to as a moving body) in the captured image by detecting a face included in the subject, for example. Whether or not there is a moving object (hereinafter referred to as a moving object) is determined by, for example, a known technique for extracting a skin color region in a captured image based on the luminance signal Y supplied from the signal processing unit 12, It is determined whether or not a face is included in each region by a known technique such as matching with reference data stored in a memory (not shown). Note that the moving object region detection unit 14 may include a memory that stores one or a plurality of fields, and may detect whether or not there is an object to be a moving object based on the one or more fields.

  Then, the moving body region detection unit 14 uses a region including the face, that is, a region determined to have a moving target as the first moving body region, and determines whether or not each region is the first moving body region. 15 is supplied. However, the method for detecting the moving object is not limited to the method for detecting the face, and the method for detecting the face is not limited to the method for extracting the skin color region.

  The camera shake amount detection unit 15 extracts the motion vector of the region that is the first moving object region from the moving object region detection unit 14 as the reference motion vector V1. Then, the camera shake amount detection unit 15 includes the reference motion vector V1 and the motion vector of a region determined to be effective by the correlation value data analysis unit 24 except for the first moving body region (hereinafter referred to as an effective region). Are sequentially compared to determine whether or not the motion vectors are similar to each other. For example, the similar criterion may be that the directions of the motion vectors are within a predetermined range, and the difference in magnitude between the motion vectors is equal to or less than a predetermined threshold.

  If the camera shake amount detection unit 15 determines that the two motion vectors are similar to each other, the region including the similar motion vectors is set as the second moving object region. Furthermore, the camera shake amount detection unit 15 averages the motion vectors of the effective area obtained by excluding the first moving body area and the second moving body area from the effective area, and calculates the average as the camera shake amount V2 of the entire screen. Then, the data of the camera shake amount V2 is supplied to the address control unit 16.

  If there are a plurality of first moving body regions, a plurality of reference motion vectors V1 are set, and a region having a motion vector similar to any of the plurality of reference motion vectors V1 is set as the second moving body region. Alternatively, a plurality of reference motion vectors may be averaged to form one reference motion vector V1. When the first moving object region is not detected, the second moving object region is not detected, and the camera shake amount V2 is calculated in all effective regions.

  FIG. 4 is a diagram for explaining the calculation of the amount of camera shake of the entire screen. In FIG. 4, the arrows shown in the divided areas P1 to P16 indicate motion vectors. Areas P1 to P4, P7, and P8 indicated by hatching in FIG. 4 are areas that are determined as invalid areas by the correlation value data analysis unit 24, and are excluded from areas when calculating the amount of camera shake on the entire screen. Is done.

  Regions P5, P6, and P9 to P16 are regions determined to be effective regions by the correlation value data analysis unit 24. Furthermore, the region P11 is a first moving body region detected by the moving body region detection unit 14 as a region including a human face. The motion vector in the region P11 is the reference motion vector V1. The regions P12, P15, and P16 are detected by the camera shake amount detection unit 15 as the second moving body region having a motion vector similar to the reference motion vector V1. Therefore, the camera shake amount detection unit 15 averages the motion vectors of the regions P5, P6, P9, P10, P13, and P14 in which the first moving body region and the second moving body region are excluded from the effective regions, and the camera shake of the entire screen. The amount V2 is calculated.

  The address control unit 16 calculates the address reading position of the digital image signal corrected by the vector amount corresponding to the direction opposite to the camera shake amount V2−V2 with respect to the address start position of the entire region used when the camera shake amount V2 is calculated. To do. The read address position data is supplied to the field memory 17. The digital image signal is a signal based on the luminance signal Y and the color difference signals Cb and Cr supplied from the signal processing unit 12 to the field memory 17 described later.

  The field memory 17 controls the position where the digital image signal is read from the read address position supplied from the address control unit 16 and outputs the digital image signal to a display unit (not shown). FIG. 5 is an explanatory diagram for displaying an image by controlling the read address position. In FIG. 5, a region indicated by a solid line is a region Im1 indicating the entire captured image. A region indicated by a broken line is a region Im2 cut out as regions P1 to P16 in order to calculate the camera shake amount V2. A region indicated by a one-dot chain line is a region Im3 cut out by using the address control unit 16 as a read start position R, which is a position moved by a vector corresponding to the direction −V2 opposite to the camera shake amount V2.

  The data indicating the validity of the region obtained by the correlation value data analysis unit 24 has been described as being supplied to the camera shake amount detection unit 15, but is supplied to the motion vector calculation unit 23 and invalidated. A configuration may be adopted in which motion vectors are not calculated in the divided areas. In addition, the method for determining the representative points, the number of area divisions, the method for determining the validity of the area, and the like in the first embodiment can be variously changed without departing from the spirit.

  The operation of the camera shake correction apparatus shown in FIG. 1 described above and the procedure of the camera shake correction method performed by the camera shake correction apparatus shown in FIG. 1 will be described again with reference to FIG. In step S01, the representative point storage unit 21 divides the captured image into a plurality of regions, and defines a plurality of representative points in each region. In step S02, the correlation value data calculation unit 22 calculates a correlation value from the relationship between the representative point and the pixel that minimizes the luminance difference for each representative point in each region.

  The motion vector calculation unit 23 obtains a motion vector for each representative point in step S03, averages all the motion vectors included in one region, calculates one motion vector for one region, Motion vectors are calculated for all regions. In step S04, the correlation value data analysis unit 24 determines the validity of each region from the correlation value of each representative point included in the region to obtain an effective region. In step S05, the moving object region detection unit 14 determines whether or not a moving object exists, and sets the region where the moving object exists as a first moving object region. In step S06, the camera shake amount detection unit 15 sets the motion vector of the region where the moving target exists as the reference motion vector.

  In step S07, the camera shake amount detection unit 15 determines that the region having a motion vector similar to the reference motion vector is the second moving object region. In step S08, the camera shake amount detection unit 15 calculates the camera shake amount by averaging the motion vectors of the effective regions excluding the first moving body region and the second moving body region. The address controller 16 corrects the amount of camera shake in S09 to calculate the address reading position of the digital image signal, and the field memory 17 controls the position where the digital image signal is read based on the reading address position. Is output to a display unit (not shown).

  As described above, in the first embodiment, an effective area is obtained from the captured image, and in addition to the first moving body area detected as the area where the moving object exists, the motion vector of the first moving body area is similar. Since the camera shake amount is calculated by excluding the second moving body region having the motion vector to be corrected, an accurate camera shake amount can be obtained.

<Second Embodiment>
Hereinafter, a camera shake correction apparatus and a camera shake correction method according to a second embodiment will be described with reference to the accompanying drawings. FIG. 7 is a block diagram illustrating an example of a camera shake correction apparatus according to the present invention. In FIG. 7, the same parts as those in FIG. The camera shake correction apparatus 51 employs a camera shake amount detection unit 61 and a predicted motion vector calculation unit 62 in place of the camera shake amount detection unit 15 of the camera shake correction apparatus 1 shown in FIG.

  Similar to the camera shake amount detection unit 15, the camera shake amount detection unit 61 detects a second moving body region having a motion vector similar to the reference motion vector V1 of the first moving body region detected by the moving body region detection unit 14. . Then, motion vectors are averaged in an effective area excluding the first moving body area and the second moving body area, and the amount of camera shake V2 of the entire screen is calculated. The effective area is obtained by the same method as in the first embodiment. The data of the reference motion vector V1 and the data of the camera shake amount V2 are supplied to the predicted motion vector calculation unit 62.

  The predicted motion vector calculation unit 62 calculates the difference between the reference motion vector V1 and the amount of camera shake V2, and obtains a motion vector (motion target motion vector) V3 indicating the amount of movement of the motion target itself. Also, the predicted motion vector calculation unit 62 stores a motion vector V3 for one field or a plurality of fields. Then, the predicted motion vector calculation unit 62 calculates a predicted motion vector V4 for obtaining a position where the moving target is predicted to move in the next field from the moving target motion vector V3 for one field or a plurality of fields. The predicted motion vector V4 is supplied to the camera shake amount detection unit 61.

  The camera shake amount detection unit 61 predicts the position of the moving object in the next field based on the predicted motion vector V4, and sets the area predicted to include the moving object as the third moving body area. Then, when the next field is supplied, the camera shake amount detection unit 61 calculates the camera shake amount V2 from an effective region excluding the third moving body region. By calculating the camera shake amount V2 excluding the third moving object region predicted in advance, the camera shake amount can be determined more accurately. Moreover, it is good also as a structure which excludes a 1st moving body area | region and a 2nd moving body area | region in addition to a 3rd moving body area | region. The third moving object region is a predicted region to the last, and it is possible that the moving object remains in the first moving object region or the second moving object region, and by excluding all moving object regions, This is because an accurate amount of camera shake can be obtained.

  Further, the camera shake amount detection unit 61 calculates the camera shake amount by excluding the first and second moving object regions in the odd-numbered field, and the camera shake amount by excluding the third region in the even-numbered field. Various modifications such as calculation may be used. Further, the camera shake amount detection unit 61 displays the moving target at a predicted position on a display unit (not shown) via the address control unit 16 and the field memory 17 using the predicted motion vector V4 of the moving target. Etc. are also possible.

  FIG. 8 is a diagram for explaining calculation of a motion vector of a moving target. 8A and 8C, the arrows shown in the areas P1 to P16 indicate motion vectors. In FIG. 8, it is assumed that all the regions P1 to P16 are determined as valid regions by the correlation value analysis data unit 24. The region P11 is a first moving body region detected by the moving body region detection unit 14 as a region including a human face. The motion vector in the region P11 is the reference motion vector V1.

  Then, the divided areas P12, P15, and P16 are detected by the camera shake amount detection unit 15 as the second moving body area having a motion vector similar to the reference motion vector V1. Therefore, the camera shake amount detection unit 15 calculates the motion vectors of the areas P5, P6, P9, P10, P13, and P14 obtained by excluding the first moving body area and the second moving body area from the effective area for the current field. The camera shake amount V2 of the entire screen is calculated on average. The data of the reference motion vector V1 and the data of the camera shake amount V2 are supplied to the predicted motion vector calculation unit 62.

  FIG. 8B shows the enlarged reference motion vector V1 and camera shake amount V2. As shown in FIG. 8B, the predicted motion vector calculation unit 62 obtains a motion vector V3 from the difference (V1−V2) between the two.

  FIG. 8C shows the predicted position of the moving target in the next field of the captured image of FIG. FIG. 8C shows a case where the predicted motion vector calculation unit 62 calculates the predicted motion vector V4 from the motion vector V3 to be moved for one field. At this time, the predicted motion vector V4 matches the motion vector V3. As shown in FIG. 8C, in the next field, the moving object is predicted to be located over the areas P10, P11, P14, and P15, and this predicted area is set as the third moving body area.

  The operation of the camera shake correction apparatus shown in FIG. 7 described above and the procedure of the camera shake correction method performed by the camera shake correction apparatus shown in FIG. 7 will be described again with reference to FIG.

  In step S11, the predicted motion vector calculation unit 62 calculates a motion vector V3 from the difference between the reference motion vector V1 calculated in step S05 and the camera shake amount V2 calculated in step S08. The predicted motion vector calculation unit 62 stores the motion vector V3 of the moving object for one or more fields, and in step S12, the predicted motion vector V4 of the moving object in the next field from the motion vector V3 of the moving object. Is calculated.

  In step S13, the camera shake amount detection unit 61 predicts the moving position of the moving object in the next field based on the predicted motion vector V4, and sets the area including the predicted position of the moving object as the third moving body area. . Further, the camera shake amount detection unit 61 calculates the camera shake amount V2 excluding the third moving body region when the next field is supplied in step S14. Note that the process in step S14 may be variously modified methods such as calculating the camera shake amount V2 excluding the first moving body region and the second moving body region including the third moving body region as described above.

  As described above, in the second embodiment, the amount of camera shake is calculated by excluding the third moving body region detected as the region predicted as the position of the moving target in the next field. The amount of camera shake can be obtained.

1 is a block diagram illustrating an overall configuration of a camera shake correction apparatus according to a first embodiment of the present invention. It is a figure which shows an example of the area | region division of the picked-up image which concerns on the 1st, 2nd embodiment of this invention. It is a figure which shows an example of the representative point in each area | region of the picked-up image which concerns on the 1st, 2nd embodiment of this invention. It is a figure for demonstrating calculation of the camera-shake amount of the whole screen which concerns on the 1st Embodiment of this invention. It is a figure which shows the relationship between the whole area | region of the picked-up image based on 1st, 2nd embodiment of this invention, and a cutting frame. It is a flowchart which shows the whole flow of the camera-shake correction method which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the whole structure of the camera-shake correction apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows the example of the motion vector of the moving object which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows the whole flow of the camera-shake correction method which concerns on the 2nd Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 51 Camera shake correction apparatus 11 Imaging part 12 Signal processing part 13 Motion vector detection part 14 Moving body area | region detection part 15, 61 Camera shake amount detection part 16 Address control part 17 Field memory 21 Representative point memory | storage part 22 Correlation value data calculation part 23 Motion Vector calculation unit 24 Correlation value data analysis unit 62 Predicted motion vector calculation unit

Claims (6)

A calculation unit that calculates a motion vector of each region obtained by dividing the image acquired by the imaging unit;
It is determined whether or not there is a moving target that can be a moving subject in the image. If it is determined that there is a moving target, the region where the moving target exists is set as a first moving body region, and each region obtained by dividing the image is divided. A moving object region detection unit for detecting from among,
An extraction unit that extracts, as a reference motion vector, a motion vector calculated by the calculation unit for the first moving object region detected by the moving object region detection unit;
A region in which the motion vector calculated by the calculation unit for the region excluding the first moving body region from the image is similar to the reference motion vector, and the motion vector determined to be similar is calculated A specifying unit for specifying as a second moving body region;
An image pickup comprising: a shake amount calculation unit that calculates a shake amount of the entire image based on a motion vector calculated by the calculation unit for a region obtained by removing the first moving body region and the second moving body region from the image. apparatus. A prediction motion vector calculation unit that calculates a prediction motion vector for predicting the motion of the moving object based on the reference motion vector and the amount of camera shake of the entire image;
The specifying unit further specifies a region predicted to include the moving object as a third moving object region for an image temporally later than the image based on the predicted motion vector,
The image pickup apparatus further includes a camera shake amount calculation unit that calculates a camera shake amount of the entire subsequent image based on a motion vector of an area obtained by removing at least the third moving body area from the subsequent image . The said moving body area | region detection part discriminate | determines the presence or absence of a face for every said area | region, The area | region discriminate | determined that the face exists is made into the 1st moving body area | region where the said moving object exists . Imaging device. A calculation step of calculating a motion vector of each region obtained by dividing the image acquired from the imaging unit;
It is determined whether or not there is a moving target that can be a moving subject in the image. If it is determined that there is a moving target, the region where the moving target exists is set as a first moving body region, and each region obtained by dividing the image is divided. A moving object region detecting step for detecting from among them,
An extraction step for extracting, as a reference motion vector, the motion vector calculated in the calculation step for the first moving body region detected in the moving body region detection step;
It is determined whether or not the motion vector calculated in the calculation step for the region excluding the first moving body region from the image is similar to the reference motion vector, and the motion vector determined to be similar is calculated A specific step of identifying the region as the second moving body region;
A camera shake amount calculating step of calculating a camera shake amount of the entire image based on the motion vector calculated in the calculating step for the region excluding the first moving object region and the second moving object region from the image. Imaging method. A predicted motion vector calculating step of calculating a predicted motion vector for predicting the motion of the moving object based on the reference motion vector and the amount of camera shake of the entire image;
The specifying step further specifies a region predicted to include the moving object as a third moving object region for an image temporally later than the image based on the predicted motion vector,
5. The camera shake amount calculation step further calculates a camera shake amount of the entire subsequent image based on a motion vector of an area obtained by removing at least the third moving body area from the subsequent image. The imaging method described in 1 . 6. The moving object region detecting step determines the presence or absence of a face for each of the regions, and sets the region determined to have a face as a first moving object region in which a moving object exists . Imaging method.

Images