JP2007043228A - Electronic blurring correction apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an electronic blurring correction apparatus capable of carrying out blurring correction with high accuracy corresponding to a wide shutter speed while eliminating the need for a complicated mechanical mechanism for blurring correction. <P>SOLUTION: The electronic blurring correction apparatus includes: a solid-state imaging element 1 including an imaging section wherein a plurality of pixels are arranged in a matrix form; a register section including a first transfer register that latches a first image imaged by the imaging section and transfers the first image in a first direction and a second transfer register that latches a second image photographed by the imaging section at a time different from that photographed when the first image is photographed and transfers the second image in a second direction; angular velocity sensors 19, 20 for detecting a relative blurring between the first and second images; and a CPU 7 controls the first and second transfer registers to relatively transfer the first and second images to each other for correcting the blurring and thereafter controls the first and second transfer registers to analogically summate the first and second images. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an electronic shake correction apparatus that generates a corrected image by correcting a shake of a plurality of images taken continuously in time.

  Many imaging devices that capture a still image or a moving image using a solid-state imaging device are configured to be captured by hand. In such an imaging apparatus, for example, when the luminance of a subject is low, it is known that camera shake is likely to occur because the shutter speed becomes slow. In addition, it is known that in-vehicle cameras and the like may be shaken similarly due to vibration during traveling.

  Various techniques for correcting such blur have been proposed, and some examples of typical blur correction mechanisms will be described below.

(1) Electronic blur correction (Japanese Patent Laid-Open No. 6-46316, etc.)
In the electronic blur correction, the effective pixel area is set to an area considerably smaller than the maximum pixel area that can be imaged using the image sensor. Then, when shooting is performed with a digital camera equipped with an electronic blur correction function, the shot image is not temporarily recorded on the memory card, but temporarily temporarily stored in the buffer memory. The digital camera captures the next image immediately without leaving a time interval. Similarly to the first photographed image, the second photographed image is not immediately recorded on the memory card but is first stored in the buffer memory. At this time, if the position of the subject in the second photographed image is shifted (blurred) from the position of the subject in the first photographed image, the digital camera detects that the first image and 2 Compared with the first image, the data area to be used is shifted so that the subject fits in the screen with the same composition, and this is used as the effective pixel area in the second image. By using such a technique, it is possible to record image data in which the position of the subject does not shift between the images (that is, no blur occurs between the images).

(2) Optical blur correction (JP-A-10-336510, etc.)
In the optical shake correction, when the vibration gyroscope senses the movement of the camera, the mainstream method is to move a part of the lens in a direction to cancel the shake at the position where the light reaches. By using such technology, even if the camera is likely to move and shake, the position of the light that first reaches the image sensor after the shutter opens and the light that reaches the image sensor just before the shutter closes. It is possible to prevent the camera shake by setting the same position to the same position.

(3) Sensor shift type blur correction (Japanese Patent Laid-Open No. 2004-56581)
The sensor shift type shake correction is a technique for correcting the shake by moving the image sensor by the shake amount of the light arrival position on the imaging surface when the vibration gyroscope senses the movement of the camera. This technique has an advantage that blur correction can be performed without basically deteriorating image quality because a part of the lens does not move. Since this technology can be applied to the camera body side of an interchangeable lens camera, it is possible to perform blur correction even using a conventional interchangeable lens that does not consider the blur correction technology. There are advantages you can do.

(4) Japanese Patent Laid-Open No. 2001-86398 The image pickup apparatus described in this publication reads a plurality of images taken at exposure time intervals at which no blurring occurs from the image pickup device and stores them in a buffer memory. After correcting the image blur (mutual image position shift) of the plurality of images stored in the memory, the corrected images are added to generate an image without blur. This technique has an advantage that a complicated mechanical mechanism is not required because the electronic shake correction is performed electronically.

By the way, Japanese Patent Application Laid-Open No. 7-226696 describes a technique related to a solid-state imaging device and a manufacturing method thereof, and shows an example of wiring between electrodes.
JP-A-6-46316 Japanese Patent Laid-Open No. 10-336510 JP 2004-56581 A JP 2001-86398 A Japanese Patent Laid-Open No. 7-226696

  However, the electronic blur correction technique as described above is a technique for preventing the composition and the subject position from blurring when a plurality of image data are continuously viewed, and occurs only in one image. It is not a technology that corrects shaking. That is, even if each image is shaken due to, for example, a long exposure time, it is not a technique for suppressing the shake, and cannot be said to be a technique that can be effectively applied as it is to shake correction in a still image. .

  On the other hand, the optical blur correction as described above can be applied to the blur correction in a still image, but the design is very difficult and the cost is high. In addition, although the blur is corrected, for example, chromatic aberration is somewhat increased, and strictly speaking, optical performance is deteriorated. Further, unlike the electronic blur correction technique, it is necessary to incorporate a mechanism for moving the lens, so that it is difficult to reduce the size of a digital camera or the like incorporating the mechanism. In addition, when this technology is applied to, for example, an interchangeable lens camera, it is necessary to incorporate at least a part of the blur correction mechanism on the lens side. Therefore, in an interchangeable lens that does not support the conventional blur correction function, Camera shake cannot be corrected.

  In addition, since the sensor shift type blur correction as described above mechanically drives the image sensor, it is difficult to reduce the size of a digital camera or the like incorporating the mechanism. This is the same as in the case of formula blur correction. In addition, when the image sensor is large, the amount of movement is large, and thus the mechanism is inevitably increased in size. It is important for the image sensor to move only on the imaging plane perpendicular to the optical axis of the photographic lens. However, the occurrence of backlash, the tilt of the image sensor with respect to the optical axis, the image sensor around the optical axis, It is difficult to prevent rotation and the like.

Furthermore, the technique described in the above-mentioned Japanese Patent Application Laid-Open No. 2001-86398 needs to read a plurality of times from the image sensor in order to obtain one piece of image data. The readout from the image sensor requires a longer time as the number of pixels constituting the image sensor increases. As a specific example, in the case where pixel signals are read out from a 6-million-pixel imaging device at a cycle of 30 [MHz], it takes 200 [ms] to read out signals of all pixels once. Become. Therefore, when reading is performed a plurality of times, an integral multiple of 200 [ms] (or more) is required. On the other hand, in a 35 mm film camera, for example, when the focal length of the photographic lens is f [mm], the longest exposure time that can substantially suppress the occurrence of blurring is empirically 1 / f [second]. Are known. For example, when a standard 50 [mm] photographing lens is used, the longest exposure time is 20 [ms]. But,
Even the time 200 [ms] for reading the image signal once from the image sensor becomes extremely longer than the longest exposure time 20 [ms]. Too long. Thus, for example, the technique described in the publication discloses that the number of pixels of the image sensor is small, and the longest exposure time that can substantially suppress the occurrence of blurring is long (for example, the focal length of the photographing lens is short). It is considered practical only under limited conditions. In addition, according to the technique described in Japanese Patent Application Laid-Open No. 2001-86398, each image obtained by divided shooting is shot with an exposure time that does not cause blurring. Is set based on empirical data as described above, and blur correction may not be sufficient depending on individual differences and photographing conditions.

  The present invention has been made in view of the above circumstances, and provides an electronic camera shake correction device that enables highly accurate camera shake correction corresponding to a wide range of shutter speeds while eliminating the need for a complicated mechanical mechanism for camera shake correction. The purpose is that.

  In order to achieve the above object, the electronic shake correction apparatus according to the first aspect of the present invention performs analog addition while sequentially correcting shakes of a plurality of images taken continuously in time inside the solid-state imaging device. An electronic shake correction apparatus that generates an image with a shake corrected by the above-described solid-state imaging device, and has a plurality of pixels arranged in a matrix, and an image signal is output by the plurality of pixels. An imaging unit to be generated and a first image captured by the imaging unit and stored in the solid-state imaging device separately from the imaging unit and storing the first image in the first direction A second image that is captured by the first transfer register and the imaging unit at a time different from the first image is stored, and the second image is transferred in a second direction that intersects the first direction. Transfer register A register unit, a blur detection unit for detecting relative blur between the first image and the second image, and the first transfer register and the second transfer for the first image and the second image, respectively. A control unit that relatively transfers in the register to correct relative blur between the first image and the second image, and then controls the first image and the second image to be analog-added; Is provided.

  The electronic shake correction apparatus according to the second invention is the electronic shake correction apparatus according to the first invention, wherein the control unit analog-adds an image obtained by analog addition of the first image and the second image. 2 is controlled to be stored in the transfer register.

  Furthermore, the electronic shake correction apparatus according to a third aspect of the invention is the electronic shake correction apparatus according to the first aspect of the invention, wherein the shake detection unit is configured to make the first relative to the first image and the second image relative to each other. In the second direction, and the control unit transfers the first image in the first transfer register based on the blur in the first direction. And the second image is controlled to be transferred in the second transfer register based on the blur in the second direction.

  According to a fourth aspect of the present invention, in the electronic shake correction apparatus according to the first aspect of the invention, the solid-state imaging device transfers an image picked up by the image pickup unit to the register unit. .

  According to a fifth aspect of the present invention, there is provided an electronic image blur correction apparatus according to the first invention, wherein the first transfer register and the second transfer register are each one pixel which is a minimum unit constituting an image. Each of the four electrodes is configured to transfer charges by applying different driving pulses to the four electrodes.

  An electronic shake correction apparatus according to a sixth invention is the electronic shake correction apparatus according to the fifth invention, wherein each of the first transfer register and the second transfer register has a relatively large area; Electrodes having a relatively small area are alternately arranged.

  According to the electronic shake correction apparatus of the present invention, it is possible to perform high-precision shake correction corresponding to a wide range of shutter speeds while eliminating the need for a complicated mechanical mechanism for shake correction.

  Before describing the embodiment in detail, first, the principle for correcting blur will be briefly described.

  For example, it is assumed that the proper exposure time (total exposure time) obtained by photometry of the subject is 1/15 seconds. Then, it is assumed that blurring occurs in the exposure time (shutter speed) of 1/15 seconds. On the other hand, when the exposure time (shutter speed) is 1/125 seconds, no blur occurs or the generated blur can be substantially ignored. In such a case, the total exposure time of 1/15 seconds described above is time-divided into an exposure time of 1/125 seconds, and shooting is performed 8 times by the time-division exposure, and this time-division shooting is obtained. By adding the eight images, one image having an appropriate exposure time of 1/15 seconds is obtained. However, since the shake is not corrected by simply adding the time-division exposure images taken at 1/125 seconds, the shakes of the respective time-division exposure images are corrected and then added. At this time, if the time-division exposure interval is long, it is not possible to continuously shoot with a high-speed exposure time without blurring. Therefore, in the embodiment described below, the shake is corrected at high speed inside the solid-state imaging device, and the corrected image is added.

  Specifically, first, an image photographed by time-division exposure is stored in a second transfer register provided in the image sensor, and an image photographed by the next time-division exposure is stored in a first transfer register provided in the image sensor. . Based on the amount of shake detected by the shake detection unit, one of the first transfer register and the second transfer register is shifted in the X direction and the other is shifted in the Y direction. The images are added and the added image is stored in the second transfer register. Subsequently, the image captured by the next time-division exposure is stored again in the first transfer register, and the above-described correction and addition are repeated until the entire exposure amount becomes appropriate.

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

[Embodiment 1]
FIG. 1 to FIG. 19 show Embodiment 1 of the present invention, and FIG. 1 is a block diagram mainly showing an electrical configuration of an electronic camera. In the present embodiment, an electronic shake correction apparatus is applied to an electronic camera.

  This electronic camera includes a solid-state image sensor (hereinafter abbreviated as an image sensor as appropriate) 1, a correlated double sampling circuit (CDS) 2, a gain control amplifier (AMP) 3, and an A / D converter. Unit 4, timing generator (TG) 5, signal generator (SG) 6, CPU 7, information processing unit 8, DRAM 9, compression / decompression unit 10, recording medium 11, liquid crystal display unit 12, and interface Unit 13, lens drive system 14, photographing lens 15, aperture drive system 16, aperture 17, first release switch 18a and second release switch 18b, angular velocity sensor (angular velocity sensor A) 19, and angular velocity sensor ( Angular velocity sensor B) 20, A / D converter 21 and A / D converter 22, distance detection unit 23, and shooting mode setting unit 25. Includes the imaging condition setting unit 26, a.

  The photographing lens 15 is an imaging optical system for forming a subject image on the imaging surface of the imaging device 1.

  The diaphragm 17 is an optical diaphragm for adjusting the amount of light by defining the passage range of the imaging light beam from the photographing lens 15.

  The image sensor 1 photoelectrically converts the subject image formed by the photographing lens 15 via the diaphragm 17 and outputs it as an electrical signal. The imaging device 1 includes an imaging unit 41 (see FIG. 2) that photoelectrically converts an optical image of a subject, and a pixel that can shift and add two consecutively captured images in an arbitrary direction. And a correction / addition unit 40 (see FIG. 2) for transferring the charges in the horizontal direction and the vertical direction. A more detailed configuration and operation of the image sensor 1 will be described later.

  The TG 5 supplies a transfer pulse for driving the image sensor 1.

  The CDS 2 is driven according to the sample hold pulse supplied from the TG 5 and removes reset noise by performing processing such as correlated double sampling on the image signal output from the image sensor 1.

  SG6 generates a synchronization signal based on the control of the CPU 7 and outputs it to the TG5.

  The gain control amplifier (AMP) 3 amplifies an analog signal output from the CDS 2.

  The A / D converter 4 converts an analog signal output from the gain control amplifier (AMP) 3 into a digital signal in accordance with a signal supplied from the TG 5.

  The information processing unit 8 processes the pixel signal output from the A / D converter 4 to generate image data.

  The DRAM 9 temporarily stores the image data output from the information processing unit 8, and the image data obtained by expanding the compressed image data read from the recording medium 11 by the compression / decompression unit 10. It is a temporary memory.

  The compression / decompression unit 10 compresses image data stored in the DRAM 9 and decompresses compressed image data read from the recording medium 11.

  The recording medium 11 records the image data compressed by the compression / decompression unit 10 and is, for example, a nonvolatile recording medium.

  The liquid crystal display unit 12 displays image data output from the information processing unit 8 or expanded image data output from the DRAM 9.

  The interface unit 13 is an interface including terminals for exchanging data with an external device such as a monitor or a personal computer. It is possible to output image data or the like supplied from the information processing unit 8 or the DRAM 9 to an external device via the interface unit 13, or in some cases, image data or the like is taken into the device from the external device. Is possible.

  The lens driving system 14 is configured to drive the photographing lens 15 to the in-focus position by receiving a command from the CPU 7 based on the subject distance detected by the distance detecting unit 23. Such processing is known as so-called autofocus control. Here, although autofocus control is performed based on the output from the distance detection unit 23, the CPU 7 uses a high-pass filter or the like for the luminance component of the image data for one frame (one screen) stored in the DRAM 9. An AF evaluation value corresponding to a contour component on the high frequency region side is calculated by extracting a high frequency component, calculating a cumulative addition value of the extracted high frequency component, etc., and performing focus detection based on the AF evaluation value It doesn't matter.

  The diaphragm drive system 16 is a diaphragm control means for driving the diaphragm 17 to change the aperture diameter by the CPU 7 performing an exposure calculation based on the image data stored in the DRAM 9 and receiving a command based on the result from the CPU 7. It is. Such processing is known as so-called AE (automatic exposure) control.

  The angular velocity sensor 19 is obtained when the electronic camera is rotated about the X axis direction as a rotation center when the right direction in the left and right directions when the electronic camera is viewed from the subject side is the X axis direction (see FIG. 14). This is for detecting the angular velocity and is a shake detection means.

  On the other hand, the angular velocity sensor 20 detects an angular velocity when the electronic camera is rotated with the Y-axis direction as the center of rotation when the upward direction in the vertical direction of the electronic camera is the Y-axis direction (see FIG. 14). And shake detection means.

  The A / D converter 21 converts an analog signal representing the angular velocity detected by the angular velocity sensor 19 into a digital signal at a predetermined time interval (sampling interval).

  Similarly, the A / D converter 22 converts an analog signal representing the angular velocity detected by the angular velocity sensor 20 into a digital signal at a predetermined time interval (sampling interval).

  The CPU 7 performs processing for time integration of the digital signal converted by the A / D converter 21. This time-integrated digital signal corresponds to a rotation amount about the X axis of the camera body as a rotation center. Whether the rotation direction around the X axis is clockwise or counterclockwise is determined depending on whether the analog output signal of the angular velocity sensor 19 is positive or negative.

  Similarly, the CPU 7 performs processing for time integration of the digital signal converted by the A / D converter 22. This time-integrated digital signal corresponds to a rotation amount around the Y axis of the camera body. Whether the rotation direction around the Y axis is clockwise or counterclockwise is determined depending on whether the analog output signal of the angular velocity sensor 20 is positive or negative.

  The first release switch 18a is the first stage of a release switch which is an automatic return type two-stage switch for inputting an imaging operation. When the release switch is pushed in and the first release switch 18a is turned on, a distance measuring operation and a photometric operation are performed.

  The second release switch 18b is the second stage of a release switch which is an automatic return type two-stage switch for inputting an image capturing operation. When the release switch is further pushed in and the second release switch 18b is turned on, an image pickup operation is performed by the image pickup device 1, and the image data is generated as described above, compressed, and recorded on the recording medium 11. It has become so.

  The distance detection unit 23 is for detecting the distance to the subject, and can employ a known configuration as appropriate.

  The shooting mode setting unit 25 is for selecting one of the shutter priority shooting mode, the aperture priority shooting mode, and the program shooting mode.

  The photographing condition setting unit 26 is for setting various conditions necessary for photographing such as a shutter speed (exposure time), an aperture value, and ISO sensitivity.

  The CPU 7 incorporates an EEPROM 24 as a non-volatile memory for storing a program diagram and other information to be described later. The CPU 7 includes a signal from the first release switch 18a, a signal from the second release switch 18b, a signal from the angular velocity sensor 19 via the A / D converter 21, and an angular velocity sensor via the A / D converter 22. 20, a signal from the shooting mode setting unit 25, and a signal from the shooting condition setting unit 26 are input. Then, the CPU 7 outputs commands to TG5 and SG6. Further, the CPU 7 is bi-directionally connected to the information processing unit 8, the DRAM 9, the lens driving system 14, the aperture driving system 16, and the distance detecting unit 23, and serves as a control unit that controls the entire electronic camera including them. ing. Specifically, the CPU 7 performs the autofocus control and the AE control as described above, and based on a signal instructing capturing of a still image from the first release switch 18a and the second release switch 18b, The drive mode is switched. Further, the CPU 7 performs control for changing the aperture of the diaphragm 17 and exposure time control for the image sensor 1. The CPU 7 sets the shooting mode of the electronic camera based on the input from the shooting mode setting unit 25, and sets the shooting condition for the electronic camera based on the input from the shooting condition setting unit 26. In addition, the CPU 7 also calculates a shake amount based on the outputs from the angular velocity sensors 19 and 20, and also serves as a shake detection unit.

  Next, FIG. 2 is a diagram showing an outline of the overall configuration of the solid-state imaging device 1.

  As illustrated in FIG. 2, the imaging element 1 includes an imaging unit 41, a correction / addition unit 40, and a horizontal transfer unit 42.

  The imaging unit 41 converts the optical image formed by the photographing lens 15 into an electric signal. In other words, the imaging unit 41 includes a plurality of (substantially many) photodiodes (PD) 32 (see FIG. 3) arranged in a matrix, and adjacent to each column of the photodiodes 32. A vertical transfer unit 33 composed of vertical transfer CCDs arranged in the direction is provided. The vertical transfer unit 33 is a first transfer unit for transferring charges accumulated by photoelectric conversion by each photodiode 32 to a correction / addition unit 40 (= register unit) described below.

  The correction / addition unit 40 is disposed adjacent to the imaging unit 41, corrects the relative blurring of a plurality of time-division images transferred from the imaging unit 41, and corrects the time-division images to each other. Addition is performed, and the added image is stored. The configuration of the correction / addition unit 40 will be described later in detail with reference to FIG.

  The horizontal transfer unit 42 is a second transfer unit for reading out an image whose shake has been corrected by the correction / addition unit 40 to the outside of the image sensor 1.

  The signal read from the horizontal transfer unit 42 is analog amplified by the amplifier 43.

  As is apparent from the configuration shown in FIG. 2, the image sensor 1 is an application of the frame interline transfer method (FIT method). As shown, it is configured by an interline transfer system.

  Next, FIG. 3 is an enlarged view showing the configuration of the imaging unit 41 and the correction / addition unit 40 of the solid-state imaging device 1.

  Note that IφV0 to IφV3, SφH0 to SφH3, S1φV0 to S1φV3, S2φV0 to S2φV3 indicate drive pulses for charge transfer as described later. For example, the description of “transfer electrode IφV0” is “drive pulse IφV0”. Means an electrode to which is applied.

  The imaging unit 41 includes a plurality of photodiodes (PD) 32 arranged in a matrix, and a vertical transfer unit 33 for transferring charges accumulated by photoelectric conversion by these photodiodes 32 to the correction / addition unit 40. It is equipped with. Here, the vertical transfer unit 33 includes vertical transfer electrodes IφV0 to IφV3 for applying transfer pulses.

  Further, one pixel 31 of the imaging unit 41 includes one photodiode 32 and two vertical transfer electrodes. At this time, there are two types of one pixel 31; the first includes a photodiode 32 and vertical transfer electrodes IφV0 and IφV3, and the second includes a photodiode 32 and vertical transfer electrodes IφV1 and IφV2. This one pixel 31 has a horizontal size (X direction) of Lx and a vertical direction (Y direction) of Ly.

  The correction / addition unit 40 includes a Y direction transfer unit 34, a divided image transfer unit 35, and an X direction transfer unit 36.

  The divided image transfer unit 35 is a transfer unit that transfers the charge accumulated in the photodiode 32 to the correction / addition unit 40, and the vertical transfer arranged in the vertical direction so as to be continuous with the vertical transfer unit 33 of the imaging unit 41. The electrode S1φV0 to S1φV3 are included. These vertical transfer electrodes S1φV0 to S1φV3 are electrodes for applying a pulse for transferring the charge transferred from the imaging unit 41 to the correction / addition unit 40.

  The X-direction transfer unit 36 is a transfer unit (first transfer register) that transfers the charges transferred by the divided image transfer unit 35 in the horizontal direction, and extends in the horizontal direction so as to intersect the divided image transfer unit 35 described above. The horizontal transfer electrodes SφH0 to SφH3 are arranged. These horizontal transfer electrodes S.phi.H0 to S.phi.H3 are arranged in the horizontal direction by the amount of charge transferred below the electrodes S1.phi.V0 and S1.phi.V2 of the vertical transfer electrodes S1.phi. It is an electrode for applying a transfer pulse to be transferred to.

  The Y-direction transfer unit 34 is a transfer unit (second transfer register) that adds a plurality of charges by time-division exposure and transfers the added charges in the vertical direction. The Y-direction transfer unit 34 in the vertical direction of the divided image transfer unit 35 described above. The vertical transfer electrodes S2φV0 to S2φV3 arranged in the vertical direction so as to cross the X-direction transfer unit 36 are adjacent to each arrangement. These vertical transfer electrodes S2φV0 to S2φV3 store an image obtained by adding a plurality of time-division images corrected for blurring, and the added image is an amount corresponding to the blurring in the Y direction of the electronic camera. It is an electrode for applying a transfer pulse to be transferred in the vertical direction.

  Of the above-described electrodes, the electrode SφH2 and the electrode S1φV0, the electrode SφH2 and the electrode S1φV2, and the electrode SφH0 and the electrode S2φV0 are common.

  Moreover, each electrode mentioned above is comprised from the polycrystalline silicon, and adjacent electrodes are arrange | positioned through the interlayer insulation layer. The adjacent electrodes are arranged such that the end faces partially overlap each other when viewed from the direction perpendicular to the imaging surface. In FIG. 3, the end surface of the electrode indicated by a broken line means that the end surface portion of the electrode is disposed below the other electrode.

  The electrodes to which the same drive pulse is applied are connected to each other by a predetermined wiring layer via a contact portion. However, in FIG. 3, in order to clarify the arrangement of the electrodes, the wiring between the electrodes is not shown.

  The correction / addition unit 40 includes a plurality of electrode groups 37 as shown in a range surrounded by a one-dot chain line in FIG. 3 corresponding to each pixel 31 of the imaging unit 41 (note that the imaging of the present embodiment) As will be described later, the element 1 can perform field reading suitable for moving image shooting and can also perform all-pixel reading suitable for still image shooting. When performing, the number of electrode groups 37 that is half of all the electrode groups 37 provided in the correction / addition unit 40 (the number in the horizontal direction is the same, but the number in the vertical direction is half) is used. On the other hand, when all the pixels are read out, all the electrode groups 37 provided in the correction / addition unit 40 are used.

  The electrode group 37 also has two types of electrode configurations.

  First, the first electrode group 37 is an electrode group including electrodes SφH0, SφH3, SφH2, SφH1, electrodes S1φV0, S1φV1, and electrodes S2φV0 to S2φV3. Among these, since there are two common electrodes as described above, the number of electrodes included in the first electrode group 37 is eight.

  The second electrode group 37 is an electrode group including electrodes SφH0, SφH3, SφH2, SφH1, electrodes S1φV2, S1φV3, and electrodes S2φV0 to S2φV3. Since there are two common electrodes among them, the number of electrodes included in the second electrode group 37 is also eight.

  Accordingly, the horizontal transfer electrodes SφH0 to SφH3 constituting the X-direction transfer unit 36 and the vertical transfer electrodes S2φV0 to S2φV3 constituting the Y-direction transfer unit 34 are defined by the first electrode group 37 and the second electrode group 37. In any case, all kinds of electrodes are arranged one by one. This is because the charges stored under the horizontal transfer electrodes SφH0 to SφH3 and the charges stored under the vertical transfer electrodes S2φV0 to S2φV3 do not interfere with each other when transferring charges in response to shaking. It is to make it.

  On the other hand, the vertical transfer electrodes S1φV0 to S1φV3 constituting the divided image transfer unit 35 do not need to have a function of storing charges immediately before addition (this function is provided by the X-direction transfer unit 36 and the Y-direction transfer unit 34). Therefore, only two types of electrodes are arranged in one electrode group 37. Therefore, when one first electrode group 37 and one second electrode group 37 are combined, all kinds of vertical transfer electrodes S1φV0 to S1φV3 constituting the divided image transfer unit 35 are formed one by one. Will be aligned. In this way, the configuration of the electrode is made as simple as possible.

  In such a configuration, readout of pixel charges from the correction / addition unit 40 to the horizontal transfer unit 42 can be performed by two types of paths. That is, both the Y-direction transfer unit 34 and the divided image transfer unit 35 are connected to the horizontal transfer unit 42 and can transfer pixel charges to the horizontal transfer unit 42. Accordingly, the first path is a path that is read out from the horizontal transfer unit 42 to the outside of the image sensor 1 via the Y direction transfer unit 34, and the second path is that from the horizontal transfer unit 42 via the divided image transfer unit 35. This is a path read out to the outside of the image sensor 1.

  The first path is used when transferring the electric charge accumulated under the vertical transfer electrode S2φV2 of the Y-direction transfer unit 34, as will be described later, taken in a time-division manner, shake-corrected, added.

  The second path is used when normal photographing (one-time exposure) is performed and the charge is transferred via the divided image transfer unit 35 in the same manner as the normal frame interline transfer method. It is done. As will be described later, when the exposure time is extremely short compared to the time of blur correction processing by time-division shooting, normal shooting is performed, and at this time, this second path is used.

  A vertical transfer CCD including a vertical transfer electrode for transferring charges in the vertical direction is called a vertical transfer register. A horizontal transfer CCD including a horizontal transfer electrode for transferring charges in the horizontal direction is called a horizontal transfer register. Therefore, the vertical transfer unit 33, the divided image transfer unit 35, and the Y direction transfer unit 34 are vertical transfer registers. The X direction transfer unit 36 and the horizontal transfer unit 42 are horizontal transfer registers. Of these, the Y-direction transfer unit 34 and the X-direction transfer unit 36 further serve to store the charge to be added. The correction / addition unit 40 including the divided image transfer unit 35 and the Y-direction transfer unit 34 that are vertical transfer registers and the X-direction transfer unit 36 that is a horizontal transfer register is a register unit.

  By the way, when the imaging device 1 is increased in number of pixels so that a fine still image can be captured, in moving image shooting that requires fewer pixels than a still image, in order to increase the speed and improve the image quality, A process of adding and reading out pixel data is often performed. For example, when such pixel addition is performed in the imaging unit 41 (specifically, when the readout is performed by adding the charges of the odd lines and the charges of the even lines in the field readout), compared to the case where the addition is not performed. Therefore, the correction of the shake amount is different. When three or more pieces of pixel data are added, similarly, the correction of the blur amount differs depending on the number of additions. Further, when pixel addition is performed, the number of registers necessary for storing image data in the correction / addition unit 40 is also different. If the control of time-division shooting and pixel value addition in the correction / addition unit 40 are made different in accordance with such individual cases, the control becomes complicated.

  Therefore, in order to simplify the control, for example, it is conceivable to perform time-division shooting and pixel value addition in the correction / addition unit 40 only in the case of all-pixel readout frequently used in still image shooting. At this time, time division imaging is not performed in other readout (for example, field readout) (however, of course, the present invention is not limited to this. An example of performing addition will also be described.) When a plurality of time-division shootings are performed, the image of the X direction transfer unit 36 and the image of the Y direction transfer unit 34 are added and accumulated in the Y direction transfer unit 34, and finally the Y direction transfer unit. The added image stored in 34 is transferred to the horizontal transfer unit 42. On the other hand, when time-division shooting is not performed even when all pixels are read, or when other reading is performed, an image transferred from the vertical transfer unit 33 to the divided image transfer unit 35 is directly directly horizontal. The data is transferred to the transfer unit 42. As a result, an image sensor that can cope with still image blur correction and has functions such as pixel addition similar to those of a conventional image sensor and is relatively easy to control. At this time, the X-direction transfer unit 36 only needs to have a function of transferring the time-division image twice or more (that is, the X-direction transfer unit 36 does not need a function of accumulating the added pixels. Therefore, the area of the electrodes in the X-direction transfer unit 36 can be considerably reduced. Here, when focusing on reducing the area of the entire image pickup device 1, the shape and size of each electrode constituting the X-direction transfer unit 36 are set so that the vertical width of the X-direction transfer unit 36 is shortened. Just decide. As a result, the area of the entire correction / addition unit 40 can be reduced. Such a configuration can be similarly applied to FIG. 7 described later.

  Next, FIG. 4 is a timing chart showing the basic operation of the solid-state imaging device 1 during imaging.

  In FIG. 4, VSUB is a substrate applied high voltage pulse for forcibly discharging charges accumulated in the photodiode 32 to a semiconductor substrate (substrate). Further, φV and φH are descriptions collectively referring to all transfer electrodes as shown in FIG.

  When the photographing operation is started, a high voltage pulse is applied to VSUB and a high-frequency transfer pulse is applied to electrodes φV and φH from time t01 at a predetermined timing to exposure (charge accumulation) start time t02. Then, all the charges accumulated in the photodiode 32 and the charges remaining in each transfer CCD are discharged.

  Thereafter, when an exposure operation start signal is received, the application of the high voltage pulse to VSUB is stopped at the timing of time t02, whereby charges corresponding to the intensity of the received light are accumulated in the photodiode 32. The accumulation of charges in the photodiode 32 is continuously performed during the exposure time TExp from time t02 to time t03. Then, this exposure time TExp is time-divided into time intervals so as not to cause blurring, and each time-division image is repeatedly transferred and added according to the blurring amount, so that 1 of the exposure time TExp for which blurring correction has been performed. A sheet of image data is obtained. The operation of the image sensor 1 during the exposure time TExp will be described in more detail later with reference to FIGS.

  When the exposure ends at time t03, the blur accumulated in the potential below the vertical transfer electrode S2φV2 of the Y-direction transfer unit 34 of the correction / addition unit 40 (hereinafter, “potential” is omitted) is corrected. An image is read from the image sensor 1 at a normal reading speed via the horizontal transfer unit 42 (see FIG. 2).

  Next, FIG. 5 is a timing chart illustrating an operation of transferring an image captured by the imaging unit 41 of the solid-state imaging device 1 to the correction / addition unit 40.

  Note that the voltage of the drive pulse applied to the electrodes is held under a high voltage VH (15 [V]) serving as a transfer pulse for reading, a medium voltage VM (0 [V]) for holding charges, and different electrodes. This is a ternary value of a low voltage VL (−5 [V]) for forming a potential barrier that prevents the mixed charges from being mixed. However, among these three values, the high voltage VH is applied only to the electrodes IφV0 and IφV2 that also serve as the gate electrodes.

  Assuming that one time-division exposure of a plurality of time-division exposures is completed, a high voltage VH (15 [V]) is applied to the electrode IφV0 and the electrode IφV2 at time t1. At this time, the electrodes IφV1 and IφV3 and the electrodes S1φV0 to S1φV3 are all set to the low voltage VL (−5 [V]). As a result, the charges accumulated in the photodiode 32 are shifted down to the electrodes IφV0 and IφV2 of the vertical transfer unit 33, respectively. When the shift is performed, the voltage applied to the electrodes IφV0 and IφV2 becomes the medium voltage VM (0 [V]), and the read charges are held (accumulated) under the electrode φH2.

  At the same time as the voltage applied to the electrodes IφV0 and IφV2 changes from the high voltage VH (15 [V]) to the medium voltage VM (0 [V]), the photodiode 32 starts accumulating charges for the next photographing. The That is, the next time-division shooting is started.

  At time t2, since the electrode IφV1 becomes the medium voltage VM and the electrodes IφV0 to IφV2 are at the medium voltage VM, the charges shifted from the photodiode 32 to IφV0 and IφV2 at time t1 are below the electrodes IφV0 to IφV2. Spread and mix. Therefore, here, the charges are mixed for the two pixels that are vertically continuous. This is because the field readout is performed in the same manner as the operation of the normal frame interline transfer side CCD. Therefore, in the next field, reading is performed with different combinations of two vertically continuous pixels.

  At time t3, the electrode IφV0 becomes the low voltage VL, and the electrodes IφV1 and IφV2 are at the medium voltage VM. Therefore, the charges accumulated under the electrodes IφV0 to IφV2 at time t2 move below the electrodes IφV1 and IφV2. To do.

  At time t4, since the electrode IφV3 becomes the medium voltage VM and the electrodes IφV1 to IφV3 are the medium voltage VM, the charges accumulated under the electrodes IφV1 and IφV2 at the time t3 are below the electrodes IφV1 to IφV3. Moving. Further, at this time, the electrode S1φV0 of the correction / addition unit 40 becomes the medium voltage VM, and the charge under the electrodes IφV1 to IφV3 corresponding to the final pixel of the imaging unit 41 is the first electrode S1φV0 of the correction / addition unit 40. It spreads down.

  At time t5, the electrode IφV1 becomes the low voltage VL, and the electrodes IφV2 and IφV3 are at the medium voltage VM. Therefore, the charges accumulated under the electrodes IφV1 to IφV3 at time t4 move below the electrodes IφV2 and IφV3. To do. Further, the charge of the pixel at the final stage of the imaging unit 41 moves below the electrodes IφV2 and IφV3 and the first electrode S1φV0 of the correction / addition unit 40.

  At time t6, the electrode IφV0 becomes the medium voltage VM, and the electrodes IφV0, IφV2, IφV3 are the medium voltage VM. Therefore, the charges accumulated under the electrodes IφV2, IφV3 at the time t5 are Move down. In addition, the charge of the pixel at the final stage of the imaging unit 41 remains accumulated under the electrodes IφV2 and IφV3 and the first electrode S1φV0 of the correction / addition unit 40.

  At time t7, since the electrode IφV2 becomes the low voltage VL and the electrodes IφV0 and IφV3 are the medium voltage VM, the charges accumulated under the electrodes IφV0, IφV2 and IφV3 at the time t6 are the electrodes IφV0, Move down to IφV3. Further, the charge of the pixel at the final stage of the imaging unit 41 moves below the electrode IφV3 and the first electrode S1φV0 of the correction / addition unit 40.

  At time t8, since the electrode IφV1 becomes the medium voltage VM and the electrodes IφV0, IφV1, and IφV3 are the medium voltage VM, the charges accumulated under the electrodes IφV0 and IφV3 at time t7 Move down to IφV3. Further, the charge of the pixel at the final stage of the imaging unit 41 remains accumulated under the electrode IφV3 and the first electrode S1φV0 of the correction / addition unit 40.

  At time t9, since the electrode IφV3 becomes the low voltage VL and the electrodes IφV0 and IφV1 are the medium voltage VM, the charges accumulated under the electrodes IφV0, IφV1 and IφV3 at time t8 Move to. Further, at this time, since the electrode S1φV1 of the correction / addition unit 40 becomes the medium voltage VM and the electrodes S1φV0 and S1φV1 of the correction / addition unit 40 are the medium voltage VM, they are stored under the electrodes IφV3 and S1φV0 at time t8. The charges that have been transferred move below the electrodes S1φV0 and S1φV1.

  At time t10, the electrode IφV2 is at the intermediate voltage VM and the electrodes IφV0 to IφV2 are at the intermediate voltage VM. Therefore, the charges accumulated under the electrodes IφV0 and IφV1 at time t9 move below the electrodes IφV0 to IφV2. To do. Further, at this time, the electrode S1φV0 of the correction / addition unit 40 becomes the low voltage VL, the electrode S1φV2 becomes the medium voltage VM, and the electrodes S1φV1 and S1φV2 are the medium voltage VM. The accumulated charge moves below the electrodes S1φV1 and S1φV2.

  By repeating the above operation, the charges transferred in the field accumulation mode are accumulated under the electrodes S1φV0 to S1φV3 of the correction / addition unit 40 (the divided image transfer unit 35 shown in FIG. 3).

  At the time when the image transfer from the imaging unit 41 to the correction / addition unit 40 is completed, in the correction / addition unit 40, the electrodes S1φV0 and S1φV2 are at the medium voltage VM, and the other electrodes S1φV1 and S1φV3 are at the low voltage VL. It has become. Therefore, electric charges are accumulated under the electrodes S1φV0 and S1φV2.

  In addition, in FIG. 5, the example in which the field reading is performed from the image sensor 1 has been described. However, the image sensor 1 according to the present embodiment may perform all pixel reading by changing the pulse applied to each electrode. It can be done. Here, all pixel readout means that pixel charges are transferred all at once from all the photodiodes 32 provided in the imaging unit 41 to the vertical transfer unit 33, and thereafter, the two pixels as described at the time t2 are transferred to the two pixels. This is readout in which each pixel charge is transferred to the divided image transfer unit 35 of the correction / addition unit 40 without performing such charge mixing (mixing of charges related to field readout). When all the pixels are read out, at the time when the image transfer from the imaging unit 41 to the correction / addition unit 40 is completed, in the correction / addition unit 40, the electrode S1φV0 and the electrode S1φV2 are at the medium voltage VM, and other The electrodes S1φV1 and S1φV3 have a low voltage VL. Accordingly, charges are accumulated under the electrodes S1φV0 and S1φV2.

  Next, FIG. 6 is a timing chart showing a shake correction operation and an addition operation performed in the correction / addition unit 40.

  First, the outline of the shake correction operation and the addition operation in the correction / addition unit 40 will be described.

  In the electrode configuration of the correction / addition unit 40 as described above, the added image is stored (accumulated) under the electrode S2φV2 of the Y-direction transfer unit 34. Then, a vertical transfer pulse is applied to the electrodes S2φV0 to S2φV3 to move the accumulated charges in the vertical direction (Y direction), thereby correcting the blur in the vertical direction of the image. Further, the latest image is stored under the electrode SφH2 of the X-direction transfer unit 36. The accumulated charges are moved in the horizontal direction (X direction) by applying a horizontal transfer pulse to the electrodes SφH0 to SφH3, thereby correcting the horizontal blur of the image. Thereby, the relative blurring of the added image and the latest image in the vertical direction and the horizontal direction is corrected.

  The added image and the latest image whose blur is corrected in this way are, for example, an electrode SφH0 which is an electrode at a position where the X-direction transfer unit 36 and the Y-direction transfer unit 34 intersect (this electrode is an electrode as described above). It is common with S2φV0.)

  The images thus added are stored (accumulated) in the electrode S2φV2 of the Y-direction transfer unit 34 as described above. The position of the electrode S2φV2 does not get in the way when the next shake correction / addition operation is performed (that is, it is not directly adjacent to the X-direction transfer unit 36, and the electrodes S2φV1 and S2φV3 are kept at the low voltage VL). In this position, unnecessary mixing with other charges does not occur.

Therefore, the shake correction / addition operation in the present embodiment is basically performed according to the following procedures (1) to (4).
(1) The charge obtained by the latest time-division imaging is transferred via the X-direction transfer unit 36 according to the amount of shake in the X direction.
(2) The added image obtained by adding the images obtained by the time-division shooting so far is transferred via the Y-direction transfer unit 34 according to the amount of blur in the Y-direction.
(3) The image transferred in (1) and the image transferred in (2) are added under the electrode SφH0 where the X-direction transfer unit 36 and the Y-direction transfer unit 34 intersect.
(4) The image added in (3) is transferred and stored under the electrode S2φV2 closest to the bottom of the Y-direction transfer unit 34 (belonging to the same electrode group 37 as the electrode SφH0 on which the addition has been performed).

  Subsequently, such an operation will be described more specifically with reference to the timing chart of FIG. FIG. 6 shows an example in which the latest image is shifted by one pixel in the left direction on the paper, and then the added image is shifted by one pixel in the upward direction on the paper.

  At time t_1, the electrode SφH2 and the electrode S2φV2 are at the medium voltage VM, and the latest image is accumulated under the electrode SφH2, and the added image is accumulated under the electrode S2φV2. The charge related to the added image among these is held (stored) as it is under the electrode S2φV2 until the later time t_6.

  At time t_2, the electrode SφH3 is at the intermediate voltage VM and the electrodes SφH2 and SφH3 are at the intermediate voltage VM. Therefore, the charge related to the latest image stored under the electrode SφH2 at time t_1 is below the electrodes SφH2 and SφH3. Move to.

  At time t_3, since the electrode SφH2 becomes the low voltage VL, the electrode SφH0 becomes the medium voltage VM, and the electrodes SφH0 and SφH3 are the medium voltage VM, the latest image accumulated under the electrodes SφH2 and SφH3 at the time t_2. The electric charge related to moves to the lower side of the electrodes SφH0 and SφH3.

  At time t_4, the electrode SφH3 becomes the low voltage VL, the electrode SφH1 becomes the medium voltage VM, and the electrodes SφH0 and SφH1 are the medium voltage VM. Therefore, the charges accumulated under the electrodes SφH0 and SφH3 at time t_3 are , Move below the electrodes SφH0 and SφH1.

  At time t_5, the electrode SφH0 is at the low voltage VL, the electrode SφH2 is at the medium voltage VM, and the electrodes SφH1 and SφH2 are at the medium voltage VM, so that the charges accumulated under the electrodes SφH0 and SφH1 at time t_4 are , Move below the electrodes SφH1 and SφH2.

  At time t_6, since the electrode SφH1 is at the low voltage VL and the electrode SφH2 is at the intermediate voltage VM, the charge accumulated under the electrodes SφH1 and SφH2 at time t_5 moves below the electrode SφH2.

  By performing such an operation, the latest image is moved by one pixel in the horizontal (left) direction, and a relative blur (for example, a blur for one pixel) in the horizontal direction is corrected.

  Next, at time t_7, since the electrode S2φV1 becomes the medium voltage VM and the electrodes S2φV1 and S2φV2 are the medium voltage VM, the charges related to the added image accumulated under the electrode S2φV2 until then are the electrodes S2φV1, Move down S2φV2. At this time, since the electrode SφH2 remains at the medium voltage VM, the latest image remains stored under the electrode SφH2. The charge related to the latest image is held (stored) as it is under the electrode SφH2 until a later time t_10.

  At time t_8, the electrode S2φV2 becomes the low voltage VL, the electrode S2φV0 becomes the medium voltage VM, and the electrodes S2φV0 and S2φV1 are the medium voltage VM. Therefore, the charges accumulated under the electrodes S2φV1 and S2φV2 are It moves below the electrodes S2φV0 and S2φV1. At this time, in the X-direction transfer unit 36, not only the electrode SφH2 but also the electrode SφH0 has the medium voltage VM, but another electrode (electrode SφH1) having a low voltage is provided between the electrode SφH0 and the electrode SφH2. And the electrode SφH3), the charge related to the latest image stored under the electrode SφH2 is affected by the charge related to the addition image transferred through the Y-direction transfer unit 34 at the time t_7. There is no.

  At time t_9, the electrode S2φV1 becomes the low voltage VL, the electrode S2φV3 becomes the medium voltage VM, and the electrodes S2φV0 and S2φV3 are the medium voltage VM. Therefore, the charges accumulated up to that time under the electrodes S2φV0 and S2φV1 are It moves below the electrodes S2φV0 and S2φV3. At this time as well, the electrode SφH0 becomes the medium voltage VM, but for the reason described at the time t_8, the charge related to the latest image stored under the electrode SφH2 is transferred through the Y-direction transfer unit 34 at the time t_9. It is not affected by the charge related to the added image.

  At time t_10, the electrode S2φV0 becomes the low voltage VL, the electrode S2φV2 becomes the medium voltage VM, and the electrodes S2φV2 and S2φV3 are the medium voltage VM. Therefore, the charges accumulated up to that time under the electrodes S2φV0 and S2φV3 are It moves below the electrodes S2φV2 and S2φV3.

  At time t_11, the electrode S2φV3 becomes the low voltage VL, the electrode S2φV1 becomes the medium voltage VM, and the electrodes S2φV1 and S2φV2 are the medium voltage VM. Are transferred to the lower side of the electrodes S2φV1 and S2φV2. At this time, since the electrode SφH3 is at the intermediate voltage VM and the electrodes SφH2 and SφH3 are at the intermediate voltage VM in the X-direction transfer unit 36, the charges accumulated under the electrode SφH2 until time t_10 are below the electrodes SφH2 and SφH3. Move to.

  By performing such an operation, the charge related to the addition image stored under a certain electrode S2φV2 is moved under the electrode S2φV2 that is one pixel above, and the relative blur in the vertical direction (for example, one pixel) Blur) is corrected.

  At time t_12, the electrode S2φV2 becomes the low voltage VL, the electrode S2φV0 becomes the medium voltage VM, and the electrodes S2φV0 and S2φV1 are the medium voltage VM. Are transferred to the lower side of the electrodes S2φV0 and S2φV1. Further, at this time, the electrode SφH2 becomes the low voltage VL, the electrode SφH0 is common with the electrode S2φV0, and thus becomes the intermediate voltage VM, and the electrodes SφH0 and SφH3 are the intermediate voltage VM, so that the electrodes SφH2 and SφH3 at time t_11. The charges accumulated below move below the electrodes SφH0 and SφH3. Therefore, at time t_12, the added image and the latest image are mixed via the electrode SφH0 (that is, the electrode S2φV0) that is the common electrode, and a new added charge is generated. This new added charge is held (accumulated) under three electrodes, that is, electrode SφH0 (that is, electrode S2φV0), electrode SφH3, and electrode S2φV1.

  At time t_13, the electrode SφH3 becomes the low voltage VL, and the electrodes S2φV0 and S2φV1 remain at the medium voltage VM. The charged charges move below the electrodes S2φV0 and S2φV1.

  At time t_14, the electrode S2φV0 becomes the low voltage VL, the electrode S2φV2 becomes the medium voltage VM, and the electrodes S2φV1 and S2φV2 are the medium voltage VM. Therefore, the charges accumulated under the electrodes S2φV0 and S2φV1 at time t_13 are Are transferred to the lower side of the electrodes S2φV1 and S2φV2. At this time, no electric charge remains under the electrodes SφH0 to SφH3 of the X-direction transfer unit 36.

  At time t_15, since the electrode S2φV1 becomes the low voltage VL and the electrode S2φV2 is at the medium voltage VM, the charges accumulated under the electrodes S2φV1 and S2φV2 at the time t_14 are transferred to the lower side of the electrode S2φV2.

  By such an operation, the latest time-division image and the image added so far are newly added after mutual blurring is corrected and stored under the electrode S2φV2. The newly added image is stored under the electrode S2φV2 at time t_15, as described above, when the next time-division image is transferred in the X-direction transfer unit 36 (FIG. 3). This is to prevent the already added image and the latest image from interfering with each other.

  In the above description, an example in which the blur amount in the X direction is for one pixel and the blur amount in the Y direction is for one pixel has been described. However, in the case of a more general blur amount, the blur amount depends on the blur amount. It is only necessary to transfer the same amount. Moreover, even if the shake in the X direction is either the positive direction or the negative direction in the X direction, the shake correction can be freely performed by controlling the phase of the pulse applied to the electrode. The same applies to the blur in the Y direction.

  In this way, four electrodes for the X-direction transfer unit 36 and the Y-direction transfer unit 34 in the correction / addition unit 40 of the image sensor 1 are provided for one pixel (however, the electrodes (SφH0, S2φV0) at the intersection positions are common). Therefore, when transferring for blur correction, the charge related to the new image in the X-direction transfer unit 36 and the charge related to the added image of the Y-direction transfer unit 34 interfere with each other. So that there is no.

  Next, a modification of the image sensor 1 will be described with reference to FIG. FIG. 7 is an enlarged view of the configuration of the imaging unit 41 and the correction / addition unit 40 of the imaging device 1 in the modification.

  First, the overall configuration of the image sensor 1 of this modification is the same as that shown in FIG. 2 described above. Furthermore, the configuration of the imaging unit 41 of the imaging device 1 of this modification is the same as that of the imaging device 1 shown in FIG. In addition, the correction / addition unit 40 of the image pickup device 1 of this modification includes a divided image transfer unit 35, an X direction transfer unit 36, and a Y direction transfer unit 34. This is the same as the image sensor 1 shown in FIG.

  However, the correction / addition unit 40 of the image pickup device 1 of this modification is configured such that the configuration of the X-direction transfer unit 36 and the configuration of the Y-direction transfer unit 34 are slightly different from those of the image pickup device 1 shown in FIG. It has become.

  First, the X-direction transfer unit 36 includes four transfer electrodes SφH0 to SφH3 per pixel, and the Y-direction transfer unit 34 includes four transfer electrodes S2φV0 to S2φV3 per pixel. The so-called four-phase CCD that is driven by an independent transfer pulse is the same as that of the image sensor 1 shown in FIG.

  On the other hand, in the image pickup device 1 of this modification, the area of the electrode SφH1 and the electrode SφH3 of the X-direction transfer unit 36 is compared with the area of the electrode SφH0 and the electrode SφH2 as compared with the image pickup device 1 shown in FIG. The difference is that the areas of the electrodes S2φV1 and S2φV3 of the Y-direction transfer section 34 are narrower than the areas of the electrodes S2φV0 and S2φV2. Thus, the X-direction transfer unit 36 and the Y-direction transfer unit 34 are four-phase CCDs configured by alternately arranging electrodes having a relatively large area and electrodes having a relatively small area.

  It is to be noted that each electrode is made of polycrystalline silicon, and the end face shown by the broken line of the electrode is arranged below the other electrode, as in the example shown in FIG. Also in FIG. 7, as in FIG. 3 described above, in order to clarify the arrangement of the electrodes, the wiring between the electrodes is not shown, but the electrodes to which the same drive pulse is applied are not shown. These are connected to each other by a predetermined wiring layer through a contact portion.

  The electrodes SφH1 and SφH3 of the X-direction transfer unit 36 are used only for charge transfer, and charge accumulation is performed only under the electrodes SφH0 and SφH2. Similarly, the electrodes S2φV1 and S2φV3 of the Y-direction transfer unit 34 are used only for charge transfer, and charge accumulation is performed only under the electrodes S2φV0 and S2φV2.

  By adopting such a non-uniform electrode configuration, it is possible to increase the charge storage capacity in the X-direction transfer unit 36 and the Y-direction transfer unit 34, and to reduce the area of the correction / addition unit 40. Accordingly, it is possible to reduce the size of the image sensor. In addition, reducing the substrate area improves the yield of the image sensor at the time of manufacturing and contributes to lowering the manufacturing cost.

  Next, the operation of the image sensor 1 of this modification will be described. First, the overall operation of the image sensor 1 is the same as that described with reference to FIG. Next, the operation of transferring an image from the imaging unit 41 of the imaging device 1 to the divided image transfer unit 35 of the correction / addition unit 40 is the same as that described with reference to FIG.

  On the other hand, the outline of the operation of the correction / addition unit 40 of the imaging element 1 of the modification is as follows.

  The operation of the image sensor 1 as described with reference to FIG. 3 described above corrects the amount of blurring in the horizontal direction from the previous imaging for the latest image transferred from the imaging unit 41 to the correction / addition unit 40. After the image is transferred in the horizontal direction in the X-direction transfer unit 36, the previous added image stored in the Y-direction transfer unit 34 is vertically converted in the Y-direction transfer unit 34 so as to correct the amount of blur in the vertical direction. Then, the latest image and the added image are added under the electrodes (SφH0, S2φV0) where the X direction transfer unit 36 and the Y direction transfer unit 34 intersect, and then the Y direction transfer is performed. The data was again stored under the electrode S2φV2 of the part 34.

  On the other hand, in the operation of the image sensor 1 of this modification, first, the Y-direction transfer unit 34 is used to correct the shake amount in the vertical direction from the previous addition image stored in the Y-direction transfer unit 34. After the image is transferred in the vertical direction, the latest image transferred from the imaging unit 41 to the correction / addition unit 40 and stored in the X-direction transfer unit 36 is corrected for the amount of horizontal blurring from the previous imaging. Thus, the image is transferred in the horizontal direction in the X-direction transfer unit 36, and then the latest image and the added image are added and stored under the electrode S2φV2 of the Y-direction transfer unit 34.

  Such an operation of the correction / addition unit 40 of the imaging device 1 according to the modification will be described in detail with reference to FIG. FIG. 8 is a timing chart showing a shake correction operation and an addition operation performed in the correction / addition unit 40. FIG. 8 shows an example in which the added image is shifted by one pixel in the upward direction on the paper, and the latest image is shifted by one pixel in the left direction on the paper.

  At time t−1, the electrode SφH2 and the electrode S2φV2 are at the medium voltage VM, and the latest image is displayed below the electrode SφH2 (electrode S1φV0 and electrode S1φV2) in the X-direction transfer unit 36 of the correction / addition unit 40. The added images are stored under the electrode S2φV2 in the direction transfer unit 34, respectively. Of these, the charge related to the latest image is held (stored) as it is under the electrode SφH2 until a later time t-9.

  At the time t−2, the electrode S2φV1 becomes the medium voltage VM and the electrodes S2φV1 and S2φV2 are the medium voltage VM. Therefore, the image (addition image) stored under the electrode S2φV2 at the time t−1 is the electrode Move down S2φV1, S2φV2.

  At time t-3, the electrode S2φV0 becomes the medium voltage VM, and the electrodes S2φV0 to S2φV2 are the medium voltage VM, so that the image (addition image) stored under the electrodes S2φV1 and the electrode S2φV2 at the time t-2 Move below the electrodes S2φV0 to S2φV2.

  At time t-4, since the electrode S2φV2 becomes the low voltage VL and the electrodes S2φV0 and S2φV1 are the medium voltage VM, the image (addition image) stored under the electrodes S2φV0 to S2φV2 at the time t-3 is , Move to electrodes S2φV0 and S2φV1.

  At time t-5, the electrode S2φV3 is at the intermediate voltage VM, and the electrodes S2φV0, S2φV1, and S2φV3 are at the intermediate voltage VM. Therefore, the image stored under the electrodes S2φV0 and the electrode S2φV1 at time t-4 (addition) The image) moves below the electrodes S2φV0, S2φV1, S2φV3.

  At time t-6, since the electrode S2φV1 becomes the low voltage VL and the electrodes S2φV0, S2φV3 are at the medium voltage VM, the images (additional images) stored under the electrodes S2φV0, S2φV1, S2φV3 at the time t-5 ) Moves below the electrodes S2φV0 and S2φV3.

  At time t-7, the electrode S2φV2 is at the medium voltage VM and the electrodes S2φV0, S2φV2, and S2φV3 are at the medium voltage VM, so the image (addition image) stored under the electrodes S2φV0 and S2φV3 at the time t-6 ) Moves below the electrodes S2φV0, S2φV2, and S2φV3.

  At time t-8, since the electrode S2φV0 becomes the low voltage VL and the electrodes S2φV2 and S2φV3 are at the medium voltage VM, the images (additional images) stored under the electrodes S2φV0, S2φV2, and S2φV3 at time t-7 ) Moves below the electrodes S2φV2 and S2φV3.

  At time t-9, since the electrode S2φV3 is at the low voltage VL and the electrode S2φV2 is at the medium voltage VM, the image (added image) stored under the electrodes S2φV2 and S2φV3 at time t-8 is Move down S2φV2.

  By performing such an operation, the added image stored in the Y-direction transfer unit 34 is shifted by one pixel in the upward direction in FIG. In addition, when moving for a plurality of pixels, the same operation may be repeated for the number of pixels. Then, the charge related to the added image under the electrode S2φV2 is held (stored) as it is until a later time t-12.

  In this way, when the addition image is transferred by a predetermined number of pixels via the Y-direction transfer unit 34, the electrode SφH3 of the X-direction transfer unit 36 is set to the medium voltage VM in synchronization with the electrode S2φV3 becoming the low voltage VL.

  At time t-10, since the electrodes S.phi.H2 and S.phi.H3 are at the medium voltage VM, the latest image transferred from the imaging unit 41 to the correction / addition unit 40 and stored under the electrode S.phi.H2 until time t-9. Moves below the electrodes SφH2, SφH3.

  At time t-11, the electrode SφH0 is at the medium voltage VM and the electrodes SφH0, SφH2, SφH3 are at the medium voltage VM, so that the image stored under the electrodes SφH2, SφH3 at time t-10 It moves under SφH0, SφH2, and SφH3.

  At time t-12, since the electrode SφH2 is at the low voltage VL and the electrodes SφH0, SφH3 are at the medium voltage VM, the image stored under the electrodes SφH0, SφH2, SφH3 at the time t-11 It moves under SφH0 and SφH3.

  At time t-13, the electrode S2φV1 is at the intermediate voltage VM, and the electrodes SφH0 (S2φV0), SφH3, S2φV1, and S2φV2 are at the intermediate voltage VM, so that they are stored under the electrodes SφH0 and SφH3 at the time t-12. The charge related to the latest image and the charge related to the image (added image) stored under the electrode S2φV2 are diffused under the electrodes SφH0 (S2φV0), SφH3, S2φV1, and S2φV2. Therefore, at this time, the charge related to the latest image and the charge related to the added image are mixed.

  At time t-14, the electrode SφH3 is at the low voltage VL, and the electrodes SφH0 (S2φV0), S2φV1, and S2φV2 are at the medium voltage VM. The image stored below moves below the electrodes SφH0 (S2φV0), S2φV1, and S2φV2.

  At time t-15, since the electrode SφH0 (S2φV0) becomes the low voltage VL and the electrodes S2φV1 and S2φV2 are at the medium voltage VM, they are stored under the electrodes SφH0 (S2φV0), S2φV1, S2φV2 at time t-14. The image that has been moved moves below the electrodes S2φV1 and S2φV2.

  At time t-16, since the electrode S2φV1 is at the low voltage VL and the electrode S2φV2 is at the medium voltage VM, the image stored under the electrodes S2φV1 and S2φV2 at time t-15 moves below the electrode S2φV2. To do.

  By performing such an operation, the latest image and the time-division image (added image) obtained by correcting and adding the shake up to the previous time are added, and are added under the electrode S2φV2 of the Y-direction transfer unit 34. Is remembered.

  By the way, in the present embodiment, since the charge of the pixel from a plurality of shootings is added, the charge of the pixel exceeds the storage capacity of the register of the Y-direction transfer unit 34 as shown in FIG. 3 or FIG. There is a possibility of overflowing. Below, the structure for preventing this is demonstrated with reference to FIG. 9 and FIG. FIG. 9 is a cross-sectional view for explaining that the imaging device has a vertical overflow structure, and FIG. 10 shows a state in which the reverse charge voltage is changed according to the number of time-division exposures to control the amount of accumulated charge that causes overflow. FIG.

  When strong light is incident on the photodiode 32, the charge overflowing from the photodiode 32 may flow into the adjacent photodiode. In order to prevent this blooming phenomenon, various devices have been conventionally made. In the present embodiment, in order to add charges inside the image sensor 1, for example, each exposure amount in n (n is an integer of 1 or more) time-division shooting is set to 1 / of the exposure amount in normal shooting. Even if n is set, the charge of the image obtained by adding the time-division images may overflow beyond the storage capacity of the Y-direction transfer unit 34. Therefore, it is necessary to prevent the charge of the image after the addition from exceeding the storage capacity of the Y-direction transfer unit 34.

  For this reason, in the present embodiment, a CCD having a vertical overflow structure is employed to discharge surplus charges to the substrate. As shown in FIG. 9, this vertical overflow CCD has a photodiode in a p-type diffusion region (p-well) formed on the surface of an n-type substrate (for example, an n-type silicon substrate). (PD), and the p well is completely depleted by the reverse bias voltage VSUB between the p well and the n-type substrate.

  As shown in FIG. 10, the gate for reading from the photodiode (PD) to the vertical transfer CCD (vertical transfer unit 33) is common to the transfer electrodes IφV0 and IφV2. The read gate potential φVRG is lower than the photodiode potential φVPD by a predetermined potential when a low voltage (VL) or a medium voltage (VM) is applied to the transfer electrode. On the other hand, when a high voltage (VH) pulse is applied to the transfer electrode, the potential φVRG of the readout gate becomes higher than the potential φVPD of the photodiode, and the signal charge in the photodiode is transferred to the vertical transfer CCD (vertical transfer unit 33). ).

  At this time, the vertical transfer CCD has a limited handling charge amount, which is the maximum charge amount that can be transferred, and therefore requires an operation of discharging excess charge when strong light is incident. This excess signal charge is absorbed into the substrate through an n + pn path in the vertical direction (substrate thickness direction) before overflowing to the transfer CCD side or another photodiode side. That is, the potential φVOFG of the p-well where overflow occurs is higher than the read gate voltage φVRG when the voltage of the transfer electrode that also serves as the read gate is VM (0 [V]) or VL (−5 [V]). Is set to This suppresses the phenomenon of blooming in which excessive charge leaks to the vertical transfer CCD during storage.

  Further, in the present embodiment, the reverse bias voltage (substrate voltage) VSUB between the p-well and the n-type substrate is set to a higher voltage (FIG. 10) than in normal imaging according to the number m of time-division imaging. (That is, VSUB 'depends on m), the potential of the p-well where overflow occurs is set to φVOFG' (> φVOFG), and the photo is taken during normal shooting. Excess charge exceeding 1 / m of the maximum charge accumulated in the diode is discharged to the substrate side. In this way, even if images obtained by m time-division photographing are added, the total charge amount after the addition can be suppressed within the handling charge amount of the Y-direction transfer unit 34 shown in FIG. 3 or FIG. it can.

  FIG. 11 is a diagram showing several examples of how the maximum value Q of the accumulated charge amount of the photodiode 32 that changes according to the substrate voltage VSUB is set according to the number m of time-division imaging. In FIG. 11, the horizontal axis represents the substrate voltage VSUB, and the vertical axis represents the maximum value Q (max) of the accumulated charge amount of the photodiode 32.

  In the present embodiment, ten values of m = 1 to 10 can be set as the number m of time-division shooting (in FIG. 11, examples of m = 1, 2, 4, 8). At this time, VSUB is V1, V2, V4, and V8 as shown in the figure). The maximum value Q (max) of the stored charge amount of the photodiode 32 is 1 when m = 1, and is 1 / m for each m when m takes a value of 2 to 10 (m The rule that Q (max) is 1 / m can be applied as it is even when the value is set to take a value of 11 or more.) Further, m = 1 corresponds to the maximum value of the accumulated charge amount of the photodiode 32 in the case of normal shooting (shooting not based on time division shooting).

  The reason why the maximum value of m is set to 10 is that if the value is too large, it is difficult to maintain the S / N of the image data at a high value.

  In addition, since the technique of the present embodiment is to perform image division m times with the blur limit exposure time TLimit and add the images, conventionally, the exposure time that can tolerate blur is set to TExp0 (for example, a 35 mm film camera). When the converted focal length of the photographic lens is 300 mm and the shutter speed is approximately 1/300 seconds, in the present embodiment, it is possible to shoot up to an exposure time of m · TExp0 (as described above, photographic When the focal length of the lens is 300 mm, it is approximately 1/30 second) (that is, m · TExp0 is the longest exposure time for which blur correction is possible). That is, as the value of m is larger, it is possible to perform blur correction for longer time exposures. However, as described above, if the value of m is increased too much, the S / N of image data may be lowered. There is. Therefore, it is desirable to determine the value of m in consideration of the balance between the stability of control and the S / N of images obtained by divided shooting. Thus, although the maximum value of m is set to 10 in the present embodiment, it is not necessary to be limited to this.

  Next, FIG. 12 is a flowchart illustrating processing when an image is captured and recorded by the electronic camera.

  When the power of the electronic camera is turned on (for example, the battery is replaced) or an operation start switch (not shown) (for example, a power switch) is operated, the operation of the electronic camera starts.

  When the process is started, after setting a predetermined initial value or the like, it is first determined whether or not the first release switch 18a is closed by the release operation of the photographer (step S1).

  If the first release switch 18a is not closed, the process branches to J1 and the detection of the first release switch 18a is repeated in the same manner. However, actually, an operation for performing display or detecting other key input states (not shown) is performed between J1 and step S1, but these general operations are also appropriately described below. The description will be omitted.

  If it is detected in step S1 that the first release switch 18a is closed, the blur limit exposure time TLimit is then calculated (step S2). This blur limit exposure time TLimit is a time when the blur amount from the start of exposure reaches the allowable limit blur amount.

  Here, the blur limit exposure time TLimit will be described. As a long-standing empirical rule for a so-called Leica frame (also known as double frame) camera with a length of 24 mm and a width of 36 mm (diagonal 43.28 mm) in a 35 mm film camera, when the focal length of the taking lens in millimeters is f It is known that the blur limit exposure time TLimit becomes TLimit≈1 / f (seconds). In the present embodiment, this rule of thumb is applied in consideration of the size of the photographic image frame set in the effective imaging area of the image sensor of the electronic camera. In the following description, description about the unit mm is omitted as appropriate, and only numerical values are described.

  Since the subject imaging range (shooting angle of view) in the electronic camera is inversely proportional to the focal length of the taking lens and proportional to the size of the frame, the relative image magnification relative to the frame is proportional to the focal length and the frame. Is inversely proportional to the size of Therefore, image blur (relative to the frame) is also inversely proportional to the frame size. Therefore, it is only necessary to multiply the empirical rule by a conversion factor corresponding to the frame size ratio and apply this. In this case, it is necessary to consider the difference in aspect ratio, but in the electronic camera according to the present embodiment, diagonal conversion is used.

When the diagonal length of the photographic image frame is d [mm], TLimit is expressed by the following Equation 1.
[Equation 1]

  When Expression 1 is expressed as TLimit = 1 / f ', f' is generally a focal length of a photographic lens having the same angle of view as that of a photographic lens of a 35 mm film camera.

  As is clear from Equation 1, when the photographing lens 15 of the electronic camera is a single focal length photographing lens, only one value corresponding to the focal length needs to be applied as the blur limit exposure time TLimit. However, when the photographing lens 15 is a zoom lens, it is necessary to apply different values according to the focal length.

  Note that the blur limit exposure time TLimit does not necessarily need to use the value given by Equation 1, in short, an exposure time that does not substantially cause blurring may be used. Therefore, the blur limit exposure time TLimit may be a time shorter than the exposure time given by Equation 1 in general.

  Next, the brightness of the subject is measured (step S3). In this photometry, the level of an image signal repeatedly output from the image sensor 1 is monitored, and the brightness of the subject is calculated. That is, the image signal read from the image sensor 1 is processed by the CDS 2, converted to a digital value by the A / D converter 4, and temporarily stored in the DRAM 9 through the information processing unit 8. Among the image signals stored in the DRAM 9, an image signal in a predetermined area, for example, near the center of the entire image is read by the CPU 7, and an average value of the levels is obtained. Then, the CPU 7 calculates the brightness of the subject based on the obtained addition average value.

  Subsequently, the CPU 7 calculates the shutter speed value (exposure time) TExp and the aperture value of the aperture 17 necessary for obtaining proper exposure, and sets the aperture setting of the aperture 17 via the aperture drive system 16 based on the calculation result. A subroutine for performing the process (see FIG. 13) is executed (step S4).

  FIG. 13 is a flowchart showing a subroutine for calculating and setting TExp and aperture value.

  When this subroutine is entered, it is first determined whether or not the shutter priority shooting mode is set (step S21).

If it is the shutter priority shooting mode,
TExp> m (max) · TLimit
It is determined whether or not (step S22). Here, m (max) is the maximum value of the number m of time-division exposures as described above. In the present embodiment, since m can be any one of 1 to 10, m (max) = 10 here.

  Here, when TExp> m (max) · TLimit, if the exposure time TExp set by the photographer is used as it is, the number of time-division exposures becomes m (max) or more, and the image S / S N cannot be secured. Therefore, in this case, the limit value m (max) · TLimit is stored in the memory [TExp] for storing the exposure time TExp (step S23). [] Means a memory for storing data in parentheses.

When the process of step S23 is completed, an aperture value (Av) is calculated (step S24). Based on the Apex (APEX (Additive System of Photographic Exposure)) method, the exposure value, exposure time, aperture value, subject brightness, and ISO sensitivity are expressed as an index based on the APEX method, respectively Ev (Exosure Vaulue), When Tv (Time Value), Av (Aperture Value), Bv (Brightness value), and Sv (Sensitive Value) are used, the following equation holds.
Ev = Tv + Av = Bv + Sv

  Therefore, if the brightness (Bv) of the subject obtained by photometry and the ISO sensitivity (Sv) set by the photographing condition setting unit 26 are known, the exposure value (Ev) is determined, and Tv and Av are calculated. can do. Here, since the exposure time TExp (and thus Tv) has already been set in step S23, Av can be obtained, and the aperture value corresponding to the obtained Av is calculated.

  Next, the diaphragm 17 is set via the diaphragm drive system 16 so that the diaphragm value obtained in step S24 is obtained (step S25), and thereafter, the process returns to the process shown in FIG.

  If it is determined in step S22 that TExp> m (max) · TLimit is not satisfied, it is next determined whether or not TExp ≦ TLimit (step S26).

  Here, when TExp ≦ TLimit, since time-division exposure is not performed, TExp uses the value set by the photographer as it is. Then, the process proceeds to J31, and the aperture value Av in step S24 and the aperture 17 in step S25 are set in the same manner as described above.

  On the other hand, when it is determined in step S26 that TExp ≦ TLimit does not hold, time-division shooting is performed, so the photographer selects from m · TLimit (m = 2 to 10). The closest exposure time TExp is selected as the exposure time (step S27). More specifically, first, TExp / TLimit is calculated, m value closest to the calculated value is selected, m · TLimit is calculated, and this calculated value m · TLimit is set as the exposure time TExp. Note that the value of m calculated here is also used to calculate the reverse bias voltage VSUB 'as described above. As a result, it is possible to prevent the added charge from overflowing beyond the charge handled by the CCD of the Y-direction transfer unit 34 as described above. When step S27 ends, the process proceeds to J31, and the processes of steps S24 and S25 are performed in the same manner as described above.

  As described above, in the shutter priority mode, the exposure time and the aperture value are set with the highest priority given to the exposure time set by the photographer. At this time, an overflow of electric charges in the Y-direction transfer unit 34 in time-division shooting is also prevented.

  On the other hand, if it is determined in step S21 described above that the shutter priority mode is not set, it is determined whether or not the aperture priority mode is set (step S28).

  If it is the aperture priority mode, the exposure time TExp is calculated based on the apex calculation (step S29).

  Then, it is determined whether TExp> m (max) · TLimit (step S30).

  If TExp> m (max) · TLimit, the limit value m (max) · TLimit is set to the exposure time TExp, that is, TExp = m (max) · TLimit (step S31). Thereafter, the process proceeds to J31, and the processes of step S24 and step S25 are performed as described above.

  If it is determined in step S30 described above that TExp> m (max) · TLimit, it is determined whether TExp ≦ TLimit (step S32).

  Here, when TExp ≦ TLimit, since the time-division exposure is not performed, TExp uses the value obtained in step S29 as it is. And it transfers to J31 and performs the process of step S24 and step S25 similarly to the above-mentioned.

  On the other hand, if it is determined in step S32 that TExp ≦ TLimit does not hold, time-division shooting is performed, so that the exposure time TExp obtained in step S29 from m · TLimit (m = 2 to 10) is obtained. Is selected as the exposure time (step S33). More specifically, first, TExp / TLimit is calculated, m value closest to the calculated value is selected, m · TLimit is calculated, and this calculated value m · TLimit is set as the exposure time TExp. Note that the value of m calculated here is also used to calculate the reverse bias voltage VSUB 'as described above. Thereby, the overflow of the charge of the CCD of the Y-direction transfer unit 34 can be suppressed. When step S33 ends, the process proceeds to J31, and the processes of step S24 and step S25 are performed as described above.

  As described above, in the aperture priority mode, the aperture value and exposure time are set with the highest priority given to the aperture value set by the photographer. At this time, an overflow of electric charges in the Y-direction transfer unit 34 in time-division shooting is also prevented.

  If it is determined in step S28 that the aperture priority shooting mode is not selected (that is, in the program mode), it is determined whether TExp> m (max) · TLimit (step S34).

  If TExp> m (max) · TLimit, the limit value m (max) · TLimit is set to the exposure time TExp, that is, TExp = m (max) · TLimit (step S35). Thereafter, the process proceeds to J31, and the processes of step S24 and step S25 are performed as described above.

  On the other hand, if it is determined in step S34 that TExp> m (max) · TLimit is not satisfied, it is determined whether TExp ≦ TLimit (step S36).

  Here, when TExp ≦ TLimit, since time-division exposure is not performed, TExp and the aperture value are obtained from the program diagram (step S37). The program diagram describes the relationship between the exposure value Ev and the combination of Tv (exposure time apex value) and Av (aperture value apex value) necessary for optimal exposure control. Yes, as described above, it is stored in the EEPROM 24 built in the CPU 7. Thereafter, the process goes to step S25, and the diaphragm 17 is set via the diaphragm drive system 16 so that the diaphragm value obtained in step S37 is obtained.

  If it is determined in step S36 that TExp ≦ TLimit, since time-division shooting is performed, refer to the program diagram from m · TLimit (m = 2 to 10). The exposure time closest to the obtained exposure time TExp is selected as the exposure time (step S38). More specifically, first, TExp / TLimit is calculated using the exposure time TExp obtained with reference to the program diagram, m · TLimit is calculated by selecting the value of m closest to the calculated value, This calculated value m · TLimit is set as the exposure time TExp. Note that the value of m calculated here is also used to calculate the reverse bias voltage VSUB 'as described above. Thereby, the overflow of the charge of the CCD of the Y-direction transfer unit 34 can be suppressed. When step S38 is completed, the process proceeds to J31, and the processes of step S24 and step S25 are performed as described above.

  As described above, in the program photographing mode, the aperture value and the exposure time are set with the highest priority given to the combination of the exposure time and the aperture value obtained with reference to the program diagram. At this time, an overflow of electric charges in the Y-direction transfer unit 34 in time-division shooting is also prevented.

  In this way, a subroutine for calculating and setting TExp and aperture value is performed.

  Returning to the description of FIG. 12, when the subroutine of step S4 as described above is completed, it is next detected whether or not the second release switch 18b is closed (step S5). If the second release switch 18b is not closed, as long as the first release switch 18a is closed, the process branches to J2 and waits for the second release switch 18b to close.

  In this way, when it is detected that the second release switch 18b is closed, the initial value “0” is stored in the memory [n] that stores the number n of times of actual exposure by the blur limit exposure time TLimit. (Step S6). Here, the blur limit exposure time TLimit is an exposure time that is less than or equal to a level at which the occurrence of blur can be substantially ignored, as will be described in detail later. Moreover, as described above, [] means a memory for storing data in parentheses. Therefore, [n] means a memory for storing the variable n in parentheses.

  Next, exposure is started (step S7). As shown in FIG. 4, the imaging device 1 has a substrate applied high voltage pulse VSUB for forcibly discharging the charge accumulated in the photodiode 32 to the semiconductor substrate (substrate = vertical overflow drain VOFD) immediately before the start of exposure. The time when the application of the VSUB pulse is completed and the value of VSUB is set to a value corresponding to m is the time when the exposure is started.

  Next, it is determined whether or not one time-division exposure has been completed (step S8). Here, until one time-division exposure is completed, the process branches to J3 and waits for completion of the time-division exposure.

  When one time-division exposure is completed, a subroutine “pixel value synthesis” (see FIG. 17) for synthesizing an image obtained by time-division shooting is executed (step S9).

  Before describing this pixel value synthesis subroutine, the detection of the amount of blur in the electronic camera will be described with reference to FIGS. First, FIG. 14 is a diagram showing the coordinate axes set in the electronic camera and the arrangement of two angular velocity sensors.

  In FIG. 14, at a certain time, the subject side along the optical axis O of the photographing lens 15 is viewed in the positive direction of the Z axis, and the electronic camera is viewed from the subject side in the lateral direction of the electronic camera in the imaging plane perpendicular to the Z axis. The X axis, the Y axis, and the right direction of the electronic camera passing through the intersection of the Z axis and the X axis and orthogonal to the Z axis and the X axis are the positive direction of the Y axis. The Z axis is set. The rotation angles around the X, Y, and Z axes are denoted by θx, θy, and θz, respectively. Note that the optical axis O and the Z axis of the photographing lens 15 coincide with each other at a certain time described above, but the optical axis O of the photographing lens 15 is generally set when there is a shake at another time. It does not match the Z axis.

  The camera body 51 of the electronic camera is provided with the angular velocity sensor 19 and the angular velocity sensor 20 as described above.

  The angular velocity sensor 19 is for detecting an angular velocity that is a change amount per unit time of the rotation angle θx around the X axis.

  The angular velocity sensor 20 is for detecting an angular velocity that is a change amount per unit time of the rotation angle θy about the Y axis.

  As shown in FIG. 14, the two angular velocity sensors 19 and 20 are provided so as to be arranged in relation to the above-described X axis, Y axis, and Z axis in the electronic camera.

  The detection signals output from the two angular velocity sensors 19 and 20 are converted into digital data by the A / D converters 21 and 22, respectively, and input to the CPU 7.

  The CPU 7 acquires information on the focal length f from the photographic lens 15 (for example, when the photographic lens 15 is an electric zoom, the CPU 7 acquires information via the lens driving system 14, or the photographic lens 15. If is an interchangeable lens barrel, information is acquired via a communication contact or the like).

  In addition, the CPU 7 acquires subject distance information from the distance detection unit 23.

  The information on the focal length f and the subject distance information are used for calculating the blur amount in the X direction and calculating the blur amount in the Y direction, as will be described later.

  Next, the relationship between the shake amount and the shake correction amount will be described with reference to FIG. FIG. 15 is a diagram illustrating a moving state of the image of the subject 63 on the imaging surface when the camera body 51 is shaken by the rotation angle θx.

  Assuming that the electronic camera is rotated by the rotation angle θx due to shaking or the like, the photographing lens 15 rotates and moves to the position indicated by reference numeral 15 ′, and the imaging surface 61 of the imaging device 1 is also tilted by the angle θx. Rotate to face position.

  Further, when there is no blurring, the image of the subject 63 at the center position indicated by reference numeral 62 moves to the position indicated by reference numeral 62 'on the imaging plane CD after the blurring of the rotation angle θx occurs. .

と、を用いることにより、上記移動量ΔＹは、次の数式３に示すように算出される。
［数３］

ここに、βは撮影倍率を示し、ｆ／Ｌである。また、数式３を算出するに当たっては、θx が微小量であるとして、θx の１次のオーダーまでの近似を行っている。 Here, the focal length of the photographic lens 15 is f, the distance from the object space focal point of the photographic lens 15 to the subject 63 when no blur occurs, and the image space of the photographic lens 15 when no blur occurs. Assuming that the distance from the focal point to the image position is L ′, and the amount of movement of the image position due to the shake is ΔY, the geometric positional relationship as shown in FIG. Statue formula [Equation 2]

The movement amount ΔY is calculated as shown in the following Equation 3.
[Equation 3]

Here, β indicates the photographing magnification, and is f / L. Further, when calculating Equation 3, it is assumed that θx is a minute amount, and approximation to the first order of θx is performed.

  As described above, the value f in Equation 3 is input to the CPU 7 as lens information. Further, the distance L necessary for calculating β can be calculated based on the information from the distance detection unit 23 as shown in FIG. Furthermore, the angle θx in Equation 3 can be calculated based on the output from the angular velocity sensor 19 as described above.

  Thus, even if blurring occurs in the electronic camera by performing substantial correction related to the movement amount ΔY obtained based on Equation 3, the image related to the image signal output from the image sensor 1 is affected by blurring. Can be prevented from occurring.

  As described above, since the angle θx is a minute amount, even if the imaging surface CD is inclined about the X axis by the angle θx with respect to the Y axis as shown in FIG. The effect on the generated image is not a problem other than the movement amount ΔY described above.

Further, the amount of movement ΔX of the image position when the blurring occurs by the rotation angle θy around the Y axis is also obtained as shown in the following equation 4, similarly to the above equation 3.
[Equation 4]

By differentiating both sides of Equation 3 described above with respect to time, the following Equation 5 is obtained.
[Equation 5]

  In Expression 5, since d (θx) / dt on the right side is the angular velocity itself around the X axis, the output of the angular velocity sensor 19 can be used as it is. Further, d (ΔY) / dt on the left side of Equation 5 is the image movement speed Vy in the Y-axis direction when an angular velocity of d (θx) / dt occurs.

Similarly, with respect to the amount of movement ΔX of the image position in the X-axis direction when a shake occurs around the Y-axis by the rotation angle θy, the following formula 6 is obtained by differentiating both sides of formula 4 with respect to time.
[Equation 6]

  In Equation 6, since d (θy) / dt on the right side is the angular velocity itself around the Y axis, the output of the angular velocity sensor 20 can be used as it is. Further, d (ΔX) / dt on the left side of Equation 6 is the image moving speed Vx in the X-axis direction when an angular velocity of d (θy) / dt occurs.

Now, a predetermined time ΔT (this ΔT is a sampling interval at which the A / D converters 21 and 22 convert the outputs of the angular velocity sensors 19 and 20 into digital signals, and is equal to the blur limit exposure time TLimit, or It is desirable that the time is shorter than that.) If the output d (θx) / dt of the angular velocity sensor 19 detected in the period is ωx1, ωx2, ωx3,..., Ωx (n−1), ωxn, The movement amount ΔY of the image position in the Y-axis direction after the time of n × ΔT has elapsed is given by the following formula 7.
[Equation 7]

Similarly, assume that the output d (θy) / dt of the angular velocity sensor 20 detected every predetermined time ΔT (in a predetermined time ΔT cycle) is ωy1, ωy2, ωy3,..., Ωy (n−1), ωyn. , The moving amount ΔX of the image position in the X-axis direction after elapse of n × ΔT is given by the following equation 8.
[Equation 8]

  The blur amount of two images exposed by the image sensor 1 at a time interval of n × ΔT can be calculated by Equation 7 and Equation 8. Therefore, based on the movement amounts (blurring amounts) ΔX and ΔY calculated by these mathematical formulas, pixel charges are moved so as to correct the blurring of the two images, and then added to obtain an image without blurring. Can be generated.

  FIG. 16 is a flowchart showing the flow of processing for calculating the movement amounts ΔX, ΔY and the like by the CPU 7. This process is a process independent from the process shown in FIG. 12, the process shown in FIG. 13, or the process shown in FIG. 17 to be described later, from the time when the second release switch 18b is closed until the exposure is completed. It is supposed to be executed.

  That is, when this process is started, first, it waits for the second release switch 18b to close (that is, branches to J41 until the second release switch 18b is closed) (step S41).

  If it is detected that the second release switch 18b is closed, next, the focal length f of the taking lens 15 and the subject distance L are input (step S42). These focal length f and subject distance L may be calculated in the process as shown in FIG. 16, but a separate processor or the like is used in order to calculate the blur amount at a faster cycle. The focal length f and the subject distance L are used to calculate, and the CPU 7 may input the calculated data in step S42. As a result, it is possible to increase the processing speed and achieve high real-time followability.

  Next, the CPU 7 inputs the angular velocities ωx and ωy by reading the outputs of the angular velocity sensors 19 and 20 through the A / D converters 21 and 22 (step S43).

  Then, by adding the input angular velocities ωx and ωy to the cumulative addition value up to the previously detected value, the cumulative addition values Σωx and Σωy up to the currently detected value are calculated (step S44).

  By substituting the cumulative addition values Σωx and Σωy calculated in step S44 into the above-described Equations 7 and 8, the movement amount of the image position from the end of photographing of the first time-division image among the plurality of time-division images. ΔY and ΔX are respectively calculated (step S45).

  Next, it is determined whether or not readout from the image sensor 1 is set to field readout (step S46). If the readout is set to field readout, Px = “ΔX / Lx” and Py = “ ΔY / (2 · Ly) ”is calculated (step S47a). Lx and Ly represent the sizes in the X direction and Y direction of one pixel 31 of the imaging unit 41 shown in FIG. 3, respectively, and “” means an integer value rounded off to the nearest whole number. Therefore, Px and Py are the amounts of movement ΔX and ΔY of the image position from the end of photographing of the first time-division image, and the pixels of the correction / addition unit 40 (the electrode group 37 shown in FIG. This is expressed in units of 40 pixels.

  If it is determined in step S46 that the field readout is not performed, that is, if it is determined that all the pixels are read out, Px = “ΔX / Lx” and Py = “ΔY / Ly” are calculated. (Step S47b).

  When step S47a or step S47b is thus completed, the above Px and Py are stored in the corresponding memories [Px] and [Py], respectively (step S48). As described above, the symbol [] indicates a memory that stores data in parentheses.

  Thereafter, it is determined whether or not the exposure of the exposure time TExp has been completed (step S49). If the exposure has not been completed, the process branches to J42, and the same processing as described above is repeated from step S43.

  On the other hand, if it is determined in step S49 that the exposure has ended, this process ends.

  FIG. 17 is a flowchart showing a subroutine for pixel value synthesis.

  When this processing is started in step S9 of FIG. 12, first, the charge of the photodiode 32 is shifted to the vertical transfer unit 33 in the imaging unit 41, and further transferred to the divided image transfer unit 35 of the correction / addition unit 40 ( Step S51). Note that the photodiode 32 starts the next time-division exposure immediately after shifting the pixel charge to the vertical transfer unit 33.

  Next, a transfer amount (transfer amount converted into the number of pixels in the correction / addition unit 40) Sy (= (Py−Sy0)) is calculated for transferring the previously captured image in the vertical direction (Y direction). (Step S52). Here, Py is the amount of blurring in the Y direction immediately after the start of the first exposure, and Sy0 is the amount of blurring correction in the Y direction performed immediately after the start of the first time-division exposure until the previous time-division exposure ( Cumulative blur correction amount). Therefore, in this step S52, the difference between the amount of shake in the Y direction from immediately after the start of the first time-division exposure to the current time-division exposure and the amount of shake actually corrected before the previous time-division exposure is calculated. The obtained value is used as the correction amount related to the current time-division exposure. By using such means, it is possible to calculate the amount of blurring of the subsequent image on the basis of the amount of blurring of the previous image in the two consecutive exposure operations before and after. Accumulation of errors can be prevented. As described above, in the calculations in steps S47a and S47b, rounding is performed to the unit of the number of pixels. Therefore, it is effective to employ the processing as shown in step S52.

  Subsequently, similarly, the transfer amount for transferring the latest image taken this time in the horizontal direction (X direction) (as described above, the transfer amount converted into the number of pixels in the correction / addition unit 40) -Sx (= -Px ) Is calculated (step S53). Here, as described above, Px is the amount of shake in the X direction from the start of the first time-division exposure to the current time-division exposure. In addition, the minus sign is added to the expression indicating the number of transfer pixels, and the pixel charge is moved in the direction opposite to the blur so that a new readout pixel charge is brought closer to the pixel charge side added so far. Because.

  In the processing described here, the added image is not moved from the same Y direction transfer unit 34 in principle after being stored in the Y direction transfer unit 34 (X direction transfer unit). Therefore, in step S53 described above, cumulative blurring up to the previous time corresponding to Sy0 as described in step S52 is performed. There is no need to calculate a correction amount (for example, Sx0).

  Next, the latest image captured this time is transferred in the horizontal direction (X direction) by the number of pixels of −Sx via the X direction transfer unit 36 (step S54).

  Subsequently, the images captured up to the previous time (images obtained by correcting the relative blur and added) are transferred in the vertical direction (Y direction) by the number of pixels of Sy via the Y direction transfer unit 34 (step). S55).

  Then, the image stored in the X-direction transfer unit 36 and the image stored in the Y-direction transfer unit 34 are added as described with reference to FIG. 6 or FIG. 8 (step S56).

  Thereafter, the added image is transferred below the electrode S2φV2 below the paper surface closest to the electrode at the position where the X-direction transfer unit 36 and the Y-direction transfer unit 34 intersect (step S57). As described above, since the X-direction transfer electrode SφH0 and the Y-direction transfer electrode S2φV0 are combined, when the image taken by the next time-division shooting is transferred to the X-direction transfer unit 36, This is to prevent interference with the added image.

  Next, the transfer amount Sy and the previous transfer amount Sy0 are added and stored in the memory [Sy0], so that the added value is newly set as Sy0 (step S58). As a result, the memory [Sy0] stores the cumulative addition value of the transfer amount in the Y direction until the last (previous) time-division exposure.

  When the process of step S58 is performed, the process returns from the pixel value synthesis subroutine to the process shown in FIG.

  Note that the processing in steps S54 and S55 described above is processing for correcting an integer multiple of one pixel and adding two images. However, in the case of a single-plate color image pickup device, that is, an image pickup device in which a mosaic color filter is arranged in front of the image pickup surface, it must be further devised so that color signal mixing does not occur. For example, in a Bayer color filter, a mosaic filter having a 2 × 2 color filter as a unit is repeatedly arranged. Therefore, when a blur of odd pixels is corrected and added, signals of different colors are added. As a result, normal colors cannot be reproduced. Therefore, in order to solve such a problem, blurring is corrected with the minimum repeat period in the horizontal direction and the minimum repeat period in the vertical direction (two pixels in the case of Bayer array) as the minimum unit. Just do it. For example, when all pixels are read out from an image sensor having a Bayer color filter, the vertical blur correction amount is 2 · “Sy / 2” and the horizontal blur correction amount is 2 · “ -Sx / 2 ". Here, “” means an integer value obtained by rounding off fractions as in the above, and Sx and Sy are values obtained in steps S52 and S53 described above.

  Returning to the description of FIG. 12, when the pixel value composition process is completed in step S9, n + 1 is stored in a memory [n] for storing the number n of time-division photographing that has already been performed (step S10).

  Then, it is determined whether or not the number n of time-division photographing that has already been performed is equal to the set number m of time-division photographing (step S11).

  If n = m has not yet been reached, the process branches to J3, and the time-division exposure process and the pixel value composition process are repeated as described above.

  If it is determined in step S11 that n = m, the image signal in the divided image transfer unit 35 (in the case of normal exposure that is not time division exposure) or the Y direction transfer unit 34 (in the case of time division exposure). Is read out to the outside of the image sensor 1 via the horizontal transfer unit 42 and the amplifier 43 (FIG. 2) (step S12).

  Next, the image signal read from the image sensor 1 is processed by the CDS 2, amplified by the gain control amplifier (AMP) 3, and then converted into a digital signal by the A / D converter 4 (step S13).

  Subsequently, predetermined signal processing is performed by the information processing unit 8 on the image data obtained by digitizing the image signal (step S14).

  Further, the image-processed image data is temporarily stored in the DRAM 9 (step S15).

  Thereafter, the image data stored in the DRAM 9 is compressed by the compression / decompression unit 10 (step S16), and the compressed image data is recorded on the recording medium 11 (step S17).

  At this time, Sx, Sy0 as described above (or Sx, Sy, etc. at the time-division photographing) may be stored in the recording medium 11 as the auxiliary information of the image data. Thereby, after the image data is read from the recording medium 11, it is possible to remove a portion where the time-division image is not normally added and display or print it. Specifically, for example, when the image blur is ΔX in the X positive direction and ΔY in the Y negative direction, the range of ΔX at the end in the X negative direction of the image data and the ΔY at the end in the Y positive direction Since the time-division image is not normally added to the range, the processing for removing this portion can be performed by, for example, a separate personal computer. Such removal processing of the abnormal portion is not limited to being performed after reading from the recording medium 11, and in the signal processing in step S14, the blur correction amounts Sx, Sy0 (or each time division by the information processing unit 8). Needless to say, unnecessary image portions may be removed based on Sx and Sy at the time of photographing.

  In the above description, as described in step S2 of FIG. 12, the blur limit exposure time TLimit is set to a value uniquely determined from the focal length f of the photographic lens as shown in Equation 1. However, the value shown in Formula 1 is an average value obtained by empirical rules. On the other hand, the amount of blur actually generated per unit time varies depending on individual differences, the size of the camera, the shape of the camera, and the like. An example (modification) of a technique that can cope with such variations will be described below.

  In this modification, the blur limit exposure time TLimit is actually measured before the actual blur is corrected, and more accurate blur correction considering the above-described variation is performed.

  In other words, in this modification, instead of the process in step S2 in FIG. 12 described above, a process for obtaining the blur limit exposure time TLimit by actual measurement is performed. Accordingly, the entire process is the same as the process of FIG. 12 except for step S2, and the description thereof is omitted.

  Here, the blur limit exposure time TLimit is the time TLimit (x) until the blur amount in the X direction reaches a predetermined allowable blur amount (allowable limit blur amount) αx, and the blur amount in the Y direction is a predetermined allowable value. It is set as the shorter time of the time TLimit (y) until the limit shake amount αy is reached.

  Here, the allowable limit blur amounts αx and αy will be described.

であれば、ぶれは印画紙を観察している観察者の眼には知覚されない。 Consider a case where an image is printed on photographic paper and observed. Assume that the distance from the observer's eyes to the photographic paper is K, the diagonal length of the photographic paper is Dp, and the amount of image blur is the distance ΔDp on the diagonal line of the photographic paper. If ΔDp / K [radian] is less than the two-point discrimination resolution ΔΦ of the observer's naked eye,
[Equation 9]

If so, the blur is not perceived by the eyes of an observer observing the photographic paper.

By the way, assuming that Di is the diagonal length of the image sensor (effective imaging area) and the amount of blur on the imaging surface of the image sensor (length corresponding to the above ΔDp) is ΔDi, the blur on the photographic paper with respect to the size of the photographic paper. Since the magnitude of the amount and the magnitude of the blur on the image sensor with respect to the size of the image sensor coincide with each other, the following Expression 10 is established.
[Equation 10]

Therefore, when ΔDp is eliminated from Equation 9 using the relationship shown in Equation 10, the following Equation 11 is obtained.
[Equation 11]

  As is apparent from Equation 11, ΔDi, which is an allowable blur amount on the image sensor, is a value that is uniquely determined by the size (Di) of the image sensor and the image observation conditions (Dp, K). Exactly the same as the example of the allowable blur amount in the diagonal direction, the allowable blur amount ΔX i of the image in the X direction and the allowable blur amount ΔY i of the image in the Y direction on the imaging surface of the image sensor can be obtained. it can.

  Now, it is assumed that the allowable blur amount ΔXi in the X direction is equal to the allowable blur amount ΔYi in the Y direction, and ΔXi = ΔYi = γ.

  In general, the two-point identification resolution ΔΦ for the naked eye is approximately 1 minute, but this value may vary depending on the type of image, the mode of blurring, and individual differences. ΔΦ may be selected appropriately.

  Next, with reference to FIGS. 18 and 19, a modified example of measuring the blur limit exposure time TLimit will be described. First, FIG. 18 is a flowchart showing a blur amount calculation process.

  When this process is started, first, it waits for the first release switch 18a to close (step S61).

  If it is detected that the first release switch 18a is closed, next, the focal length f of the taking lens 15 and the subject distance L are input (step S62). These focal length f and subject distance L may be calculated in the process as shown in FIG. 18, but in the same way as described above, The focal length f and the subject distance L are calculated using a separate processor or the like, and the CPU 7 may input the calculated data in step S42. As a result, it is possible to increase the processing speed and achieve high real-time followability.

  The subsequent steps S63 to S65 are the same as the steps S43 to S45 shown in FIG.

  Thereafter, ΔX calculated in step S65 is stored in the memory [ΔX], and ΔY is stored in the memory [ΔY] (step S66).

  It is then determined whether FLG1 is 1 (step S67). This FLG1 is assumed to be initialized to 0 in advance when the entire processing as shown in FIG. 12 is started. Then, as will be described later with reference to FIG. 19, it is set to 1 when the measurement of the blur limit exposure time TLimit ends.

  If it is determined in this step S67 that FLG1 is not 1 (0), the process branches to J61, and the above-described blur amount calculation and cumulative addition value calculation are repeated.

  On the other hand, if it is determined in step S67 that FLG1 is 1, this process ends.

  FIG. 19 is a flowchart showing a subroutine for measuring TLimit. The subroutine shown in FIG. 19 is executed in place of the process of calculating TLimit in step S2 of FIG.

  When this process is started, first, a timer counter (not shown) built in the CPU 7 is reset (step S71).

  Next, ΔX and ΔY calculated by the processing shown in FIG. 18 are read from the corresponding memories [ΔX] and [ΔY], and the read ΔX and ΔY values are read from the memories [ΔX (ini)] and [ [Delta Y (ini)] are newly stored (step S72). These ΔX (ini) and ΔY (ini) are initial values of ΔX and ΔY.

  Subsequently, the absolute value of the difference between ΔX (ini) and ΔX newly read from the memory is obtained and stored in the memory [ΔΔX]. Similarly, the absolute value of the difference between ΔY (ini) and ΔY newly read from the memory is obtained and stored in the memory [ΔΔY] (step S73).

  Then, it is determined whether or not ΔΔX <γ and ΔΔY <γ (step S74). Here, γ is an allowable shake amount γ when the X direction and the Y direction are equal, as described above with reference to Equations 9 to 11 and the like.

  If ΔΔX <γ and ΔΔY <γ, the process branches to J71, and the above-described processing in step S73 is repeated. By performing such processing, it is possible to obtain the shake amount ΔΔX in the X direction and the shake amount ΔΔY in the Y direction immediately after the timer counter is reset in step S71 described above.

  In step S74, if at least one of ΔΔX and ΔΔY is not less than γ (at least one is equal to or greater than the allowable blur amount γ), information on the timer counter is stored in the memory [TLimit] (step S75). . Therefore, this TLimit is the time until the blur amount reaches the allowable limit blur amount γ, that is, the blur limit exposure time.

  Thereafter, 1 is stored in the memory [FLG1] (step S76), and the process returns to the process shown in FIG.

  The charge under the row and column transfer electrodes located at the end (both ends) in the region of the correction / addition unit 40 that performs shake correction of the image sensor 1 further transfers the charge in the end face direction. Therefore, the charge transferred for blur correction stays under this electrode, and the image information is not stored correctly. Accordingly, the image signal of the transfer register at the end where the image information including the transfer register is not correct is invalidated after being read and is not displayed or recorded.

  Further, in the above description, the pixel arrangement and the transfer operation are described using the relative relationship between “horizontal” and “vertical” or “X” and “Y”, but if the image sensor is rotated 90 degrees, “ “Horizontal” and “vertical” have a relationship of “vertical” and “horizontal”, and the relationship between “X” and “Y” becomes a relationship of “Y” and “X”. Even if “vertical”, “X”, and “Y” are interchanged and read, the same is true. Therefore, for example, if one of the X-direction transfer unit 36 and the Y-direction transfer unit 34 is a first transfer register, the other is a second transfer register.

  In the above description, the example in which the solid-state image sensor is used for blur correction has been described. However, the use of the solid-state image sensor having the above-described configuration is not limited to this. It is also possible to combine two images with different exposure levels taken continuously to obtain a wide dynamic range image, and the present invention is not limited to these examples. It can be used for applications.

  According to the first embodiment, it is possible to perform highly accurate blur correction corresponding to a wide range of shutter speeds without requiring a complicated mechanical mechanism or the like for correcting blur.

  At this time, since the two images captured continuously in time are stored in the second transfer register and the first transfer register, and the relative positions are adjusted, and then added, It is possible to perform blur correction at an extremely high speed.

  In addition, the first transfer register and the second transfer register are each configured to have four charge transfer electrodes per pixel, and charge is transferred by applying four different drive pulses to each electrode. As a result, the charge stored in the first transfer register and the charge stored in the second transfer register can be transferred independently.

  In addition, since the electrode having a small area is simply used as a gate electrode for transferring charges, and the charge is accumulated and held in the potential below the electrode having a large area, accumulation in the first transfer register and the second transfer register is performed. The capacity can be increased.

  Furthermore, because the image transfer control in the register unit when images are added is different from the image transfer control in the register unit when images are not added, the relative position correction between images is made extremely fast. This makes it possible to perform image transfer control in the register unit. In addition, the same function as a conventional solid-state image sensor can be achieved, and versatility is enhanced.

  The above-described technique is a high-speed blur correction technique that can cope with a shake in which the position of the subject light incident on the imaging surface changes during the exposure period, and is particularly suitable for blur correction in still image shooting. It becomes.

  In the present embodiment, a frame interline transfer type imaging device is employed, and the imaging unit and the correction / addition unit for accumulating images and correcting blurring are separately provided on the imaging device. In addition, the configuration of a conventional imaging device can be applied as it is as the imaging unit. Therefore, the advantages of the frame interline transfer type CCD image sensor and the manufacturing know-how of the solid-state imaging device accumulated for many years can be used as they are. Moreover, the versatility of the solid-state imaging device can be enhanced.

  In addition, in any of the shutter priority shooting mode, aperture priority shooting mode, and program shooting mode, an image without blurring (less blurring) can be obtained while giving the highest priority to the intention of the photographer who has selected these modes. Can do.

  It should be noted that the present invention is not limited to the above-described embodiments, and various modifications and applications can be made without departing from the spirit of the invention.

  The present invention can be suitably used for an electronic blur correction apparatus that electronically corrects image blur.

1 is a block diagram mainly showing an electrical configuration of an electronic camera according to Embodiment 1 of the present invention. FIG. 3 is a diagram illustrating an outline of a configuration of the entire solid-state imaging element in the first embodiment. FIG. 3 is an enlarged view showing configurations of an imaging unit and a correction / addition unit of the solid-state imaging device in the first embodiment. In the said Embodiment 1, the timing chart which shows the basic operation | movement of the solid-state image sensor at the time of imaging. 4 is a timing chart illustrating an operation of transferring an image captured by the imaging unit of the solid-state imaging device according to the first embodiment to a correction / addition unit. 4 is a timing chart illustrating a shake correction operation and an addition operation performed in the correction / addition unit of the first embodiment. The figure which expands and shows the structure of the imaging part and correction | amendment / addition part of the image pick-up element in the modification of the said Embodiment 1. FIG. 4 is a timing chart illustrating a shake correction operation and an addition operation performed in the correction / addition unit of the first embodiment. Sectional drawing for demonstrating that an image sensor is a vertical overflow structure in the said Embodiment 1. FIG. FIG. 3 is a diagram illustrating a state in which the amount of accumulated charge that causes overflow is controlled by changing the reverse bias voltage according to the number of time-division exposures in the first embodiment. In the said Embodiment 1, the diagram which shows some examples of a mode that the maximum value Q of the accumulation charge amount of the photodiode which changes with the board | substrate voltage VSUB is set according to the frequency | count m of time division imaging. 5 is a flowchart illustrating processing when an image is captured and recorded by the electronic camera in the first embodiment. 7 is a flowchart illustrating a subroutine for calculating and setting TExp and an aperture value in the first embodiment. In the said Embodiment 1, the figure which shows the coordinate axis set to the electronic camera, and arrangement | positioning of two angular velocity sensors. FIG. 3 is a diagram illustrating a moving state of an object image on an imaging surface when the camera body is shaken by a rotation angle θx in the first embodiment. 6 is a flowchart showing a flow of processing for calculating movement amounts ΔX, ΔY and the like by the CPU in the first embodiment. 5 is a flowchart showing a pixel value synthesis subroutine in the first embodiment. 5 is a flowchart showing a shake amount calculation process in the first embodiment. 5 is a flowchart showing a subroutine for measuring TLimit in the first embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Solid-state image sensor 2 ... Correlated double sampling circuit (CDS)
3. Gain control amplifier (AMP)
4 ... A / D converter 5 ... Timing generator (TG)
6 ... Signal generator (SG)
7 ... CPU (control unit, shake detection unit)
8 ... Information processing unit 9 ... DRAM
DESCRIPTION OF SYMBOLS 10 ... Compression / decompression part 11 ... Recording medium 12 ... Liquid crystal display part 13 ... Interface part 14 ... Lens drive system 15 ... Shooting lens 16 ... Aperture drive system 17 ... Aperture 18a ... 1st release switch 18b ... 2nd release switch 19,20 ... Angular velocity sensor (blur detector)
21, 22 ... A / D converter 23 ... Distance detector 24 ... EEPROM
25 ... Shooting mode setting unit 26 ... Shooting condition setting unit 31 ... 1 pixel 32 ... Photodiode (PD)
33. Vertical transfer unit (first transfer unit)
34 ... Y-direction transfer unit (second transfer register)
35 ... Divided image transfer unit 36 ... X direction transfer unit (first transfer register)
37 ... Electrode group 40 ... Correction / addition unit (register unit)
41 ... Imaging unit 42 ... Horizontal transfer unit (second transfer unit)
43 ... Amplifier 51 ... Camera body 61 ... Imaging surface when no blurring occurs 63 ... Subject

Claims (6)

An electronic shake correction device that generates a shake-corrected image by performing analog addition while sequentially correcting the shake of a plurality of images taken continuously in time inside a solid-state imaging device,
An imaging unit which is provided inside the solid-state imaging device and has a plurality of pixels arranged in a matrix, and generates an image signal by the plurality of pixels;
A first transfer register provided in the solid-state imaging device separately from the imaging unit, storing a first image captured by the imaging unit and transferring the first image in a first direction; A second transfer register for storing a second image taken by the imaging unit at a time different from the first image and transferring the second image in a second direction intersecting the first direction; A register unit having
A shake detection unit that detects a relative shake between the first image and the second image;
Relatively transferring the first image and the second image in the first transfer register and the second transfer register, respectively, to correct a relative blur between the first image and the second image; Thereafter, a control unit that controls the first image and the second image to be analog-added,
An electronic shake correction apparatus comprising:   2. The electronic blur according to claim 1, wherein the control unit controls an image obtained by analog addition of the first image and the second image to be stored in the second transfer register. 3. Correction device.   The shake detection unit detects a shake in the first direction and a shake in the second direction relative to the first image and the second image. The control unit One image is controlled to be transferred in the first transfer register based on the blur in the first direction, and the second image is controlled in the second transfer register based on the blur in the second direction. 2. The electronic shake correction apparatus according to claim 1, wherein the electronic shake correction apparatus is controlled so as to be transferred.   The electronic blur correction device according to claim 1, wherein the solid-state imaging element is configured to transfer an image captured by the imaging unit to the register unit.   Each of the first transfer register and the second transfer register has four electrodes per pixel, which is a minimum unit constituting an image, and charges are transferred by applying different drive pulses to the four electrodes. The electronic shake correction apparatus according to claim 1, wherein the electronic shake correction apparatus is configured to perform the following.   The first transfer register and the second transfer register are each configured by alternately arranging electrodes having a relatively large area and electrodes having a relatively small area. The electronic blur correction apparatus according to claim 5.

Images