JPH04349440A - Blurring detecting device for camera - Google Patents

Abstract

PURPOSE:To offer a blurring detecting device for camera in which a photographer can freely select what positioned image pickup element is used to detect blurring in the camera which executes multipoint range-finding and detects the blurring by using the image pickup element for detecting focus. CONSTITUTION:An AF sensor 11 is composed of plural image pickup elements to perform the multipoint range-finding. A range-finding part 12 outputs image pickup data from the sensor 11 to a microcomputer 10. The photographer optionally selects what image pickup element is used among the plural image pickup elements installed in the AF sensor 11 to detect the blurring by a manual island selection part 21. Based on the image pickup data of the image pickup element selected by the photographer, the microcomputer 10 executes blurring operation. With such constitution, the photographer freely selects what positioned image pickup element is used to detect the blurring.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a camera shake detection device, and more particularly to a camera shake detection device that utilizes an image sensor included in a camera focus detection device.

0002

2. Description of the Related Art Conventionally, cameras have been proposed that detect camera shake using an automatic focus detection device (hereinafter referred to as an AF device), as disclosed in Japanese Patent Application Laid-open No. 166910/1983. These cameras perform blur detection using an image sensor included in the AF device, and determine the relative movement amount of a subject from image data at a certain time and image data after a predetermined period of time.

On the other hand, in recent years, cameras capable of multi-point distance measurement have been proposed, which can measure object distances at a plurality of points within a photographic screen. This uses a plurality of AF devices to measure the distance, determines the position of the main subject on the photographic screen based on the measured distance data, and then focuses on the main subject. Furthermore, some cameras allow the photographer to freely select the position at which focus detection is to be performed depending on the position of the main subject. This multi-point distance measurement makes it possible to
Unlike the F camera, it is no longer necessary to center the main subject, allowing for free composition.

Now, when installing the above-mentioned blur detection device on such a camera capable of multi-point distance measurement, it is necessary to use a plurality of A.
The problem is which of the F devices should use imaging data to detect blur. On the other hand, with conventional cameras that are capable of multi-point distance measurement and use a focus detection image sensor to detect blur, the camera shake is detected based on image data from an AF device that performs distance measurement in the center of the shooting screen. Ta. However, with this camera, if the main subject is located off-center, background blur will be detected. Therefore, in order to detect blur, it is necessary to place the main subject in the center, and as mentioned above, the ability to freely compose a picture through multi-point distance measurement cannot be utilized at all. In addition, with a camera that allows the photographer to freely select the focus detection position, even if the main subject is set to a position off the center and the camera is set to focus on that position, the blur detection will still detect blur in the background. There was a problem with it being put away.

[0005]

SUMMARY OF THE INVENTION It is therefore an object of the present invention to eliminate the drawbacks of the above-mentioned conventional examples, and to provide a camera that is capable of multi-point distance measurement and that performs blur detection using a focus detection image sensor. An object of the present invention is to provide a camera in which it is possible to freely select which position of an image sensor to use for blur detection.

[0006]

[Means for Solving the Problems] In order to achieve the above object, the blur detection device of the present invention includes a plurality of focus detection image sensors, and uses any one of the image sensors to detect the relative relationship between the camera and the subject. In a camera that calculates blur, the photographer manually selects which imaging device to use for blur detection, and calculates blur data based on the imaging data of the selected device. .

[0007]

[Operation] The blur detection device of the present invention allows a photographer to select which of the plurality of image sensors should be used to detect blur in a camera that is capable of multi-point distance measurement and uses a focus detection image sensor to detect blur. You can freely choose what to do. Therefore, it becomes possible for the photographer to freely set the desired composition.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Examples of the present invention will be described below with reference to the drawings. First, a focus detection device used in the present invention will be explained. Figure 1 shows the so-called TTL (Through T
This figure shows the principle of distance measurement of a camera using a phase difference detection method (he Lens). In the figure, incident light that has passed through a photographic lens 1 forms an image on a film equivalent surface 2, passes through a condenser lens 3, and is split into two light beams by an aperture mask 4. The divided light flux is imaged by the separator lens 5 into two areas, a standard area and a reference area, set on the image sensor 6. Then, the amount of defocus is detected based on the interval between the two images formed on the image sensor 6. That is, the image sensor 6
The focus state, that is, the front focus or rear focus state, is determined based on the magnitude of the image interval appearing in the reference area and the reference area with respect to a predetermined interval.

FIG. 2(a) is a diagram showing the distance measurement range in the field finder of a camera capable of multi-point distance measurement to which the present invention is applied. This embodiment has four distance measurement zones, and uses four focus detection devices as shown in FIG. 1 to obtain subject distance data at points corresponding to each distance measurement zone. The camera of the present invention calculates the photographing magnification from these four distance measurement data and the focal length of the photographic lens, and then automatically determines in which zone the main subject is located, taking into account the size of the subject, etc. Then, the photographic lens is driven to focus on that zone.

FIG. 2(b) shows a CCD line sensor group corresponding to each distance measurement zone, which functions as the image pickup device 6 of FIG. However, this figure shows Figure 1.
This is a group of CCD line sensors in the reference area of the image sensor 6. Furthermore, an illuminance monitor (not shown) is provided near this sensor to control the integration time of CCD integration. As shown in the figure, the four CCD line sensor groups are respectively designated as the first island to the fourth island, and the arrangement direction of the line sensors on the first and third islands and the arrangement direction of the line sensors on the second and fourth islands are mutually different. Orthogonal. In this example, the first and third islands have 40 pixels,
The second island is composed of a line sensor of 88 pixels, and the fourth island is composed of a line sensor of 60 pixels. During automatic focus detection, as described above, it is determined which island captures the main subject based on distance measurement data from the sensors of these four islands and the size of the subject. Thereafter, the lens driving amount is determined based on the distance measurement data of the island determined to capture the main subject (hereinafter referred to as the AF selection island). By driving the photographing lens using the lens drive amount calculated in this way, the main subject is brought into focus. In addition, although the distance measurement points are set to four in this example, it is not necessary that there are four points.
Any number may be used as long as multi-point distance measurement is possible. The same applies to the number of pixels constituting each island, and the number of pixels is not particularly limited to the number described above.

[0011] In the present invention, blur detection is performed using a CCD line sensor used for automatic focus detection. Note that horizontal shake is detected using the sensor on the second or fourth island, and vertical shake is detected using the sensor on the first or third island.

FIG. 3 is a block diagram of a camera to which the present invention is applied. In FIG. 3, the microcomputer 10 (
The microcomputer (hereinafter referred to as a microcomputer) controls a series of shooting sequences such as lens drive and exposure, and also performs calculations related to shooting such as exposure calculations, distance measurement calculations, and blur calculations. The AF sensor 11 is a group of four CCD line sensors shown in FIG. 2, and the distance measuring section 12 outputs distance measurement data to the microcomputer 10 based on the detection results of this sensor. Furthermore, since the present invention performs blur detection using image data of the AF sensor, the distance measuring section 12 also outputs image data for blur detection. The photometry unit 14 is a photoelectric conversion element SPD13 (Silicon Ph
The luminance of the subject is measured based on the output of the diodes, and the photometric data is output to the microcomputer 10.

The lens control section 16 controls general lens operations such as driving the photographing lens 15, and also controls the photographing lens 15.
Lens data such as focal length and aperture value outputted from the microcomputer 5 (more specifically, from the lens ROM) are outputted to the microcomputer 10. The shutter curtain 17 of the focal plane shutter is
It includes a leading curtain and a trailing curtain, and is driven by the shutter control section 18. The blur data output unit 19 is an output terminal for outputting blur data calculated by the microcomputer 10. The output blur data is used to display the amount of blur to show how much the subject is blurred, and to correct blur using an optical system or the like. Mode switching section 20
This mode changes modes, such as portrait mode and landscape mode, in order to perform control more suitable for the subject and shooting environment. This may be switched by a dial type switch or by setting an external storage medium such as an IC card, and the format is not particularly limited. The manual island selection unit 21 is an input device that allows the photographer to select an island to be used for blur detection. The manual island selection section 21 not only allows the microcomputer to automatically select an island, but also allows the photographer to select which island to use for blur detection.

[0014] S0 is a power switch for the camera, and when turned on, operation of the camera is permitted. S1 is a switch that is turned on by pressing the first stroke of a release button (not shown) and starts photometry and distance measurement. S2 is a switch that is turned on by pressing the second stroke (deeper than the first stroke) of the release button and starts the release operation.

FIG. 4 is a diagram showing the blur detection sequence of the blur detection device of the present invention. The blur detection of the present invention is based on AF.
Since the camera uses a CCD that is used in the AF system, blur detection is performed between AF operations as shown in the figure. To explain this diagram in more detail, first, the release button of the camera is pressed halfway, and when S1 is turned on, distance measurement is started. When the distance measurement data for all the islands is obtained through this distance measurement operation, the camera shake detection is performed first before starting the focusing operation.
Let's go around. In this case, blur detection is performed using four cycles as one set. Note that this cycle will be described later. Next, AF operation is started. This AF operation represents the operation from distance measurement to calculation of the lens drive amount to driving the photographing lens. After this, the AF operation of "distance measurement, lens drive amount calculation, and lens drive" is repeated, and the camera waits for 0.5 seconds or more to elapse from the start of the first blur detection. FIG. 4 shows an example in which the AF operation is repeated twice in 0.5 seconds. The number of times this AF operation is performed depends on the brightness of the subject; when the brightness of the subject is low, the integration time of the CCD becomes longer and the number of times decreases.

When 0.5 seconds or more have elapsed, the shake detection operation is started again, and this time, shake detection is performed with one set of 8 cycles. Thereafter, the AF operation and the blur detection, each set of 8 cycles, are performed in sequence. However, as mentioned above, there is a gap of at least 0.5 seconds between one shake detection and the next shake detection.
During this time, the AF operation is repeated. Note that this 0.
The reason for waiting 5 seconds is that if the blur detection cycle is made faster, it will interfere with AF and other operations. In other words, if the cycle of blur detection is made faster and the number of times is increased, more time will be spent on blur detection, and other operations will be delayed accordingly. This sequence is activated when the release button is further pressed (S
2 on) and continues until the release operation begins.

FIG. 5 is a flowchart showing the sequence of camera operations from S1 on to release. When S1 is turned on in #2, distance measurement and photometry are performed in #4. #
6, the subject image used for distance measurement is low contrast (
(hereinafter referred to as low contrast). If the contrast of the subject image is low, distance measurement is impossible, so a low contrast scan is performed. This low-contrast scan is a method in which a lack of contrast in a subject image is often due to a large out-of-focus condition, and the contrast is found by driving the photographic lens for focusing. When the contrast is found, the lens is stopped and the distance is measured again. If the focus is not low in #6, it is determined in #8 whether the focus is on. If it is in focus, it enters static motion determination, and if it is not in focus, it proceeds to #10. Note that the static motion determination will be described later. S2 is determined in #10, and if it is off, four cycles of blur detection are performed in #12. After detecting blur, the lens is driven in #14. If S2 is on, the lens is driven without performing blur detection. In #16, distance measurement and photometry are performed again, and it is similarly determined whether or not the contrast is low. If the contrast is low, measure the distance again. Note that low contrast scanning is not performed here. #20
In-focus is determined, and if it is in focus, the process proceeds to static/motion determination, and if it is not in focus, the lens is driven.

Next, referring to FIG. 6, the sequence of operations after the static motion determination will be explained. If S2 is on in #30, the release operation immediately begins. If S2 is off, it is determined in #32 whether the subject is a still object or a moving object. A moving object in this case means an object moving in the optical axis direction of the camera.

If the subject is a still body, no more focusing is performed (AF lock). Distance measurement with #36
Perform photometry and perform low contrast judgment in #38. #40
S2 is determined, and if it is on, the release operation begins. If it is off, it is determined in #42 whether 0.5 seconds or more has elapsed since the previous shake detection. If 0.5 seconds or more has elapsed, 8 cycles of blur detection are started in #44, and if not, distance measurement and photometry are performed again (#36). From now on S2
Repeat this operation until it is turned on.

[0020] When the subject is a moving object, focusing continues as the subject moves (continuous).
. Distance measurement and photometry are performed in #48, and low contrast determination is performed in #50. It is determined in #52 whether or not the image is in focus, and if it is in focus, S2 is determined in #54. If it is not in focus, drive the lens and measure distance and light again. If S2 is on in #54, a release operation is started, and if it is off, lens driving is performed. This lens drive #58 is for correcting the fact that even if the lens is once in focus, the focus may shift over time due to movement of the subject. In #60, it is determined whether the lens driving for the correction has ended or not, and if the lens is still being driven, distance measurement and photometry are started again. If the lens has stopped, it is determined in #62 whether 0.5 seconds or more has elapsed since the previous shake detection. If 0.5 seconds or more has elapsed, 8 cycles of blur detection begin in #64, and if not, the process returns to #48. From now on, this operation is repeated until S2 is turned on. When S2 is turned on, a release operation begins, but since no blur detection is performed after this point, the explanation will be omitted.

Next, the actual blur detection operation will be explained. The blur detection according to the present invention is roughly divided into four operations. The four actions are "island selection",
These are "block selection", "blur calculation", and "average processing". These four operations will be explained in order below.

[1] Island Selection Island selection means selecting an island suitable for use in blur detection from among the four islands described above. In the present invention, in order to detect the amount of blur in the horizontal and vertical directions, two of the four islands that are perpendicular to each other are selected. One selection method is to select an AF selection island, and the other is to select an island that is perpendicular to the AF selection island and has a difference in defocus amount (hereinafter referred to as DF) from the AF selection island within 100 μm. This is because the purpose of the present invention is to detect the relative blur of the main subject. To achieve this, first select the AF selection island that is determined to capture the main subject, and then This is to select an island that is perpendicular to the island. The island that captures the first selected main subject is called the main island, and the second island that is orthogonal to it is called the secondary island. FIGS. 7A to 7D are diagrams showing two islands selected by this selection method.

Note that there is no island that satisfies the above condition among the islands perpendicular to the AF selection island (D
If the difference in F is greater than 100 μm, DF cannot be detected due to low brightness, etc.), the AF selection island is the main island,
No secondary islands are selected. Then, blurring is detected only on the main island. This is shown in FIGS. 7(e) and 7(f). In (e), the fourth island is the AF selection island, and the first and third islands perpendicular to it both capture the background, so they do not satisfy the above condition. Therefore, in this case, blur detection is performed only on the fourth island. In (f), the third island is the AF selection island, and in this case as well, neither the second nor fourth islands that are perpendicular to the third island satisfy the above conditions, and the blur detection is performed by the third island.
Perform only on islands.

The above is the island selection method according to the present invention. However, the method just described is a basic selection method, with the following exceptions.

[0025] When a photographer uses a focus aid (hereinafter referred to as FA) to perform manual photography, the second island is set as the main island, and no sub-island is selected. Then, blur detection is performed only on the main island (second island). To explain this selection method, in the case of FA photography, since the lens is extended manually, data on the lens position cannot be obtained and distance data cannot be calculated. This makes it impossible to determine where the main subject is located within the photographic screen. Therefore, blur detection is performed using the central island where the probability of the main subject being present is highest, that is, the second island.

In addition, multi-point distance measurement is possible, and the photographer can arbitrarily select an island to focus on (local A).
F function) In the camera, the island selected by the photographer is the main island, and this island is used to perform blur detection. This is because it is thought that the main subject usually exists within the island selected by the photographer. Note that when using this local AF function, the photographer selects an island using the manual island selection section 21 shown in FIG.

FIG. 8 is a flowchart showing the operation sequence for island selection as described above. The operation order of island selection will be explained with reference to this figure. However, since this flowchart shows general island selection, it is assumed that there are l horizontal islands from H1 to H1 and m vertical islands from V1 to Vm. Note that this index (1~l, 1~
m) are numbered in order of distance from the main island when the main island is selected.

In #100, the AF selection island is set to the main island. In #102, it is determined whether the set main island is a vertical island or a horizontal island. If the main island is vertical, select #10 to select the secondary island from the horizontal islands.
After setting k to 1 in Step 4, it is determined in Step #106 whether the defocus amount of the island Hk has already been detected. In the case of general photography, since the DFs of all islands are detected, the answer is YES and the process proceeds to #108. This determination is NO only in special cases, such as when only one island is used in the FA mode or local AF mode, as described above. If YES in #106, it is determined in #108 whether the Hk DF is reliable. If the subject is of low brightness, the answer is NO as there is no reliability. Main island DF (DF0) and Hk at #110
It is determined whether the difference between DF (DFHk) is less than or equal to a predetermined value. In this embodiment, this predetermined value is 100 μm. DF
If the difference is less than a predetermined value, the island H is
Set k as the secondary island and complete island selection. If the answer is NO in any of steps #106 to #110, k is incremented by 1 in #112, and the same determination is made for the next island via #114. All l horizontal islands are #106 to #110
If the above conditions are not satisfied, the answer is YES in #114, and the sub-island is determined to be absent in #130.

#1 if the main island is horizontal
A sub-island is selected from among the vertical islands by the determinations from #16 to #126. This operation is exactly the same as when the main island is in the vertical direction, so the explanation will be omitted. As described above, one island is selected from each of the vertical and horizontal islands to be used for blur detection.

What has been explained above is island selection.

[2] Block selection (including selection of standard part and reference part) Block selection is to select a plurality of pixels (in this example, 16 pixels) as a detection block. A block selected on a main island is called a main block, and a block selected on a sub island is called a sub block. This selected block is used as a reference portion, and its pixel data is used for blur calculation.

This block selection will be explained with reference to FIG. Note that after this block selection, exactly the same operations are performed in both the horizontal and vertical directions, so from this point on, horizontal blur detection will be explained. In Figure 9, for simplicity, the total number of pixels of the selected island is 21.
It is assumed that pixels (No. 1 to No. 21) are selected, and five pixels among them are selected as a detection block. In this example, there are three blocks, first to third, as shown in the figure. Even if both ends of the island are selected as reference parts, they may quickly leave the area as the subject moves, making blur detection impossible, so they should not be selected. In this embodiment, the block with the most recent defocus is selected from among these three blocks. However, if the contrast of that block is less than the predetermined value, a new first
~The block with the highest contrast among the third blocks shall be selected.

Now, the pixel data when the first CCD integration is completed after S1 is turned on are assumed to be a1 to a21, respectively, as shown in FIG. 9(a). This pixel data of a1 to a21 is dumped and stored in memory. CC again after dumping
An integration operation of D is started, and block selection is performed in parallel with this integration operation. If the second block (No. 9~
No. 13) has recent defocus, this pixel data (a9 to a13) is stored in memory as reference data.

Next, when the second integration is completed, the pixel data are designated as b1 to b21, respectively, as shown in FIG. 9(b). These data are also dumped in the same way, and after the dump is completed, integration is started again. An area (No. 5 to No. 17) expanded by four pixels left and right from the previously stored reference portion is used as a reference portion, and this pixel data (b5 to b17) is used as reference data. Thereafter, correlation calculations and interpolation calculations are performed between the standard data and reference data to calculate the movement of the subject. This operation will be explained in detail later in the blurring operation section. As a result of the calculation, for example, if the subject has moved 3 pixels to the left,
The next reference section is shifted 3 pixels to the left from the previous reference section (No. 9 to No. 13), and No. 6~No. Set it to 10. Once the reference portion is determined, the pixel data (b6 to b10) is stored as the next reference data.

Furthermore, the pixel data when the third integration is completed are respectively designated as c1 to c21 as shown in FIG. 9(c). Similarly to the above, the area (No. 2 to No. 14) is expanded by 4 pixels to the left and right from the reference part (No. 6 to No. 10).
is the reference part, standard data b6 to b10 and reference data c
A blur calculation is performed between 2 and c14.

[0036] The selection of the reference portion and the reference portion as described above is repeated in order to perform blur detection. If the contrast of the reference portion decreases as detection continues and becomes less than a predetermined value, the one with the highest contrast among the first to third blocks described above is selected as a new reference portion. Furthermore, when the reference portion reaches the edge as shown in FIG. 10 and the reference portion cannot be set, a new block with the highest contrast is similarly selected. However, even in the block with the highest contrast, if the contrast is less than a predetermined value, the reference portion becomes "none" and the next blur calculation becomes impossible.

The above is an explanation of block selection (selection of reference portion/reference portion).

[3] Blur Calculation Blur calculations can be broadly divided into correlation calculations and interpolation calculations. First, correlation calculation will be explained. Note that this blurring calculation will also be explained using the horizontal direction as an example.

The standard data is s(x) to s(x+4), the reference data is r(x-4) to r(x+8), and the correlation value is defined as follows.

[0040]

[Math 1]

In equation (1), n indicates how many pixels the standard data and reference data are shifted from each other, and this correlation value H(n
) is the minimum, and n represents the relative movement amount of the subject. That is, for example, if the subject has not moved at all and there is no camera shake, H(0) will be the minimum value, and if the subject has moved two pixels to the right, H(2) will be the minimum value.

As a specific example, a correlation calculation will be performed using the standard data (a9 to a13) and reference data (b5 to b17) shown in FIG. 9(b). According to equation (1), n is -
The correlation values from 4 to 4 are calculated as follows.

[0043]

[Math 2]

[0044]

[Math 3]

0045::

[0046]

[Math 4]

In this example, the subject has moved three pixels to the left, so H(-3) is the minimum value. By the way, the amount of blur on the film surface when the subject moves by one pixel is 60 μm, so in this case, the amount of blur on the film surface is 180 μm.
It is moving μm.

What has been explained above is the correlation calculation.

Next, interpolation calculation will be explained. FIG. 11 is an example showing the distribution of correlation calculation values, in which each correlation data is plotted with n on the horizontal axis and correlation value H(n) on the vertical axis. As a result of the correlation calculation, n=1 has the highest degree of correlation. However, as can be seen from the figure, n, which actually has the highest correlation, exists between 0 and 1. Interpolation calculation is used to find this value.

FIG. 12 shows the minimum value P(m) in P(n) (n=-4...4) and the points P(m-1) and P(m+) on both sides thereof.
1) is plotted.

First, draw a straight line passing through P(m) and the larger of P(m-1) and P(m+1) (represented by straight line 101 in FIG. 12). Next, P(m-1) and P(m+1
), and draw a straight line with the opposite sign and the same absolute value as straight line 101 (in Figure 12, straight line 1
02). Let the value of n at the intersection of the two straight lines 101 and 102 be the horizontal blur pixel amount Gx. However, in Figure 11, P(-
4) or P(4) becomes the minimum value, the above calculation cannot be performed. Therefore, the detection range of Gx is -3.
The value is limited to between 5 and +3.5, and values larger than this are considered undetectable because the amount of blur is too large, and the calculation result is invalidated. Note that this amount of blurred pixels Gx is calculated using the following formula.

[0052]

[Math 5]

[0053]

[Math 6]

The amount of blurred pixels Gx obtained in this way is “
This indicates the number of pixels shifted to the left or right. YM/C is defined as a reliability value for this amount of blurred pixels Gx. However, YM and C are the values shown in FIG.
It is defined by the following formula.

[0055]

[Math 7]

[0056]

[Math. 8]

[0057] This C represents the slope of the straight line 101,
The steeper the slope, that is, the larger C, the higher the contrast. Furthermore, the smaller the correlation value, the higher the degree of correlation, and therefore the smaller YM, the higher the reliability. From the above, the smaller YM/C is, the higher the reliability is. In this example, Y
If M/C is 1 or more, the reliability is considered low and the calculation result is invalidated.

Next, based on this horizontal direction blurred pixel amount Gx, it is calculated how much blurring actually occurs on the film surface. As mentioned above, in the camera of this embodiment, CC
When one pixel is shaken on D, the amount of shake on the film surface is 60
It is μm. Therefore, assuming that the actual amount of blur in the horizontal direction on the film surface is Bx, the amount of blur Bx is determined by the following equation.

[0059]

[Math. 9]

After the amount of blur is determined as described above, the velocity of blur Vx is then determined. This blur speed Vx is how much the subject moves on the film surface in 1 second (horizontal direction)
It is defined by the following formula.

[0061]

[Math. 10]

The amount of blur and the speed of blur are calculated as described above. Up to this point, the amount of blur and the speed of blur in the horizontal direction have been found, but since they can be found in the vertical direction using exactly the same calculations, a description of the vertical direction will be omitted. Note that the amount of blur and the speed of blur on the film surface in the vertical direction are defined as By and Vy, respectively.

[4] Averaging Process Once the blur velocity is determined by the blurring calculation as described in [3], averaging processing is then performed. In this averaging process, the above-described blur calculation is performed four or eight times, and then the data are added and averaged. In this way, one average blur speed is determined in each of the horizontal and vertical directions. However, blur speeds that have become invalid (low contrast, large YM/C, large amount of blur, etc.) are not added to the average. Furthermore, if the number of valid data is less than three, the average blur speed for that time is invalidated. Note that one set of shake detection is defined as one set of shake detection in four cycles, which calculates one average shake speed from four shake speeds.
8 cycles are called one set of blur detection, in which one average blur speed is calculated from the individual blur speeds.

Table 1 shows specific data on blur speed, and the averaging will be explained with reference to this table. Note that this table is an example in which one set is 8 cycles. There is data in which the blur speed is invalidated in the table, but this occurs when "all blocks in the island have low contrast and there is no reference part" as mentioned earlier, or when "the amount of blur is too large (3 This is a case where the blur (larger than .5 pixels) cannot be detected. These invalid data are not added to the arithmetic average. Further, as described above, when YM/C is 1 or more, these are also not added to the average because the reliability is low. Therefore, the average blur speed of this set is obtained by averaging the blur speeds at i=3, 6, and 7, and is expressed by the following formula.

[0065]

[Math. 11]

[0066]

[Table 1]

[0067] Once the average blur speed has been determined as described above, weighting is then performed on the main island and the sub-island. This weighting is a calculation for more accurately capturing the movement of the main subject, since the purpose of the present invention is to detect blur of the main subject.

Weighting will be explained with reference to Table 2.

First, a case will be described in which both the average blur velocity obtained from the main island and the average blur velocity obtained from the sub islands are valid. In this case, the output blur speed Bo, which is the final output, is determined using the following equation.

[0070]

[Math. 12]

[0071]

[Math. 13]

To explain these equations, if the average blur velocity of the main island is smaller than that of the sub-island, the data of the main island is directly used as the output blur velocity Bo. Conversely, if the average blur velocity of the main island is larger than that of the sub island, the average of the two velocity data is set as the output blur velocity Bo. In other words, more emphasis is placed on the main island than on the sub-island, and on those with small blurring speeds rather than on those with large blurring speeds. This is because, when comparing the main subject and the background, the blurring speed of the background is generally greater than that of the main subject. To give one notable example, when you take a photo while following a moving subject, the main subject is almost stationary, while the background is moving at a tremendous speed. . For this reason, if the blur speed of the main island is small, it is taken as Bo, and if the blur speed of the sub island is small, the average of the two blur speeds is taken as Bo.

Next, if the average blur velocity obtained from the main island is valid but the one obtained from the sub island is invalid, the data of the main island is used as is as Bo. Conversely, if the data on the main island is invalid and the data on the secondary island is valid, the data on the secondary island is
o. Furthermore, if both the main island and the sub-island are invalid, the output blur speed Bo is also invalid.

Furthermore, if the sub-island is not selected during island selection and blur detection is performed only on the main island, if the data of the main island is valid, it will be used as Bo, and if it is invalid, the output Bo will also be used. shall be invalidated.

By performing the weighting as described above, the output blur speed Bo, which is the final blur speed output, is calculated.

[0076]

[Table 2]

FIG. 13 shows the operational sequence of blur detection explained in [1] to [4] above, and the horizontal axis is time. Note that this figure shows one set of four cycles.

When the camera shake detection operation starts, CCD integration is started (integration (2)) and at the same time island selection is performed. This island selection has already been explained in [1]. After the integration (2) is completed, the pixel data of the two selected islands are individually dumped and stored in memory. (Dump■) However, the integration time of CCD integration is as shown in Figure 4.
As explained above, it depends on the brightness of the object, and if the brightness of the object is too low, it will take a long time to complete the integration. For this reason, in this embodiment, the longest integration time when detecting blur is 10 ms, and beyond that, the next blur calculation is impossible because the subject is low in brightness.

Next, start the CCD integration again (integration ■
) At the same time, block selection is performed in each of the two islands. This block selection has also been explained in [2]. When the integration (2) is completed, the pixel data of each island is similarly dumped. (dump■)
After the dump (■) is completed, the integration (2) is started, and at the same time, the shake calculation (2) is performed. This blurring calculation (2) performs a correlation calculation/interpolation calculation between the reference part set at the time of block selection and the reference part of the pixel data stored in the dump (2), and calculates the amount of blurring and the blurring speed. At this time, a reference portion for the next blur calculation is set in accordance with the determined amount of blur.

After the integration (2) is completed, the pixel data is similarly dumped (dump (2)), and blurring calculation (3) is performed. Correlation and interpolation calculations are performed between the reference part set in this blur calculation ■ Blur calculation ■ and the reference part of the pixel data stored in dump ■.
Find the amount of shake and the speed of shake. Thereafter, CCD integration and blur calculation are repeated in the same manner, and when blur calculation (2) is completed, averaging processing is performed. As explained in [4], this averaging process adds and averages the four blur velocities obtained in blur calculations ① to ②, and further weights the data of the main island and the data of the sub-islands.

By performing the blur detection operation as described above, the output blur speed Bo, which is the final output, can be determined. In the case of blur detection using 8 cycles as one set, blur calculations are performed from ■ to ■ in the same operating order as in the case of 4 cycles, and the data of eight blur speeds are averaged.

FIGS. 14 to 17 are flowcharts showing one set of blur detection operations. The blur detection operation will be explained in more detail with reference to these flowcharts. In #202, the microcomputer detects this blur as 1.
Determine whether it is the first time or later, and set j to 4 or 8. In #204, CCD integration is started. In parallel with CCD integration, island selection is performed in #206 and #208. In #210, it is determined whether two islands that meet the conditions have been selected. If NO, that is, there are no sub-islands that meet the conditions, the process advances to #212 and sets a flag for determining only the main island. #210 is Y
If it is ES, proceed to #213 and set both the main and sub island determination flags. After selecting an island, CC
When the D-integration is completed, the pixel data is dumped and stored in the memory in #214.

After dumping, CCD integration is started again at #216. In parallel with the integration operation, a main block is selected in #218. This main block selection selects the block with the most recent defocus or the maximum contrast within the main island. In #220, it is determined whether a main block that satisfies the condition (contrast is greater than or equal to a predetermined value) has been selected. YES with #220
If so, the pixel data of that block is stored as reference pixel data in #224, and a flag for determining the main reference portion is set in #226. Conversely, if all the blocks within the island have low contrast and no main block satisfying the conditions has been selected, the main reference portion determination flag is cleared in #222.

Next, select the sub-block, #
At step 228, it is determined whether there is a secondary island. This determination is made based on whether a flag is set in #213. If there is a sub-island, the sub-block is selected and the flag is set in steps #230 to #238. This operation is the same as that of the main block, so the explanation will be omitted. If there is no secondary island, it is not possible to select a secondary block, so the process directly proceeds to #240. C in #240
After completing the CD integration, dump the pixel data.

After completing island selection and block selection,
Starts blur calculation operation. The operation order of blur calculation will be explained with reference to FIG. In #250, i is set to 1. This i indicates the number of blur calculations, and is incremented by 1 each time a blur calculation is performed. CC in #252
D-integration is started, and in #254, a flag is checked to see if the main reference part has been determined. If the main reference portion has been determined, the main island blur calculation is performed in #256. As described above, this blurring calculation involves performing correlation/interpolation calculations between the reference part and the reference part to obtain the blurring amount, blurring speed, and YM/C. In #258, it is determined whether the determined amount of blur is within the detectable range. In this embodiment, if the blurred pixel amount Gx exceeds ±3.5 pixels, it is determined that detection is impossible and the data is invalidated. If it is within the detection range #2
At 60, the vibration speed is memorized. At this time, Y
The M/C value is also stored in memory. After the blur speed is memorized, the next reference portion is set based on the blur pixel amount Gx calculated in #262. In #264 and #266, it is determined whether the contrast of the set reference portion is greater than or equal to a predetermined value.
Then, it is checked whether the reference part is within a settable range. If any of these is NO, the process advances to #268 and the main block is reselected. This main block selection is
Select the block with the highest contrast within the main island. Main block selection in #268 is #264,2
This is not limited to the case where the answer is NO at #66, but also when the main reference portion has not been determined at #254, or when the amount of blurred pixels calculated at #258 is too large. In #270, it is determined whether a main block that satisfies the condition (contrast is equal to or greater than a predetermined value) in #268 has been selected. If the main block that meets the conditions is selected #2
Proceed to 72.

#262 to 266 or #268 to 2
If the main reference portion is set in step 70, the pixel data is stored as reference pixel data in #272, and a flag for determining the main reference portion is set in #274. Conversely, if the main reference part has not been determined, the main reference part determination flag is cleared in #276.

[0087] This completes the blur calculation on the main island and the setting of the next main reference portion. Next, a blur calculation on the sub-island and setting of the sub-reference part are performed.

At #280 in FIG. 16, it is determined whether a sub-island exists. If there is no sub-island, the process proceeds to #306 without performing any blurring calculation or reference section setting. If there is a sub-island, the blur calculation on the sub-island and the setting of the sub-reference part are performed in steps #282 to 302.
Since this operation is exactly the same as that of the main island, the explanation will be omitted here.

[0089] Once the blur calculation and reference section setting for the main and sub islands have been performed as described above, the CCD
Upon completion of the integration, each pixel data is dumped in #306. After the dump in #306 is completed, i is set to i+1 in #308, and it is determined whether i=j in #310. j
is 4 or 8, which was set in #202. i=
If it is not j, the process advances to #252 and blur calculation is started again. Thereafter, the blur calculation is repeated until i=j. If i=j, the final blur calculation starts from #320 in FIG.

At #320, the flag is checked to see if the main reference portion has been determined. If the main reference part has been determined #3
At step 22, a blur calculation is performed, and after checking the result of the calculation, the blur speed is stored in memory. (#324, 326) If the main reference portion has not been determined, the blur calculation is not performed and the process directly proceeds to #328.

In #328, it is checked whether a sub-island exists. If a sub-island exists, a flag is checked in #330 to see if the sub-reference part has been determined. If the sub-reference part has been determined, a blur calculation is performed, and after checking the calculation result, the blur speed is stored in memory. (#332~3
36) Direct #3 if there is no sub-island or sub-reference part
Proceed to 38. This completes the last blur calculation, and four or eight blur velocities for each of the main island and the sub island are stored in memory. (If there is no sub-island, then only the main island) The blur calculation is performed four or eight times by the operations described above.

Next, an average processing operation for these data is started. In #338, the blur speed data on the main island is averaged. At this time, as described above, YM/C is checked, and unreliable data is treated as invalid and not added to the averaging. Further, if the number of valid data is less than three, the average blur velocity on the main island is invalidated. After calculating the average blur speed of the main island in #338, it is determined in #340 whether or not a sub-island exists. If a sub-island exists, the blur speed data is averaged in the same way as for the main island. If there is no sub-island, the average blurring speed of the sub-island is zero. Once the average blur speeds for each of the main and sub islands are determined as described above, they are weighted in #344. This weighting has already been explained, and the magnitudes of the average blur speeds of the main and sub islands are compared to output the final output blur speed Bo. This is the end of one set of blur detection.

As explained above, the blur detection device of the present invention determines the position of the main subject in the photographic screen in a camera capable of multi-point distance measurement, and detects the image of the image sensor corresponding to that position. This method detects subject blur based on data. By using such a detection method, it is possible to accurately detect the blur of the main subject no matter where the main subject is located within the photographic screen. The blur data detected by the blur detection device of the present invention can be used for various controls related to blur, such as blur warning and display. An example of its application will be explained below.

FIG. 18 shows a display device called a dynamic indicator, which displays the texture of a photographed photograph. This display is display No. The higher the display number, the clearer and still the photo you can take; conversely, the higher the display number. This indicates that the lower the value, the more flowing and dynamic photos you can take. In this embodiment, this display device is provided within the viewfinder. Table 3 shows the display state of the dynamic indicator under various conditions. Generally speaking, even when photographing the same subject, the faster the shutter speed and the shorter the focal length of the photographic lens, the sharper and clearer the photograph will be. However, in the embodiment shown in Table 3, the display is changed not only depending on the shutter speed and focal length but also on blur data.

[0095] Referring to Table 3, this will be explained using specific numerical values. For example, the focal length of the photographic lens is 100mm.
Let us consider the case where the shutter speed is 1/125. Under these conditions, if the output blur speed Bo determined by the blur detection device of the present invention is 30 mm/s, the dynamic indicator will display the display No. Display 2. At the same focal length and shutter speed, if the blur speed Bo is 10 mm/s, the dynamic indicator is N.
o. 3 is displayed, and if the shake speed Bo is 1 mm/s, No. Display 4.

In the embodiment shown in Table 3, the focal length is short;
The faster the shutter speed and the smaller the blur speed, the more display
o. It is controlled to raise the As explained above, in this application example, the photographer can imagine the dynamic feeling of the photograph (finished photograph) taken at the time when the subject is captured through the camera's finder.

[0097]

[Table 3]

Next, as another embodiment, a panning mode using the above-mentioned blur detection device will be described. Panning is one of the photographic techniques for depicting dynamic motion, and is a photographing method in which the camera is moved in accordance with the movement of the subject while the shutter is open. In the photographs taken, the subject is still and the background is flowing, creating a sense of movement. Now, the purpose of the panning mode is to create a more dynamic feeling in the above-mentioned shooting operation. To explain the outline, the camera shake is detected using the camera shake detection device mentioned above, and if the camera shake of the main subject being chased (relative shake between the camera and the main subject) is small, the shutter speed is corrected by increasing the shutter speed. However, if the main subject is significantly shaken, the shutter speed is shortened. With this kind of control, when an experienced photographer takes a panning shot, the shutter speed becomes longer, allowing him or her to take a picture with a moving background and a moving background. Conversely, when a beginner takes a panning shot, it is possible to prevent the subject itself from becoming blurred.

The operation of the blur detection device of the present invention in the panning mode will be described below. Note that switching from normal photography to panning mode is performed by the mode switching unit 20 shown in FIG. Furthermore, in panning mode, the braking judgment shown in Figure 6 is not performed, and all continuous A
Do F.

FIG. 19 is a diagram showing a sequence of blur detection in panning mode. In normal shooting mode,
As shown in Figure 4, there was an interval of 0.5 seconds or more between one blur detection and the next, and the AF operation was repeated during that time, but in panning mode, the AF operation and blur detection were performed once as shown in Figure 19. Do this alternately, leaving no 0.5 second intervals. This is because in panning shooting, the subject is moving at high speed, so a lot of blur data is used to capture the subject's movement more accurately. This blur detection and AF operation are similarly repeated until S2 is turned on and the release operation begins.

FIG. 20 is a flowchart showing the operational sequence of blur detection in the panning mode. As shown in the figure, in the panning mode, the second island is set as the main island (#406), and no sub-island is selected. After that, blur detection is performed only on the second island. This is because in panning shots, it is necessary to capture the movement of a subject moving at high speed, so it is necessary to reduce the calculation time as much as possible. In other words, by reducing the number of islands for blur detection to one, the calculation time and the time to dump imaged data are approximately halved. Also select the second island.
The island was used when chasing a moving subject.
This is because the photographer often focuses on the subject. Furthermore, when performing panning shots, the subject is mostly moving horizontally, such as a car, and horizontal blur data is mainly required. For this reason, in the panning mode, blur detection is performed only on the second island.

The operations after the island selection are the same as those performed in the normal photographing mode when blur detection is performed only on the main island without selecting the sub-island, so the explanation will be omitted here.

Once blur data is determined by the blur detection method described above, exposure compensation is performed based on this data. FIG. 21 is an example of an exposure program in panning mode of a camera to which the present invention is applied. The solid line is a program diagram when the output vibration speed Bo is 3 mm/s. This program diagram shifts left and right depending on the magnitude of the blur. Specifically, when the amount of blurring is large, the shutter speed is shifted to a high speed side, and when the amount of blurring is small, the shutter speed is shifted to a low speed side. This situation is shown in the program diagram indicated by broken lines. However, in this embodiment, this shift range is from 1/30 to 1/125, and Bo≦1.5 mm/
The shutter speed is set to 1/30 for s, and 1/125 for Bo≧6 mm/s. By shifting the program diagram as described above, it is possible to preferentially determine the shutter speed according to the magnitude of blur for subjects having the same Ev value.

[0104] By controlling the exposure as described above,
If an experienced photographer takes a panning shot, the shutter speed will be longer, allowing him or her to take a moving photo with a flowing background. Conversely, when a beginner takes a panning shot, it is possible to prevent the subject itself from becoming blurred.

[0105]

As explained above, the blur detection device of the present invention has a plurality of focus detection image sensors and detects a subject based on the image data of the image sensor of any one of the plurality of image sensors. In a device that calculates blur data, the photographer can manually select which image sensor to use for blur detection. Therefore, no matter where the main subject is placed within the photographic screen, blur detection can be performed by selecting the image sensor corresponding to that position. This allows the photographer to freely set a desired composition.

Furthermore, in a camera that is capable of multi-point distance measurement and allows the photographer to freely manually select which image sensor to use for blur detection, the manual selection means for selecting the image sensor to perform blur detection is focus detection. By also serving as a selection means for selecting the element used for the camera, by selecting the image sensor that is capturing the image of the main subject, the photographer can focus on the main subject and at the same time accurately detect the camera shake. be able to.

Furthermore, the manual selection means is a switching means for switching to the panning mode when performing panning, and by the switching, a predetermined one of the plurality of image pickup devices is selected.
In a blur detection device that selects one image sensor, the time required for blur calculation etc. can be shortened by using only one image sensor for use in blur detection.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the principle of a focus detection device used in the present invention.

[Figure 2] (a) is a diagram showing the distance measurement range in the field of view finder of the camera, and (b) is a diagram showing the CC corresponding to the distance measurement range.
FIG. 3 is a layout diagram of a D-line sensor group.

FIG. 3 is a block diagram of a camera to which the present invention is applied.

FIG. 4 is a diagram showing a blur detection sequence of the present invention.

FIG. 5 is a flowchart showing the operation order of the camera to which the present invention is applied.

FIG. 6 is a flowchart showing the operation order of the camera to which the present invention is applied.

FIG. 7 is a diagram illustrating a method of island selection based on differences in the position of the main subject.

FIG. 8 is a flowchart showing an operation procedure for island selection.

FIG. 9 is an explanatory diagram illustrating a selection method for block selection.

FIG. 10 is an explanatory diagram illustrating a selection method for block selection.

FIG. 11 is a graph showing the distribution of correlation calculation values.

FIG. 12 is a graph used for interpolation calculations.

FIG. 13 is a diagram illustrating the operational sequence of blur detection with one set of four cycles.

FIG. 14 is a flowchart showing the operational sequence of blur detection.

FIG. 15 is a flowchart showing the operational sequence of blur detection.

FIG. 16 is a flowchart showing the operational sequence of blur detection.

FIG. 17 is a flowchart showing the operational sequence of blur detection.

FIG. 18 is a diagram showing a display state of a dynamic indicator.

FIG. 19 is a diagram showing a blur detection sequence in panning mode.

FIG. 20 is a flowchart showing the operational sequence of blur detection in panning mode.

FIG. 21 is a graph showing an exposure program in panning mode.

[Explanation of symbols]

6 Imaging device 10 Microcomputer 11 AF sensor 19 Shake data output section 21 Manual island selection section

Claims (3)

[Claims] 1. At least two or more focus detection image sensors, a manual selection means that allows a photographer to manually select which element among the plurality of image sensors is used for blur detection, and What is claimed is: 1. A camera blur detection device, comprising: calculation means for calculating relative blur data between a camera and a subject based on image data of an image sensor selected by a selection means. 2. The camera according to claim 1, wherein the manual selection means also serves as selection means for selecting which of the plurality of focus detection image sensors is used for focus detection. Shake detection device. 3. The manual selection means is a switching means for switching to a panning mode when performing panning, and when the switching means is switched, a predetermined one of the plurality of image pickup devices is selected. The camera shake detection device according to claim 1, characterized in that: