JP2814945B2 - Automatic focus detection device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic focusing system for a TTL camera.
The present invention relates to a point detection device. 2. Description of the Related Art Conventionally, a focus detection device for a TTL camera has been used.
The luminous flux coming from different areas of the pupil of the imaging optical system
The photographing optical system is determined based on the relative displacement of the plurality of subject images.
Of the so-called pupil division method that detects the focus adjustment
Point detection devices are known. For example, Japanese Patent Publication No. 57-49
No. 841 discloses a lens holder disposed near the primary image plane.
Pair with the light-receiving element array
An automatic focus detection device of this kind composed of rays is disclosed.
ing. Further, Japanese Patent Application Laid-Open No. 54-104859 discloses
Is the field lens located on the primary image plane and the primary image plane
Re-imaging lenses for re-imaging the formed image on the secondary image plane
And two image sensor arrays arranged on the secondary image plane
This type of automatic focus detection device is disclosed.
You. [0004] However, this is not the case.
In such a conventional pupil division type automatic focus detection device,
It had the following disadvantages. That is, this species
In the automatic focus detection device, the optical system on the focus detection side
Therefore, a predetermined F is set on a plane orthogonal to a predetermined position on the optical axis.
Since the number pupil is specified, the F number of the exit pupil
Is greater than the given F-number or the exit pupil
F number is the same or smaller than this predetermined F number
Also, an interchangeable lens whose exit pupil position is different from the predetermined position
When mounted on a camera body having the automatic focus detection device
In that case, the exit pupil causes vignetting in the focus detection light beam.
Vignetting may occur on the image plane on the focus detection optical system side.
If the image is not uniform, the image of the subject will be distorted,
Focus detection could not be performed. [0005] This disadvantage is disclosed in Japanese Patent Publication No. 57-49841.
1 will be described in more detail with reference to a conventional example disclosed in US Pat.
You. FIG. 1A is a schematic side view of the device, and FIG.
FIG. 4 is a front view of a conversion element. A field lens behind the taking lens (11)
(12) is located near the focal plane after the field lens (12).
A plurality of micro lenses (13), (14), (15) ...
Pairs behind them corresponding to lenses (13), (14), (15) ...
Photodetectors ((13a), (13b)), ((14
a), (14b)), ((15a), (15b))... The light receiving sections (13a)... Have the subscripts a and b
Image sensor array, each micro lens
The position of the light receiving unit, which is a pair of photoelectric conversion
The exit pupil position of the lens (11) is approximately the same for each microlens.
The curvature of each micro lens is formed so as to come to the position of the role.
You. In addition, the field lens (12)
The closer the microlens is to the lower end, the more the optical path needs to be bent.
The exit pupil of the taking lens (11) is at the predetermined position (16).
When the image on the light receiving surface of each pair of photoelectric conversion elements
So that they exist completely overlapping each other
The image of the light receiving sections (13a), (14a), (15a).
Images of the photoelectric conversion elements (13b), (14b), (15b) ... at the position (11a)
Are positioned so that they overlap each other at position (11b).
The curvature of the small lens is determined (hereinafter each photoelectric conversion element
Field lens (12), micro lens (13) etc.
Set the position where the images from the focus detection optics overlap each other
Eye position). [0008] In this automatic focus detection device,
The luminous flux that is inserted is almost
Only when there is no failure, that is, F number
Small bright lens or large F-number
Even if the exit pupil position is equal to the set position, vignetting
For a lens in which the effect of
Focus detection can only be performed effectively. For example, in the case of a 35 mm single-lens reflex camera,
If you think about it, the exit pupil of an interchangeable lens that is a shooting lens
The position is from about 50mm from the focal plane to over 400mm
The F-number is also F1.2
There are dark objects that range from F11 to F11. If an automatic focus detection device corresponding to FIG.
In this case, the set pupil position (16) is set at 100 mm from the focal plane (hereinafter referred to as the pupil position).
Thereafter, the distance between the set pupil position and the focal plane is represented by PO. did
Therefore, in this case, design at PO = 100 mm) and use it for detection.
Spread of the luminous flux to be used, that is, the light receiving portion ((13a), (13b)),
((14a), (14b)), ((15a), (15b))
Do not design the limited spread of the detection light beam to F4.
If it is darker than F4 and the distance between the exit pupil position and the focal plane
(Hereafter, this will be referred to as PO '.)
With regard to interchangeable lenses, the detection accuracy of the automatic focus detection
It will be much lower. This will be described with reference to FIG.
Compare the appearance and degree of vignetting with various shooting lenses
FIG. 4 is an explanatory diagram showing the detection light flux with respect to the above design value.
4. When the set pupil position is PO = 100mm,
Size is F6 and PO '= 100mm, 50mm, ∞
Is shown. FIG. 2A shows a case where PO '= 100 mm.
Each of the photoelectric conversion elements receiving the light beam having the spread of F4
The light-receiving parts ((15a), (15b)), ((14a), (14b))
The light beams that have passed through the pupil of F6 at the light receiving portions ((15
a), (15b)), ((14a), (14b)) ...
Equally assigned. Therefore, when the subject has uniform brightness
As shown in FIG. 2 (D), the output of each light receiving section (15a).
(15a1), (15b1), (14a1)... Become uniform. In other words, this place
The detection accuracy despite vignetting
No degradation occurs. That is, by the row of the photoelectric conversion element pairs,
It is possible to detect a shift between two images. FIG. 2B shows the case where PO '= 50 mm, and
Light passing through the pupil of the photographic lens at F6 as shown in FIG.
Light section (15a) ... distributed at different ratios for each location
You. Therefore, the output of each light receiving section (15a) at this time is shown in FIG.
As shown in (E), the output is (15a1) ...
The output to be done is significantly different. Here, the minute lenses (13) at both ends and the minute lenses (13)
The position of the lens (15) is from the center micro lens (14)
Fig. 2 for the case of + 2.5mm and -2.5mm
When the degree δ of vignetting in (E) is calculated, the average is 1
Approximately, Is a very large value. That is, the pair
The photoelectric output of the photoelectric conversion element
Regardless, the output varies greatly depending on vignetting
In such a situation, the photoelectric conversion element
It is very difficult to detect a shift between two images by a pair.
You. FIG. 2C shows a case where Po '= ∞.
The vignetting in this case is completely opposite to that in FIGS. 2 (B) and 2 (E). You
That is, the photoelectric output for a subject of uniform brightness is shown in FIG.
(F), the degree of vignetting δ is ± 2.5mm
In [0018] [Equation 2] Approximately. That is, the case of FIG.
Similarly, the photoelectric output of the pair of columns of the photoelectric conversion elements has a uniform luminance.
Despite vignetting, output varies greatly with vignetting
And it is very difficult to detect a shift between two images.
You. As described above, the accuracy of focus detection is adversely affected.
In order to avoid the situation, the set pupil position is common
With different pupil sizes, and therefore different F-numbers
Providing multiple focus detection means and switching between them
A plurality of focus detection means having different formulas and set pupil positions are provided.
The aperture of the focus detection optical system
The method of changing the focus and the situation that adversely affects the focus detection
Switch the processing system of the image sensor output when
Switching, such as changing the filter characteristics or focusing
The method used to switch the algorithm
The situation that adversely affects the focus detection is predicted in advance
Various methods of measuring and switching manually, and attached lenses
System that automatically detects the open F-number of the
Furthermore, the F-number and pupil position are detected and automatically turned off.
An alternative method is conceivable. [0021] However, these methods still have
It has the following disadvantages. That is, in the case of manual operation,
The user has to decide how to switch,
Not only is it cumbersome,
There was a risk of being wrong. Further, even if the automatic switching is performed,
It can be switched uniquely depending on the specifications of the attached lens.
Therefore, the vignetting that actually occurs depends on the open F number of the lens.
Lens and exit pupil position, as well as lens extension and aperture
Diameter and position of restricted part of lens other than, dirt on lens surface
Various factors such as non-uniform transmittance due to
In such a case, it is truly immediate
Adjustments cannot be made, and the lens
A mechanism that transmits the exit pupil position from the lens side to the camera body side
Required and expensive.
Was. The present invention eliminates the above-mentioned conventional disadvantages.
And a pair of focus detection optical systems such as vignetting.
Detects imbalance in the subject image and automatically corrects it
Focus detection device that enables accurate and accurate focus detection
It is intended to provide. In order to achieve the above object,
The main point is that the pupil of the imaging optical system has different areas
Focus detection that forms a pair of subject images from light beams incident from
An output optical system and a pair of images for receiving the pair of subject images.
Image sensing means having a sensor array;
From the relative displacement of a pair of outputs from the image detection means,
Focus detection means for detecting a focus adjustment state of the photographing optical system
And the degree of correlation between a pair of outputs from the image detecting means.
The correlation degree detecting means for detecting, and the correlation degree detecting means
Depending on the detected degree of correlation,
Changing means for changing the range of a pair of outputs of
An automatic focus detection device. According to the automatic focus detection device of the present invention, the correlation
Means for detecting the degree of output of a pair of outputs from the image detecting means.
Detects the degree of relevance, thereby determining which output of the image sensor
It is determined whether the focus detection accuracy of the range is poor, and the
According to the degree of correlation detected by the degree of correlation detection means,
Changing the range of a pair of outputs from the image detection means
With this, the focus calculation is performed in the area where the focus detection accuracy is good.
You. Hereinafter, various embodiments of the present invention will be described with reference to the drawings.
I will tell. FIG. 3 is a block diagram common to the embodiments of the present invention.
There is a subject image constituted by the photographing lens (11) and the like.
The field optical system (10) for forming
The focus detection optical system (20) composed of the lens (12)
The point detection optical system (20) has a pair of image sensor arrays
Image detection means (30)
Filter means (34) and focus detection means (40)
The vignetting state detection means (50) follows the output means (40), and further,
Various operation switching machines in response to the output of vignetting state detection means (50)
The detection state control means (60) for controlling the structure is a vignetting state detection means.
When connected to the step (50) and receiving the output of the focus detection means (40)
Shooting controlled by the output of the detection state control means (60)
Auto focus control means (7) for driving and controlling the optical system (10)
0) is provided, and the auto focus control means (70) is
The output (71) is output to a lens driving means (not shown).
You. A part of the output of the detection state control means (60) is indicated by a broken arrow.
(61) to the vignetting state warning display means (not shown)
Output. The vignetting state detecting means (50) is a main part of the present invention.
The details will be described later, but only the main points
5 will be described. FIG. 5 shows the image by the focus detection optical system.
The intensity distribution of the subject image formed on the
For simplicity of explanation, the shooting optical system (10) is
It is assumed that the camera is in focus. FIG. 5 (A) shows a case where there is no vignetting and the state is proper.
FIG. 5B is a diagram showing a mapping image F (x), and FIG.
A pair of images for a subject with uniform brightness when
The photoelectric conversion element array forming the image sensor array is a
Expressed as subscripts for the seri row and b sensor row
The intensity of the subject image formed on each array.
Cloth Va (x), Vb (x) (hereinafter referred to as vignetting function)
And ). Therefore, when the vignetting state occurs, FIG.
(A) image sensor through focus detection optical system
When projected onto the array, as shown in FIG.
-On the row, Fa (x) = Va (x) × F (x) forms an image and the b sensor
Fb (x) = Vb (x) × F (x) forms an image on the column. In the out-of-focus state, the object on each row
Since the body images are shifted from each other on each row, the shift amount is 2Δ
Then, Fa (x) = Va (x) × F (x +
Δ) forms an image, and Fb (x) = Vb (x) ×
F (x−Δ) forms an image. The output of the subject image in this out-of-focus state is optical.
If the system is in the proper state without vignetting, the deviation is the strength
The distributions should be identical and can be overlapped,
If there is vignetting and it is in an inappropriate state, it can not be repeated
No. Therefore, the vignetting state detection means (50)
Output a0 to aN of column a and output b0 of column b
This is to detect the vignetting state by comparing with bN. The focus detecting optical system (20) and the image detecting means (3
0) or the focus detection means (40) is the same as the vignetting state detection means (50).
Or vignetting state by the output of the detection state control means (60).
It is configured so that the characteristics can be switched to eliminate
First, the image detection means (30) is configured as such.
This case will be described as a first embodiment. Focus detecting optical system (20) and image detecting means
The stage (30) is configured, for example, as in the conventional example described with reference to FIG.
It is. In such a case, even if the focus detection optical system (20) is the same,
By simply changing the shape of the light receiving part of the image sensor,
The fixed pupil position is common and the pupil size is different
Wear. Therefore, in the following first embodiment, a different receiving
Image sensor array with optical part shape ((31a), (32a)),
((31b), (32b)) is provided, and focus detection
Assuming that the same optical system (20) is arranged,
Eliminates vignetting by switching di-sensor arrays
The description will be made assuming that the characteristics can be switched as much as possible. Image sensor array (31a) and image
It has the same characteristics as the sensor array (31b) and
None, (arrays (13a to 15a) and arrays (13b to 15b) in FIG. 1)
It is usually used. Also the image
Sensor array (32a) and image sensor array (32b)
Have the same characteristics and are paired with each other. Array (3
2a) and the array (32b)
(31a) and (31b) are smaller than those of
That is, the characteristics of the image sensor arrays (31a) and (31b)
The arrays (32a) and (32b) are vignetting.
It is used when it is broken. Therefore, the image sensor arrays (32a), (3
When using 2b), the size of the set pupil becomes smaller,
That is, since the F number of the focus detection becomes large,
Focus detection when using the image sensor arrays (31a) and (31b).
The image sensor array (32
Switching to a) or (32b) can cause vignetting of the focus detection light beam
Less likely. The image sensor arrays (31a, 31b) are switches.
Switch means (33a) and (3b) connected to the switch means (33a) and (33b).
3b) is linked, and the switch means (33a) is
Switch when it is set to the sensor array (31a) side
Switch the means (33b) to the image sensor array (31b)
It is supposed to be. Switch means (33a), switch means (33b)
Of the image sensor array (31a) switched by
Filter means (34) for adjusting the force is switch means (33a),
The focus detection means is provided after the switch means (33b).
(40) is an image detecting means (3) through a filter means (34).
0). The focus detecting means (40) is an image detecting means (30)
From the output transmitted through the filter means (34)
On the image sensor array based on the algorithm of
The pair of subject images are compared to determine the relative bias of the pair of subject images.
Calculate the position amount and convert the amount to the defocus amount
Is what you want. Next, the operation of the first embodiment will be described.
Normal switch means (33a) and (33b) are image sensor arrays
Since (31a) and (31b) are selected, the filter means (34)
Has its output sent. When no vignetting occurs
Is the output of the vignetting state detection means (50)
(For example, low-level output). Therefore, the detection state control means (60)
Received and compared with the threshold level
Status is not occurring and it is determined that the
Current switching mechanism (switch means (33a) etc.)
Output a control signal to maintain the
The currently selected state of Molly, ie the image sensor
The fact that the arrays (31a) and (31b) are selected is stored. Thus, the switch means (33a), (33b)
Continue to select the image sensor arrays (31a) and (31b),
The output is subsequently transmitted via the filter means (34)
The auto focus control means (70) is also
Out-of-focus information which is the output of (0) and detection state control means (60)
Based on the output of the
A lens drive control signal is sent to the lens drive means. On the other hand, if the vignetting state occurs,
The output of the rare condition detection means (50) indicates that an improper condition has occurred.
(For example, high level output)
Stage (60), and the detection state control means (60) outputs this output.
The vignetting state is compared with the threshold level.
Judge that the limit has been exceeded and switch each operation
Control for switching the mechanism to respond to vignetting
Outputs signals and stores them in internal memory
I do. Therefore, the switch means (33a) and (33b)
Receiving the signal of the state control means (60), the image sensor array
Select (32a) and (32b) and output the output to filter means (34)
To send to. The auto focus control means (70) is
Control of vignetting from output state control means (60)
The lens drive control mode is switched based on the signal. example
For example, a low-speed mode or a stop mode is set. On the other hand, a switched image sensor
The outputs of the arrays (32a) and (32b) are passed through the filter means (34).
Is sent to the focus detection means (40) and the vignetting state detection means (50).
Vignetting state detection means (50) further
The presence or absence of the state is judged, and the detection state control means (60)
If a control signal indicating that the
The focus control means (70) sets the proper state based on the output.
Return to the lens drive control mode and return to the image sensor array (3
2a) and (32b), the focus detection means (40)
Lens to drive the imaging optical system (10) based on the amount of misalignment.
A lens drive control signal is sent to the lens drive means. Further, the detection state control means (60)
Sensor arrays (32a) and (32b) are selected.
After a certain period of time, the image sensor arrays (31a) and (31b)
Inverts to the selected state, and then detects vignetting state again in that state
The degree of vignetting is detected from the output of the means (50). This
Vignetting due to lens replacement etc.
If the error has been resolved,
Switch to disensor array (31a), (31b) and select
Becomes possible. If the image sensor array
B) When (31a) or (31b) is selected, vignetting state detection means (50)
Detects the vignetting state, and the detection state control means (60) detects it.
If it is determined that the limit is exceeded,
(32a) and (32b) are selected again. Further, the image sensor arrays (32a), (32
b) In the state where is selected, still vignetting state detection
The output of the means (50) indicates the vignetting state, and the detection state control means
If it is determined in (60) that it exceeds the limit,
Means that the detection state control means (60)
Output to the state warning display means to warn the user. Further, a pair of image sensor arrays (31a),
(31b) and another pair of image sensor arrays (32a), (32b)
F-number for focus detection by selecting either
Is different from the relative displacement of a pair of subject images.
Since the value of the coefficient when converting to the amount differs, the detection state
The control means (60) selects any pair of the image sensor array.
Signal from the detection state control means (60).
Is sent to the point detecting means (40),
Switch the conversion factor for the quantity. In the first embodiment, the image
Shown is the one equipped with two sets of disensor arrays, but two or more sets
Provided, vignetting state detection means (50), detection state control means (60)
May be selectively switched appropriately according to the output of
No. In addition, the vignetting state detecting means (50) is connected to the switch means (33
The outputs of a) and (33b) may be directly received. Further
An apparatus disclosed in Japanese Patent Application Laid-Open No. 54-104859.
In the case of, since there are multiple sets of image sensor arrays,
Instead of the aperture and field lens in the focus detection optical system.
Switching may be performed. Next, the focus detection optical system (20) is the same as in the first embodiment.
And the image sensor array is also a set.
An example will be described. In the second embodiment, the first
Image sensor arrays (32a), (32b) and
The switch means (33a) and (33b) are unnecessary. As shown in FIG. 3, the filter means (34)
Are the first filter (34a) and the second filter having different filter characteristics.
There are two filters, Iruta (34b). For example, the first file used for the normal state
In Iruta (34a), low-frequency components like a human face
In order to enable focus detection for the included subject, FIG.
As shown in (A), the transfer function shows a fairly high value at low frequency.
H1 (w) shall be used when vignetting occurs.
Vignetting generally occurs at the second filter (34b) used
In that case, the effect will affect the low frequency components.
As shown in (B), the transmission shows a fairly small value at low frequency.
It has a function H2 (w). Note that in FIG.
Fn is the pitch of the photoelectric conversion element that constitutes the image sensor.
Nyquist frequency 1 / (2d) when h is dmm
(Book / mm). The focus detecting means (40) is provided with a first filter (3).
4a) According to the second filter (34b), the detection state control means (60)
Multiple algorithms (40
a) and (40b) are provided. The algorithms (40a) and (40b) are, for example,
As described above, depending on the presence or absence of vignetting
Switching the characteristics of the means (34) to remove low frequency components
Similarly, the output of the filter means (34) is used as the focus detection means.
The sampling interval at the time of capture is set to the focus detection means (40).
Switched when the output of the filter means (34) is input
Data at a point distant from the center of the screen
Data, i.e., make the ranging area variable.
It is configured to: Also, if vignetting occurs
In this case, the low-frequency component is affected,
The output of the filter means (34) used for point detection is, for example,
It is also conceivable to perform logarithmic conversion and focus detection. The other configuration in the second embodiment is similar to the first embodiment.
It is the same as the embodiment. Next, the operation of the second embodiment will be described.
You. The basic operation is the same as in the first embodiment. Vignetting
If no condition has occurred, the filter means (34)
One filter (34a) is selected, and the focus detection means (40) selects
Algorithm (40a) is selected.
Mode). In the normal mode, the image sensor
The subject image information output from the ray is filtered by the first filter (34a).
Filtered by the algorithm of the focus detection means (40) (4
0a) detects the out-of-focus amount, and then detects the vignetting state.
Output means (50), detection state control means (60), auto focus
The operation of the control means (70) is the same as in the first embodiment,
State control means (60) stores that the normal mode is selected.
doing. When vignetting occurs, vignetting occurs
The detection means (50) first detects this, and the detection
State control means (60) outputs a control signal, thereby
In the filter means (34), the focus detection means is provided to the second filter (34b).
In step (40), it is switched to algorithm (40b) (this
This state will be referred to as inappropriate mode hereinafter). Also, the detection status system
The control means (60) memorizes that the incorrect mode has been selected.
ing. When the mode is changed to the inappropriate mode, the same as in the first embodiment is performed.
In addition, the auto focus control means (70) controls the lens drive control mode.
Switch the mode to low speed mode or stop mode. So
To detect the vignetting state again in the inappropriate mode.
If the means (50) detects and the inappropriate state has been resolved,
In the inappropriate mode, the auto focus control means (70)
Return the lens drive control mode to the normal state and set the shooting optical system (1
0) based on the information on the amount of defocus from the focus detection means (40).
Drive. Further, the detection state control means (60)
After reverting to the normal mode and performing re-detection,
If the vignetting state is still not resolved by the normal mode,
Warning to the user by vignetting state warning display means
This is the same as in the first embodiment. The vignetting state detecting means (50) is a filter.
Output of the image sensor array without through the means (34)
You may receive power directly. In this case, the detection
The state control means (60) is in a state where the inappropriate mode is selected.
There is no need to reverse after a certain time, and vignetting state detection means
What is necessary is just to invert according to the output of (50).
The control means (70) also has vignetting output from the vignetting state detecting means (50).
The mode is switched according to the signal indicating the degree of the state
Just do it. The filter means (34) and the focus detection means
(40) and only one of them without switching at the same time
May be switched, and only the focus detection means (40)
If not, analyze the signal that does not pass through the filter means (34)
Prepare multiple algorithms that can be used. Also, filter
-Log conversion means etc. are provided before means (34) and combined
Is also good. Further, the image sensor in the first embodiment
It may be combined with one that switches rays,
-When combined with a means that switches means (34),
Even after switching to the image sensor array for the normal mode,
If the vignetting condition persists, apply the filter
It is preferable to provide a filter for removing low frequency components from the force. Next, the detailed contents of the vignetting state detecting means (50)
Will be described with reference to some examples. First, the first fruit
The principle of the embodiment is that when vignetting occurs,
The degree of correlation between the outputs of the image sensor arrays
Vignetting occurs when the maximum (or minimum) of the function
Less than maximum (or minimum) if not
(Or bigger). FIG. 6A shows an a sensor as described with reference to FIG.
Image sensor in which a row and a b-sensor row form a pair
The output of each column of the array is indicated by a triangle for the a sensor column, and
The sensor row is represented by a cross, and vignetting occurs.
And the subject image is a sensor row and b sensor row
Offset from each other on the row (overlap when in focus)
I have an image. Therefore, a correlation function is defined as in equation (1).
You. Each sensor row is N + 1 of a0 to aN and b0 to bN
, And outputs the following data: T = K− (S + 1) / 2... S: odd number, T = KS / 2... S: even number In equation (1), the range in which the variable S is moved is represented by Smin to Smax.
Then, the constants K and L satisfy the conditions of Tmin ≧ 1, Tmin + Smin ≧ 1, Tmax + L ≦ N, and Tmax + L + Smax ≦ N, respectively. In the definition of the correlation function of the equation (1),
Is assumed that the number of additions of the term of the absolute value of the difference is fixed to L.
However, the number L of addition terms may be changed by the variable S,
In that case, normalize the equation (1) with the number of addition terms L
Is also good. Thus, the correlation function in the case shown in FIG.
C (S) is the minimum value at the point Sm as shown in FIG.
Take C (Sm). C (S) is calculated for the case where S is an integer.
Therefore, the minimum value C (Sm) is generally obtained by appropriate interpolation.
It is. For example, as shown in FIG. 6B, S is an integer.
Is defined as C (So), and its adjacent correlation
The values are C (So-1) and C (So + 1), respectively.
o) and C (So-1) or C (So + 1)
When C (Sm) is interpolated by two straight lines with different signs
Can find the minimum value C (Sm) as shown in the following equation (2).
You. [Mathematical formula-see original document] Generally, when no vignetting state occurs, the following expression is obtained. Is as follows. However, when vignetting occurs,
Similarly, find the minimum value C (Sm) from the output of each sensor row.
Then, if the degree of correlation is low and the minimum value C (Sm) is also 0,
Rather, it is a considerably large value. Therefore, each sensor can be obtained by the equations (1) and (2).
The minimum value of the column correlation is always determined and monitored, and the minimum value C (S
m) is greater than a predetermined value.
Can detect that a vignetting condition has occurred.
You. And the degree of vignetting depends on the magnitude of that value
Can be detected. The value of the minimum value C (Sm) is exactly as shown in FIG.
Indicate the area of the difference between the two patterns during focusing.
And accurately reflects the degree of vignetting. If the average value of the output of the photoelectric conversion element is fixed,
AGC (Auto Gain Control) is applied
In this case, the value of C (Sm) is directly
Parameters, and such A
When GC is not applied, the output of the photoelectric conversion element
If you divide the average value, the degree of vignetting
It is always a good indicator. Using the minimum value C (Sm) as an index,
In other words, this amount is the result of the focus detection calculation in the focus detection means (40).
The amount is automatically calculated, especially for vignetting detection.
This is very convenient because it does not require any calculation. Further, the output of the photoelectric conversion element is made logarithmic.
Consider the case where the above C (Sm) is used for the data
As a result, as shown in FIG.
The logarithm of the function of the subject image on the sensor row is expressed by the following equation (2-1).
Become like (Equation 6) As shown in the above equation, the components in the vignetting state are completely
This is very convenient because it can be separated and extracted. Also,
When the absolute value of the difference in the above equation is taken, the term of the subject image is subtracted,
Only the components in the vignetting state are obtained as in the following equation. ## EQU7 ## Therefore, the logarithmic data is
The amount of C (Sm) obtained is related to the number of terms in the correlation addition.
By dividing by the amount, the vignetting state described later can be easily approximated by a straight line.
The amount corresponding to the similarity slope (γ described later) is calculated.
It is. If the minimum value C (Sm) is used, a small amount of calculation can be performed.
The amount of vignetting is monitored by the amount and the correction amount can be easily calculated.
Can be performed. In the first embodiment, the image
A sensor row and b sensor row constituting the sensor array
Was used to calculate the correlation, but the output of each sensor was
Correlation can be calculated from filtered output.
You. The value of the correlation function C (S) depends on the subject.
Can be used by standardizing C (S).
Wear. For example, according to the maximum value Cmax of the correlation function, equation (1)
By dividing the correlation function C (S) obtained in
Such a correlation function is obtained. ## EQU8 ## Further, a method for obtaining the correlation function C (S)
The correlation between the two sets of outputs even for equations other than equation (1)
Whatever is required is required. For example, the following equation (4) shows
Such a multiplication type correlation function may be used. ## EQU9 ## In this case, a point having a high correlation is a peak.
And this peak varies depending on the subject image
However, for example, the peak value is output from the a sensor or
Specified by the peak value of the autocorrelation function of the sensor output
Monitor the height of this normalized peak value.
Can detect the degree of vignetting.
You. FIG. 8A shows the vignetting state detection method described above.
FIG. 4 is a specific block diagram of the first embodiment of the stage (50). I
Image sensor array (31a), (31b) or (32a), (32b)
Or a pair of outputs of the filter means (34)
Is input to the correlation value calculating means (51), and the correlation value calculating means (5
1) is the minimum (or maximum) C of the correlation function of a pair of inputs
(Sm) is output to the comparing means (53). On the other hand, reference value generation means
(52) outputs the reference value (Qc) to the comparison means (53),
The comparing means (53) outputs a minimum value (or a maximum value) C (S
m) and the reference value (Qc) to determine the minimum value of the correlation function (and
Is the maximum value) C (Sm) is larger (or smaller) than the reference value (Qc)
In this case, vignetting is applied to the detection state control means (60).
A signal indicating that an error has occurred is transmitted. In addition, above
The notation device calculates the minimum value of the correlation function by the correlation value calculating means (51).
(Or the maximum value) C (Sm) was obtained, but the focus detection means (40)
The minimum (or maximum) C (Sm) of the correlation function inside
If the calculated correlation value is
You may lead to 3). In the vignetting state detecting means (50),
Is smaller than the maximum (or minimum) value of the correlation function.
Was used to detect the degree of
However, for example, by-pass between the a sensor row and the b sensor row
Even if there is a difference in assembling level, the correlation function
The maximum (or minimum) value decreases (or the upper
Misdetection that vignetting has occurred.
There is a risk of spilling. [0094] Vignetting state detection avoiding such fears
Next, a second embodiment of the output means (50) will be described. In this embodiment, the vignetting state function is generally
As shown in Fig. 5 (B), each sensor with the distance measurement unit at the center of the screen
It is symmetrical to the center in the sir row
Use that thing. FIG. 7 shows vignetting to the position having the highest correlation.
The output of each sensor when it is
FIG. The amount of shift from each other
From the point Sm which takes the minimum value of the correlation function C (S) shown in FIG. 6 (C)
It can be easily converted. In FIG. 7, the symbol “〇” represents the output of the sensor row a,
The crosses indicate the output of the b sensor array. Generally each other
The amount of deviation is determined by adjusting the pitch of the photoelectric conversion elements that make up the sensor row.
Since it does not become several times, δ (0 <δ <1)
It has a deviation amount. When the outputs of the two sensor arrays are compared,
On the left side, the output of the b sensor row is the output of the a sensor row
It is larger and reversed on the right from the center. Follow
If vignetting occurs, output of both sensors
Integrating the difference between
Has the opposite sign of its integral value. On the other hand, although no vignetting state has occurred,
If there is a difference in the bias level of the sensor output
Means that the integral is in the left and right regions of the center
Since the same sign is used, it is different from the case where vignetting occurs.
Can be different. In FIG. 7, the relationship between a sensor output aM and
B sensor output bM at the same position is a sensor output aM
Sensor outputs bK and b (K + 1) and sensor
-Decimal part δ of deviation amount in the case of pitch (0 ≦ δ
The value obtained by interpolation according to <1) is as shown in Expression (5). (Equation 10) Therefore, the vignetting state detection is performed as shown in the following equation (6).
Determining the output function H, vignetting depends on the magnitude of this function H.
It is possible to detect the state. Now, K = M + q (q is a shift for obtaining the maximum correlation)
The constant H is determined by a predetermined amount).
A constant M representing a point is converted to a variable n representing a general position.
By substituting, This is expressed as So this value
Vignetting is detected by detecting that H exceeds a predetermined value.
The occurrence of a state can be detected. In the second embodiment, both sections are used.
The area where the output of the sensor train is integrated
But from the center where the vignetting effect is great
Only distant sections may be used as integration sections.
More than one and compare each integral value
Is also good. Further, in the second embodiment, the sensor array
Since the difference in output is related to the size of the output,
Normalize the difference by the sum of the output of each sensor array
It may be. In addition, the outputs of both sensor rows are
The function H may be obtained from the filtered output. FIG. 8B shows a method for detecting the vignetting state described above.
FIG. 9 is a specific block diagram of a second embodiment of the stage (50). I
Image sensor array (31a), (31b) or (32a), (32b)
Pair of outputs (when not passing through the filter means (34),
Output) or a pair of outputs of the filter means (34)
It is input to the correlation value calculation means (51) and the deviation means (54). correlation
The value calculation means (51) calculates the minimum value of the correlation function of a pair of inputs (and
Is the maximum value) C (Sm) is obtained, and the deviation amount Sm at that time is calculated as the deviation value.
Output to stage (54). The displacement means (54) biases the pair of inputs.
Relative to the integration means (55) and (56).
Output. The integrating means (55) is a pair of relatively displaced inputs.
Integrate the difference between forces in one region R from the center of the input function
No, outputs the integral value Sgr to the difference means (57). Similarly, the integrating means (56) is relatively displaced.
The integration of the difference between the pair of inputs is
This is performed in the area L on the opposite side, and the integrated value Sgl is output to the difference means (57).
Power. The difference means (57) calculates the difference between the above two integrated values.
Find the difference Sg = Sgr-Sgl and output to absolute value means (58)
The absolute value converting means (58) calculates the absolute value | Sg |
Is output to the comparing means (53). On the other hand, reference value generating means (52)
Is outputting the reference value Qs to the comparing means (53).
The means (53) compares the two outputs and determines the absolute value | Sg |
If it is larger than the value Qs, the detection state control means (60)
Sends a signal indicating that vignetting has occurred
You. The correlation value C inside the focus detecting means (40)
When calculating the deviation amount Sm for obtaining (Sm), this value Sm is
Directly input to the deviation means (54) and omit the correlation value calculation means (51).
It may be omitted. Next, the third actual operation of the vignetting state detecting means (50) will be described.
An example will be described. The principle of the third embodiment is shown in FIG.
The vignetting functions Va (x) and Vb (x) shown in FIG.
In general, a straight line with a slope γ as shown in the following equation (7)
Of the output from both sensor rows.
γ is determined, and the degree of vignetting is determined by the magnitude of the value of γ.
It is to detect. (Equation 12) Therefore, the a sensor row and the b sensor row
When the subject images Fa (x) and Fb (x) are in focus,
Let F (x) be the mapping function and expressed as the following equation (8).
You. (Equation 13) From equation (8), F (x) is eliminated to obtain γ.
The following equation (9) is obtained. (14) Therefore, as the Fa (x), the n-th sensor row
The sensor outputs n and Fb (x) are the n-th element in sensor array b.
Of the sensor element from the center as output bn, x of
By substituting the distance S to equation (9), the following equation (10) is obtained.
Of the vignetting function at each sensor element position
be able to. (Equation 15) In equation (10), P is a sensor row forming element
Is the number of the sensor row central element. On the other hand, if the camera is out of focus, the second embodiment
As described in FIG. 7 in the example, the highest correlation
The output of both sensor rows to each other
Can be obtained in the same way as when focusing.
Wear. In this case, the positions of the elements in both sensor rows are required.
Since one is not necessarily offset by an integer, one output is
What is necessary is just to interpolate and obtain | require by Formula (5). Therefore, the outputs of both sensor rows are constantly monitored.
To sample the output at the appropriate location
Then, the slope γ can be easily obtained by equation (10).
Vignetting can be detected by the size of
You. Therefore, it is detected that the inclination γ is larger than the predetermined value.
To detect that vignetting is occurring
Can be. And the degree of the vignetting depends on the magnitude of this inclination γ.
Degree can also be detected. Note that the slope γ can be obtained not only at one point but also
It is better to find between the number of element positions and determine the average value.
Accuracy is good. Further, as the element position is farther from the center,
However, since the difference between the outputs is large, the accuracy increases,
It is desirable to obtain the slope γ in a region excluding the center. Similarly, the subject image F (x) has a large value.
Area, ie, the sum of Fa (x) and Fb (x) has a large value
Since it is more accurate to obtain the slope γ between the elements, an + b
Find the slope γ at a point where n is equal to or greater than a predetermined value and average it.
No. Also in the third embodiment, both cells
Filtering without directly using the output of the sensor train
Needless to say, the output may be used. Further, the gradient γ of the vignetting function is obtained.
If this is possible, feed it to the focus detection means (40).
To correct the effect of vignetting. That is, the input to the focus detection means (40)
The output of the image sensor array pair (or the filter
Assuming that the output of the switching means is An, Bn, the following equation (11) is used.
The corrected outputs An1 and Bn1 can be obtained, and
Output can be corrected. (Equation 16) N is sensor row output or filtering
Number indicating output order, c corresponds to the center of the screen among n
The number of the position where the filter is being performed, P is the element pitch or filter
Convert the ring output sampling interval to the distance on the screen
Value. FIG. 8C shows the vignetting state detecting method described above.
FIG. 14 is a specific block diagram of the third embodiment of the stage (50). I
Image sensor array (31a), (31b) or (32a), (32b)
Pair of outputs (when not passing through the filter means (34),
Output) or a pair of outputs of the filter means (34)
It is input to the correlation value calculation means (51) and the deviation means (54). correlation
The value calculation means (51) calculates the minimum value of the correlation function of a pair of inputs (and
Is the maximum value) C (Sm) is obtained, and the deviation amount Sm at that time is calculated as the deviation value.
Output to stage (54). The displacement means (54) biases the pair of inputs.
Relative displacement by the shift amount Sm and output it to the inclination detecting means (59).
You. The inclination detecting means (59) has a relatively deviated position.
The slope γ is detected based on equation (10) from a pair of inputs, and absolute
Output to the value conversion means (58). The absolute value conversion means (58)
The absolute value | γ | is output to the comparison means (53). On the other hand, the reference value
The generating means (52) outputs the reference value Qg to the comparing means (53).
Therefore, the comparing means (53) compares the two inputs and obtains the absolute value |
If γ | is larger than the reference value Qg, vignetting will occur.
Is transmitted to the detection state control means (60).
Put out. The output of the inclination detecting means (59),
The inclination γ is fed back to the focus detection means (40).
May be used to correct a pair of inputs using the slope γ.
I don't know. As described above, the three states of the vignetting state detecting means (50)
Although the examples have been described, without being limited to these,
Images that actually cause the vignetting state are superimposed
Detects vignetting from the output of a pair of disensor arrays
Should be fine. Also, the lens information (open
Combination of means to detect emission F number and exit pupil position)
You may let it. Next, the vignetting state detecting means and the detecting state
As a method of realizing control means, these are comparison, calculation
To perform functions with a microcomputer.
Can be manifested. Also, individually or integrally
It can also be realized with a circuit that has been implemented. Generally focus detection means
As a program in the microcomputer
In some cases, the sensor output can be shared on memory or
Vignetting state when considering the interface between each means
The detection means and the detection state control means are also configured as programs.
It is more advantageous to do so. In the embodiment, the unbalanced state is set.
Although explained with vignetting, a ghost or a pair of
Noise coming from performance differences due to deterioration of the image sensor array
The present invention can be applied to an unbalanced state such as noise.
Needless to say. In this case, the memory of the sensor output is
Before or after or both
The sensor output data stored
Focus using filter or filtering means output data
Detection calculation, detection state detection calculation to micro computer
Do more. According to the automatic focus detecting device according to the present invention,
For example, by detecting the degree of correlation by the degree of correlation detection means,
Which output range of the image sensor has poor focus detection accuracy
Can be determined, and focus in an area with good focus detection accuracy can be determined.
Point calculation becomes possible.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a view for explaining an automatic focus detection device, in which (A) is a schematic side view, and (B) is a front view of a pair of photoelectric conversion element rows constituting an image sensor array. . FIG. 2 is an explanatory diagram showing a state of a photoelectric conversion element and an output thereof in a case where a vignetting state occurs. FIG. 3 is a block diagram of an embodiment of the present invention. FIG. 4 is a filter characteristic diagram of a filtering means used in the embodiment. FIG. 5 is a diagram for explaining an effect of a vignetting state on a subject image. FIG. 6 is a diagram for explaining a correlation detection process of a photoelectric conversion element. FIG. 7 is a diagram superimposed to detect a correlation between subject images because the vignetting state has occurred and to detect a correlation therebetween. FIG. 8 is a block diagram showing a specific example of vignetting state detection means (50). [Explanation of References] (10) ... photographing optical system (11) ... photographing lens (12) ... field lens (13) ... minute lens (20) ... focus detection optical system (30) ... image detecting means (33a), ( 33b) Switch means (34) Filter means (40) Focus detection means (50) Vignetting state detection means (60) Detection state control means (70) Auto focus control means

Claims (1)

(57) [Claims] A focus detection optical system that forms a pair of subject images from light beams incident from different regions of the pupil of the photographing optical system; an image detection unit having a pair of image sensor arrays that receives the pair of subject images; and the image detection unit From the relative deviation amount of a pair of outputs from the focus detection means for detecting the focus adjustment state of the imaging optical system, Correlation degree detection means for detecting the degree of correlation between the pair of outputs from the image detection means, According to the degree of correlation detected by the degree of correlation detection means,
Changing means for changing the range of a pair of outputs from the image detecting means.