JP3039066B2 - Focus detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a focus detecting device,
In particular, for a photograph in which an image-forming state of an image-forming optical system is detected based on a relative positional relationship between two object images formed by a distance-measuring optical system using a light beam that has passed through a photographing lens (image-forming optical system). The present invention relates to an improvement of a focus detection device suitable for a camera, a video camera, and the like.

[0002]

2. Description of the Related Art Conventionally, as a focus detection method in a focus detection device, a so-called image shift type focus detection device has been known. In this method, it is assumed that two object images (two images) to be compared are the same images that are merely laterally shifted. For this reason, it is necessary to restrict the aperture ratio of the imaging optical system or to restrict the arrangement of the distance measurement visual field so that the distance measuring light beam does not become vignetting.

It is also known that high-pass filtering of an image signal is effective against deterioration of the image signal due to vignetting of a distance measuring light beam. However, this is mainly intended to remove the adverse effects of minute vignetting of the light beam due to manufacturing errors and the like, so that the above-mentioned drawbacks are eliminated, the restriction of the open F-number is eliminated, and the layout of the distance measuring field can be freely performed. Was not as effective.

[0004]

In the case of a camera system in which such a focus detection device is applied to, for example, a single-lens reflex camera, first, the open F number of the taking lens is generally set to F1.0 to F8. . For this reason, such a camera system is made into auto focus (auto focus)
When trying to do so, the photographing lens that can be used for autofocusing as described above is limited to a lens that is brighter than, for example, F5.6, or the distance measurement field of view is, for example, an optical axis near the optical axis where the distance measurement light flux is hard to vignet. It becomes necessary to divide the layout such that it is laid out at a position within 5 mm from the position. Therefore, it has been very difficult to realize a camera system having an autofocus that operates (distance measurement) in any photographing lens and in any region of the viewfinder field of view.

[0005] Further, the applicant of the present invention has disclosed, for example, Japanese Patent Application No. 2-17 / 1990.
In the improved distance measuring optical system disclosed in Japanese Patent No. 8848, in which a diffuser plate is arranged in the optical path of the distance measuring optical system, the distance measuring characteristic which is a feature of the distance measuring light beam is limited as long as it is not vignetting. Improved accuracy is achieved. However, vignetting is likely to occur due to the diffusion plate, and when this vignetting occurs, it tends to be difficult to measure the distance.

According to the present invention, when focus detection is performed using the image shift method, the aperture ratio of the imaging optical system can be reduced by using an appropriately configured filter means even if there is some vignetting in the distance measuring light beam. It is an object of the present invention to provide a focus detection device capable of saving focus, eliminating restrictions on the arrangement of distance measurement fields, and performing focus detection with high accuracy.

[0007]

According to the present invention, there is provided a focus detecting apparatus comprising: (a-1) a relative position relationship which changes in accordance with an imaging state of an imaging optical system to be subjected to focus detection; A re-imaging optical system that forms a second two object images, and a photoelectric conversion unit that includes a plurality of pixels that respectively output first and second signals corresponding to the first and second object images, The first and second signals are sequentially displaced relative to each other in operation, and the first and second signals are previously calculated when calculating an evaluation amount representing the degree of coincidence between the first and second signals at each relative displacement position. Is subjected to filter processing using a filter for each relative displacement position determined according to the relative displacement position, and the amount of image shift between the first and second object images is determined based on the change in the evaluation amount. Is calculated, and a defocus amount conversion operation is performed on the image shift amount, thereby forming the image forming optical system. A focus detection device for detecting an image state, wherein the defocus amount conversion calculation for the image shift amount further calculates the aperture ratio of the imaging optical system to be focus-detected as one function. I have.

(A-2) First relative position changes in accordance with the imaging state of the imaging optical system to be focused.
And a re-imaging optical system for forming two object images, and a photoelectric conversion unit including a plurality of pixels for outputting first and second signals corresponding to the first and second object images, respectively. , The first
And the second signal are sequentially displaced relative to each other in operation, and when calculating an evaluation amount representing the degree of coincidence between the first and second signals at each relative displacement position, the first and second signals are On the other hand, filter processing is performed using a filter for each relative displacement position determined according to the relative displacement position, and an image shift amount between the first and second object images is obtained based on a change in the evaluation amount, A focus detection device that detects an imaging state of the imaging optical system by performing a defocus amount conversion operation on the image shift amount, wherein the defocus amount conversion operation on the image shift amount further includes: The method is characterized in that the exit pupil position of the imaging optical system to be subjected to focus detection is calculated as one function. (A-3) A re-imaging optical system that forms a pair of object images whose relative positional relationship changes in accordance with the imaging state of the imaging optical system to be focus-detected; Corresponding first
And a plurality of photoelectric conversion means each comprising a plurality of pixels respectively outputting a second signal and a second signal. The first and second signals are sequentially displaced relative to each other in operation, and the first and second signals at each relative displacement position are calculated. When calculating the evaluation amount indicating the degree of coincidence of the two signals, the first and second signals are subjected to filter processing using a filter for each relative displacement position determined in advance according to the relative displacement position, The image shift amount of the first and second object images is obtained based on the change of the evaluation amount, and the defocus amount conversion operation is performed on the image shift amount to change the image forming state of the image forming optical system. A focus detection device for detecting, wherein a defocus amount conversion calculation for the image shift amount is determined by a photoelectric conversion unit including a plurality of pixels from an optical axis of an imaging optical system to be further subjected to focus detection. The distance representing the distance measurement field is defined as one function It is characterized by computing.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a schematic diagram showing the principle of detection of a focus detection system of a focus detection device according to the present invention. The present embodiment is characterized in that an optimum focus position in consideration of a spatial frequency of a subject image is obtained by aligning a photographing light beam and a distance measuring light beam.

First, the configuration will be described. In FIG.
40 is a subject surface, 41 is a photographic lens, and 42 is a mat surface 4 having fine irregularities near a predetermined imaging surface of the photographic lens 41.
A diffusing plate 2a; 43, a field lens (condenser lens); 48, an aperture having two openings 48a, 48b; 44, two lens units disposed symmetrically about the optical axis L2 of the taking lens 41; It is a re-imaging lens (secondary imaging lens) having 44a and 44b. This re-imaging lens 44 converges the light beam that has passed through the two aperture openings 48a and 48b, and the photoelectric conversion element array 4 arranged rearward.
6. A secondary object image of the subject is formed on 6, 47. The photoelectric conversion output obtained here is input to a microcomputer described later, and is used for focus detection of the photographing lens.

The above-mentioned field lens 43 is a diffusion plate 42
If the diffusion in the lens is ignored, it has the effect of projecting the aperture 48 and the exit pupil of the photographic lens 41 in a conjugate relationship, and the diffusion characteristics of the diffusion plate and the condensing characteristics of this field lens cause The passing area of the ranging light beam on the exit pupil is determined.

FIG. 2 is an explanatory diagram showing the diffusion characteristics of the diffusion plate 42. This shows a state in which a light beam L vertically incident from the taking lens 41 side is diffused on the mat surface 42a on the re-imaging lens side. The figure shows that the straight component emitted at the same angle as the incident angle is the strongest, and becomes weaker as it goes away from it.
It is understood that the light rays incident from the taking lens have a certain extent on the stop 48. Therefore, the light flux converging at a certain point on the stop 48 due to the principle of ray regression is
Conversely, it has some extent on the exit pupil.

FIGS. 3 and 4 are explanatory diagrams. FIG. 3 is a plan view of the stop 48, and FIG. 4 is a schematic view showing an image of the stop 48 on the exit pupil of the photographing lens 41. As shown in FIG. 3, the two aperture openings 48a and 48b have a shape obtained by dividing one circle into two. The field lens 43 forms images of the aperture openings 48 a and 48 b on the exit pupil of the photographing lens 41, but the image becomes blurred due to the action of the diffusion plate 42. FIG. 4 schematically shows the state at this time.

In FIG. 4, two aperture images 49a and 49b are provided.
Are represented by diagonally right oblique lines and diagonally left oblique lines. Due to the bleeding, these aperture images partially overlap each other, and as a whole, have a shape that spreads completely inside the exit pupil. As a result, the aperture images 49a, 4
It is possible to capture the distance measurement light beam from a wide area of the exit pupil with the center of gravity of 9b relatively small.

Next, the principle of distance measurement of the focus detecting device having the above configuration will be described.

FIGS. 5A and 5B are explanatory diagrams for this purpose, and show the details from the taking lens 41 to the aperture 48. FIG. For the diffusion of light on the mat surface 42a of the diffusion plate 42, strictly, wave optics analysis is necessary, but here, geometric optics will be described as an approximation.

In FIG. 5A, two rays A and B passing through points G and H in the aperture openings 48a and 48b and a point E on the optical axis of the mat surface 42a are considered. Considering that the condenser lens 43 is thin and is adjacent to the mat surface 42a for simplicity, the angle from the point E to the points G and H on the opening of the diaphragm 48 is θ1. Since the mat surface 42a is a diffusion surface, it is an aggregate of continuous fine irregularities. Assuming that the tangent plane of the mat surface 42a at the point E is the surface indicated by P1 in the drawing,
The light beams A and B are refracted here and become light beams A 'and B'. Points at which the two refracted light beams reach the exit pupil of the photographing lens 41 are denoted by I and J, respectively.

In FIG. 5B, a point F is set very close to a point E on the optical axis of the mat surface 42a, and the behavior of light rays there will be considered. Assume that the tangent plane at the point F is P2 as shown in the figure, and the angle from the point F to the points G and H on the aperture opening is θ2. As in FIG. 5A, two rays C and D passing through points G, H and F on the aperture of the stop 48 are refracted here to become rays C 'and D'. Points at which the two refracted rays reach the exit pupil of the taking lens 41 are denoted by K and K, respectively.
M is assumed.

Ignoring the thickness of the focus plate 42, FIG.
The angles at which the respective arrival points I, J, K, and M on the exit pupil of the photographing lens 41 shown in (A) and (B) are viewed from points E and F are θ1 ′ and θ2 ′. Points E and F are extremely close. Therefore, if both θ1 and θ2 and the inclination of the tangent plane fall within a small range, approximately θ1 ′ ≒ θ2 ′ holds.

This means that, within a range in which the angle of view of the two aperture openings from the point where the optical axis intersects the mat surface 42a is small, the distance measuring base line length is the same for rays refracted within a certain range. FIG. 2 shows an example of the diffusion characteristic due to the matte surface.
There are tangent planes in various directions around, and it can be interpreted as a characteristic obtained as a sum of them. The diffusion characteristic of the matte surface is in a range that is incompatible with the amount of light, and the description of FIG. 5 can be applied to each of these tangent planes.

In addition, in the description of FIG. 5, distance measurement of a point on the optical axis has been described for simplicity, but it is clear that the same description holds for a point outside the optical axis. Accordingly, the distance measuring base line length of the focus detecting device according to the present invention is determined by the distance between the openings of the diaphragm 48, and is not affected by the characteristics of the diffusion plate.

Due to such properties, the focus detection operation is basically the same as that of a conventional focus detection device having no diffuser. For example, when the photographing lens 41 is extended to the left in the drawing to be in a so-called front focus state, the re-imaging lens 44
The subject images (object images) projected on the light receiving surfaces of the respective photoelectric conversion element arrays 46 and 47 by a and 44b are respectively displaced in the directions of the arrows, and the photoelectric conversion elements corresponding to the relative displacement amounts of the object images The change in the output of the columns 46 and 47 indicates that the front focus state and the amount thereof are detected. Also, in the case of the back focus state, since each object image is shifted in the opposite way to the case of the front focus state, it is in the back focus state,
The amount is detected.

As described above, this focus detection device is configured to utilize a diffuser plate so as to capture a light beam having a bright F-number without increasing the distance measurement base line length more than necessary. Therefore, it is possible to detect an optimum focus position according to the spatial frequency of the subject image. Further, as described below, an application in which a focusing plate of a single-lens reflex camera is used as a diffusion plate is also possible.

FIG. 6 is a schematic view of a main part when the focus detection device of the present invention is applied to a single-lens reflex camera.

In the figure, reference numeral 80 denotes a single-lens reflex camera main body; 82, a lens barrel for holding a photographing lens 81 movably in the optical axis direction; and 83, a movable mirror. The movable mirror 83 is in a lowered state when observing the subject, and serves to deflect the light beam from the photographing lens 81 upward and guide it to the finder and the focus detection system according to the present invention. Further, at the time of photographing, the lens 83 is jumped up to a position where the light beam from the photographing lens is not obstructed immediately before the exposure of the photographic film 92 is started, and is returned to the state of FIG.

Next, regarding a finder and a focus detection system,
Reference numeral 84 denotes a focus plate on which a subject image is projected by the photographing lens 81, and has a function of simultaneously diffusing a distance measuring light beam.
85 is a condenser lens (field lens), 86
Denotes a pentaprism, 87 denotes a light splitter that splits a light beam into a focus detection system, and 88 denotes an eyepiece. On the light incident surface of the focus plate 84, there is formed a spherical portion 84a for allowing the distance measuring light beam to enter the mat surface 84c formed on the light exit surface side of the focus slope 84 at an angle nearly perpendicular to the mat surface 84c. A Fresnel lens 84b is formed in a peripheral portion outside the field of view. The mat surface 84c has a slightly convex surface at a portion corresponding to the spherical portion 84a in order to correct the curvature of the planned image forming surface. The light beam diffused by the mat surface 84c is converted into an eyepiece 88 by a condenser lens 85 disposed behind the mat surface 84c.
Is refracted to match the arrangement of Next, the light beam is applied to the pentaprism 86 and the half mirror surface 8 of the light splitter 87.
After passing through the eyepiece 88, it is deflected in the direction of the eyepiece 88 and reaches the eyes of the observer after passing through the eyepiece 88.

The light splitter 87 placed just before the eyepiece 88 reflects a part of the light to reach the eyepiece upward by the half mirror 87a, and uses the reflected light beam for focus detection. Fulfill. The portion below the light shielding mask 89 is a focus detection system, 90 is a secondary imaging lens (re-imaging lens) made of transparent plastic, 93 is a stop, 94 is a light guide prism 108, 108h1, 108h2 is a photoelectric conversion device composed of many pixels. In the pixel row of the element, the pixel row is held by a transparent resin package 95. The aperture 93 is projected onto the exit pupil of the photographing lens 81 by the secondary imaging lens 90, the condenser lens 85, and the spherical portion 84a of the focusing plate 84. The secondary imaging lens 90 is provided on the mat surface 8 of the focusing plate.
It also serves to project 4c onto the photoelectric conversion elements 108h1 and 108b2. Due to the diffusion effect of the mat surface 84c, the projected image of the subject is in a state of being spread and spread as described with reference to FIG.

FIGS. 7 and 8 show the thus constructed focus detection optical system developed along the optical axis. 7 shows a cross section in the direction of the finder-field short side, and FIG. 8 shows a cross section in the long side direction. In the drawing, reference numerals 186 and 187 denote equivalent parallel plane members having the same optical path length as the pentaprism 86 and the optical splitter 87 developed along the optical path, respectively.

The condenser lens 85 acts to project the eyepiece 88 and the exit pupil of the photographing lens 81 into a projection relationship with each other for an observation system, thereby making the entire viewfinder brighter. The aperture 93 and the exit pupil of the photographing lens 81 are similarly brought into a projection relation to obtain a wide distance measurement field of view. This can be achieved by arranging the eyepiece 88 and the stop 93 at optically equivalent positions.

When the two functions are compatible, it is difficult to limit the curvature of both sides of the condenser lens 85 because the distortion of the focus detection system affects the detection accuracy in the projection relationship of the focus detection system. . On the other hand, with respect to the projection relationship of the observation system, it is possible to add a certain restriction to the curvature as long as a predetermined power can be obtained, and to intentionally take an arbitrary value. In order to reduce the size of the camera using this characteristic, the first surface of the condenser lens 85 is formed of two curvature portions.

FIG. 9 is a plan view showing this state, in which the condenser lens 85 is observed from the light incident direction. In the figure, reference numeral 85a denotes a spherical portion having a curvature, and 85b denotes a flat portion. The diameter of the spherical portion 85a is set so as to include a range through which a light beam for distance measurement passes, and the outer flat portion is the result of limiting the curvature due to the convenience of the observation system. By providing a plane portion around the periphery in this way,
The thickness of the condenser lens 85 can be made extremely thin as compared with the case where a curvature is provided on the entire light incident surface. In this case, the boundary between the spherical portion 85a and the flat portion 85b is discontinuous. In the present embodiment, however, the discontinuity of the refractive power that varies depending on the region is solved by devising the shape of the focusing plate 84. FIG. 10A is a plan view of the focus plate 84 viewed from the light incident direction, and FIG. 10B is a sectional view thereof. FIG.
As shown in FIG. 2A, the focus plate 84 is also formed of two regions, namely, a spherical portion 84a and a Fresnel lens portion 84b. The boundary between the two regions formed on the condenser lens 85 and the focusing plate 84 is set to have a size such that they overlap each other when the photographer observes the finder from a reference position on the optical axis through the eyepiece 88. Have been.

FIG. 10B shows the focus plate 84 as described above.
2 shows a cross section of FIG. As can be seen from the two tangent lines V and W in the figure, the Fresnel lens portion 84b is attached to the spherical portion 84a. Therefore, the Fresnel lens unit 84
The refractive power of b is larger than that of the spherical portion 84a, and as a result, the refractive power of the flat portion 85b of the corresponding condenser lens 85 is compensated for the decrease in the refractive power of the spherical portion 85a. That is, in order to satisfy the above-described projection relationship with the eyepiece system in the observation system over the entire field of view, the spherical portion 84a of the focusing plate 84, the spherical portion 85a of the condenser-lens 85, and the Fresnel lens portion 84b of the focusing plate 84, The overall refractive power of the flat portion 85b of the condenser-lens 85 matches when viewed from the exit surface of the condenser-lens 85.

Next, the image forming relationship of the focus detection system will be described again with reference to FIGS. The secondary imaging lens 90 is a kind of biconvex multi-lens composed of a light incident surface integrally formed with two pairs of lens portions and a light exit surface having a curved surface coaxial with the optical axis of the photographing lens. FIG. 11 shows the secondary imaging lens 9.
0 shows the shape as viewed from the light incident direction, and the lens portions 90 above and below the central lens portions 90c and 90d.
a and 90b are arranged. 90a for multi-lens
And 90b, 90c and 90d make a pair, and the relative position changes in accordance with the imaging state of the taking lens 81.
A pair of object images is formed. Each of these multi-lenses is formed of a spherical surface, and the center of the sphere is shifted from the position on the optical axis of the matte surface 84c of the focusing plate 84 through the centers of gravity of the four apertures of the stop 93 by the respective rays. Are set so as to be substantially perpendicularly incident on each multi-lens corresponding to.

The light exit surface 90e of the secondary imaging lens 90 is a spherical surface common to the multi-lenses 90a to 90d, and its optical axis is common to the taking lens 81. The spherical center of the light exit surface 90e is set at a position optically equivalent to the vicinity of the mat surface 84c of the focusing plate 84 which is the object surface with respect to the secondary imaging lens 90. That is, the pentaprism 86 and the light splitter 8
When the optical path length of No. 7 is converted into air, the center of the mat surface 84c substantially matches the spherical center of the light exit surface 90e of the secondary imaging lens. As described above, in the secondary imaging lens 90, four light rays which are emitted from the position on the optical axis of the mat surface 84 c of the focus plate 84 on the light incident surface side and pass through the center of gravity of each aperture of the stop 93 are also incident on the incident side. The light is vertically incident on the multi-lenses 90a to 90d. Therefore, a major feature of the present invention is that the optical system emits the four light beams substantially perpendicularly from the exit surface 90e.

As described above, the secondary imaging lens 90 emits light from the center of the focus plate 84c and passes through the center of gravity of each aperture of the stop 93, that is, the light that becomes the center of light flux passing through each of the multi-lenses 90a to 90d. Photoelectric conversion element 108 without bending
It is configured to lead to. This fact means that even if the wavelength of the light from the object changes variously depending on the object, the change in the interval between the two images forming the pair can be extremely reduced. Therefore, there is almost no detection error corresponding to the color of the object due to the influence of the chromatic aberration of the focus detection system, which has conventionally been a problem.

When the configuration of the optical system as in the present invention is adopted,
In order to increase the distance measurement field of view while keeping the chip area of the photoelectric conversion element 108 small, it is desirable to set the imaging magnification by the secondary imaging lens 90 to about -0.2 to -0.5 times. FIGS. 7 and 8 show examples of about -0.2 times. In general, when such a reduced imaging system is constituted by a single convex lens, the curvature is increased on the light incident surface and the curvature on the light exit surface side is reduced in accordance with the principle of aberration sharing, thereby reducing the aberration and projecting. It is known that a point image to be formed can be reduced. As shown in FIG. 7 and FIG.
The light exit surface 90e is common to the multi-lens on the incident side, and its spherical center can have only a weak curvature due to the optical restriction near the mat surface 84c. On the other hand, the multi-lens on the light incident side having a small diameter compensates for the weak curvature on the exit side and maintains the imaging magnification, so that the curvature becomes strong. The embodiment shown in FIGS. 7 and 8 also follows the principle of aberration sharing in this sense, and can realize a small point image. The fact that a small point image can be realized in the focus detection optical system means that the distance can be measured to a finer pattern, and this is useful for improving the detection performance.

The positions of the light quantity distributions for the two pairs of formed objects and the positions of the pixel rows of the photoelectric conversion elements that receive the light must be matched with an accuracy of several μm or less in order to satisfy the ranging accuracy of the camera. There is. The greatest problem in the production of a system using a secondary imaging lens in which two pairs of lens parts are integrally formed as in the present invention and a photoelectric conversion element in which a corresponding pixel row is arranged on one chip. Are concentrated on the alignment of the optical axis in the secondary imaging lens. The secondary imaging lens in this embodiment solves the above-mentioned problem by configuring the light exit surface with one spherical surface common to two pairs of multi-lens portions.

The secondary imaging lens made of plastic is manufactured by an injection molding method or a compression molding method. The problem at this time is, as described above, the deviation between the light entrance surface and the light exit surface of each secondary imaging lens and the deviation between the respective secondary imaging lenses. In particular, when two pairs of secondary imaging lenses are used as in the present invention, if the secondary imaging lenses are disjoint and their four optical axes are at both the entrance and exit of the mold, the mutual It is very difficult to achieve accuracy.
In particular, the components constituting the light entrance surface and the light exit surface of the mold are mutually rotating components, but the precision is severe, and if there is an error, the image of each secondary imaging lens will be aberration depending on the amount. And undergo complex deformation. As a result, the point images of the secondary imaging lenses that form a paired image are not similar to each other.

The detection of the actual focus state is performed by a system in which a photographing lens 81 passes through a secondary imaging lens 90, and an image of an object to be a subject is formed in pairs on respective photoelectric conversion elements. . Here, for example, the taking lens 81 is shown in FIGS.
, The image of a paired object formed on the light receiving surface of the light conversion element shifts in a direction approaching. The change of the output of the photoelectric conversion element according to the relative shift between the images detects the front focus state and the amount thereof. In the rear focus state, the paired image is shifted in the opposite direction to that in the front focus state, so that the rear focus state and the amount thereof are detected.

As described above, since the detection of the focus state is performed with the interval between the paired images, the inconsistency of the paired images themselves appears as a ranging error. If the light exit surface is a common spherical surface with respect to each light beam as in this embodiment, there is no influence of mutual rotation at all, and it is easy to manufacture.

The light beam transmitted through the secondary imaging lens enters a stop 93 disposed thereafter. The diaphragm 93 is shown in FIGS.
As shown in (2), it is arranged slightly apart from the secondary imaging lens 90, and has two pairs of apertures corresponding to the multi-lens constituting the secondary imaging lens 90. FIG. 12 is a plan view of the stop 93, showing the shape of the opening. Openings 93a and 93b, 9 corresponding to the secondary imaging lenses 90a to 90d
3c and 93d form a pair, and the distance between the aperture centroids is an amount corresponding to the distance measurement base line length. As is apparent from the figure, the distance measurement base line length determined by the openings 93a and 93b is
It is characterized in that it is set longer than the distance measurement base line length determined by the openings 93c and 93d, and this is used to reduce the chip area of the photoelectric conversion element and make the entire system compact. .

Each of the four openings 93a to 93d has a shape surrounded by two arcs. The shape of the outer arc is such that the arc passes through the spherical portion 84 a of the focusing plate 84, the condenser lens 85 and the secondary imaging lens 90, and
Are set so that when projected onto the exit pupil, the image becomes an arc centered on the optical axis of the photographing lens 81. The inner arc is determined by translating the outer arc of the other paired opening to the other, so that the paired openings have the same shape. Accordingly, the shape of the effective diameter of the pair of secondary imaging lens systems on the taking lens 81 coincides as described later, and the light amount distribution is similar even if the object image is defocused in a range where the distance measuring light beam is not vignetting. Sex is not spoiled.

What should be particularly noted here is the shape of the arc outside the stop 93. Secondary imaging lens 90
Is an eccentric system with respect to the taking lens 81, the center of the arc outside the stop 93 is different from the point Q at which the optical axis of the taking lens 81 intersects the stop 93. Referring to FIG. 12, the center of the arc R outside the opening 93a is P, which is eccentric by the length S from the intersection Q with the photographic lens optical axis.
Here, when the outer arc is projected onto the exit pupil of the photographing lens 81 through the spherical portion 90a or the like of the secondary imaging lens 90, the image is decentered so that the image becomes an arc centered on the optical axis of the photographing lens 81. Therefore, the light beam can be captured most efficiently. FIG. 13 is an explanatory diagram of a projected image of the aperture. 1
01 is an exit pupil of the photographing lens 81, and 102a to 102d are projection images of the aperture openings 93a to 93d, respectively. The distance between the paired aperture openings corresponds to the distance measurement base line length. Since the matte surface 84c of the focus plate 84 is in the optical path, the image of the stop 93 is blurred and spreads according to the diffusion characteristics as shown in FIG. The hatching in FIG. 13 shows this bleeding. The position of the stop 93 is located behind the secondary imaging lens 90 in order to improve the uniformity of the point image with respect to the angle of view.

FIG. 14 shows a secondary imaging 90 lens 9.
This is the shape of the light-shielding mask 89 placed immediately before 0. The light-shielding mask 89 has three openings 89a to 89c, and the other portions are light-shielding portions. With this light shielding effect, ghost generated due to a step at the junction of the four multi-lenses 90a to 90d of the secondary imaging lens 90 is prevented beforehand. The focus detection optical system described in the present invention is a pentaprism 8
As described above, the entire length is long due to the interposition of 6. On the other hand, in order to improve the ranging accuracy, the aperture 93
The distance between the centers of gravity must be widened. As a result, the light quantity distributions forming a pair with the object are formed at positions far apart from each other. FIG. 15 is a diagram showing the relationship between the distance measurement field of view and the shooting screen. Distance-measuring visual fields 104a to 104 with respect to the photographing screen 103
When j is arranged in a cross shape as shown in the figure, when this is simply projected by the secondary imaging lens 90, the secondary object image forming a pair is formed at a considerably distant position, and a pixel arrangement of a large space is required. Become.

In the embodiments shown in FIGS. 6, 7 and 8, the light guide prism 94 is used to reduce the chip size of the photoelectric conversion element in consideration of the above problems. FIG.
7 and 18 are a partially enlarged view and a perspective view around the light guide prism 94. FIG. FIG. 17 shows a cross section corresponding to FIG. 7. In this cross section, the light guide prism 94 made of a transparent plastic member has six optical surfaces 94a to 94f. These six surfaces define the optical path of the taking lens 8
1 has the function of folding in the direction of the optical axis and the function of cutting unnecessary light.

In FIG. 17, stop apertures 93a and 93b are provided.
Are incident on the entrance surface 94a of the light guide prism 94. Next, the light flux is reflected twice by the total reflection surfaces 94b and 94c and the total reflection surfaces 94d and 94e, respectively, and then emitted from the light emission surface 94f toward the photoelectric conversion element 107.
On the other hand, as can be seen with reference to FIG.
And 93d pass directly through the two openings 94g and 94h opened in the light guide prism 94 to the photoelectric conversion element 107. The optical distance from the secondary imaging lens 90 to the photoelectric conversion element 107 depends on the light guide prism 9.
The difference in the optical path length occurs depending on whether or not the light passes through the reflection of 4, but since the reflected light path is in the plastic component, it is substantially shorter, and the difference between the two is small. Further, since the curvatures of the two pairs of lens portions of the multi-lens of the secondary imaging lens 90 can be independently selected, it is possible to realize an optimal image formation for the above two types of optical paths.

By using the light guiding prism 94 as described above, the size of the photoelectric conversion element 107 can be reduced efficiently. FIG. 16 shows this state. In the figure, reference numeral 107 denotes a photoelectric conversion element, pixel rows corresponding to the distance measurement fields 104a to 104e are 108a1 to 108e2, and the distance measurement field 104f.
The pixel columns corresponding to .about.104j are 108f1 .about.108j2.
It has become. The meaning of the suffix indicates the correspondence between two images formed through the paired secondary imaging lenses 90 even in the same distance measurement field of view. Here, pixel columns 108a1 to 108e1 and 108a2 to 108a2 to 108e1 corresponding to the distance measurement visual fields 104a to 104e.
108e2, the distance measurement fields 104f to 104
pixel rows 108f1 to 108j1 and 108 corresponding to j
f2 to 108j2 are located, and useless areas are eliminated. Needless to say, there is a cost advantage due to the small size of the photoelectric conversion element itself, but the introduction of the light guide prism 94 is also effective for miniaturizing the camera body itself by folding the optical path.

In particular, further miniaturization is achieved by arranging the pixel rows of the distance measurement field of view having a long distance measurement base line inside and the pixel rows of the distance measurement field of view having a short distance measurement base line outside. This is, of course, extremely advantageous in terms of cost, and is also effective for downsizing the camera.

The focus detection device described above is particularly effective for a bright lens having a small F-number (aperture ratio) requiring strict focusing. However, in a system such as a single-lens reflex camera in which the photographing lens can be exchanged, a dark photographing lens having a large F-number may be attached, and in such a case, the ranging light beam is vignetted by the photographing lens. Next, a focus detection processing method corresponding to various photographing lenses will be described.

First, FIG. 19 is a perspective view for explaining the vignetting of the distance measuring light beam. FIG. 16 shows only a distance measuring light beam to be incident on 104j in the distance measuring field of view shown in FIG. In this figure, 150a is an example of the outermost shape of the aperture images 102a and 102b shown in FIG. 13, and 151 and 152 are examples of the exit window of the photographing lens which is relatively dark.

In general, a photographic lens causes a reduction in pupil area called vignetting at a position other than the center of a photographing screen. This is because, when viewed from a certain point on the image plane, there are two or more surfaces that determine the width of the light beam in the fully opened state, and this is called an exit window. For example, this surface will be the frontmost lens and the rearmost lens. Two emission windows 151, 152 shown in the figure
Is a general example of this. However, from the viewpoint of the vignetting condition of the distance measuring light beam, these may be represented by one intermediate exit pupil, as described later.

Now, consider a state in which light reaches the distance measurement field 104j through two circles arranged at a certain distance in space as shown in the figure. Here, let us consider the passing range of the photographing light beam on the surface on which the image of the aperture opening of the focus detection optical system is formed so that the passing range of the distance measuring light beam can be fixed and handled for other distance measurement visual field positions. The light actually passes only through the region 153 shown by the broken line. Therefore, the fact that there is an image of the aperture opening outside this also means that vignetting has occurred in the distance measuring light beam. In the figure, only the distance measuring light beam of the distance measuring field 104j is shown. However, if the exit pupil is projected in the same way for other distance measuring fields, vignetting can be known. Shows a tendency that vignetting is more likely to occur.

By the way, if the defocus of the image formed by the photographing lens and the re-imaging lens increases, the point image on the photoelectric conversion element becomes closer to the shape of an aperture, and thus the distance measuring light beam is vignetted. When the point image is present, the degree of blurring of the point image is left-right asymmetric. In addition, the blurring of the secondary object image forming a pair has an inverted relationship, and as a result, the similarity between the two images is considerably reduced. As described above, when calculating the defocus amount, the relative distance between a pair of secondary object images is obtained, and the calculation accuracy for such dissimilar images is extremely poor. For this reason, open F
For a dark photographing lens having a large number, a means for correcting an asymmetrically blurred image so as not to lower the accuracy of the image shift calculation is required. Hereinafter, this correcting means will be described in detail.

First, FIGS. 20A to 20C and FIG.
(A) to (C) are diagrams for explaining how the vignetting degree of the distance measuring light beam changes depending on the difference in the characteristics of the photographing lens. FIG. 20 shows 104j in the distance measuring field shown in FIG.
FIG. 21 is a plan view of an optical path of the distance measuring optical system viewed from above, and FIG. 21 is a diagram illustrating a range through which a photographing light beam passes on a surface of the distance measuring optical system on which a stop image is formed.

In FIG. 20, an alternate long and short dash line 169 is the optical axis of the photographing lens, a broken line 156 is a projection surface of the aperture opening image, and a broken line 1 is shown.
Reference numeral 55 denotes a distance measuring light beam, which is common to FIGS. 20A to 20C. Reference numerals 157, 158, and 159 denote examples of exit pupils of the photographing lens, which represent the two exit windows shown in FIG. 19 at intermediate positions thereof. FIG. 20A shows a case where the image is relatively close to the shooting screen.
FIG. 20B shows a case where the image coincides with the surface 156 of the stop image.
(C) shows the case where it is relatively far from the shooting screen. The light beams passing through these exit pupils are solid lines indicated by 160 to 168 in the figure, the innermost one is an F-number (aperture ratio) 5.6, the middle one is an F-number 4, and the outermost one is an F-number 2.8. . In the figure, the vertical direction is enlarged for easy understanding. On the stop image plane 156, the distance measuring light beam 155
By examining the common region between the light beam that has passed through the exit pupil and the above-mentioned light beam, it is possible to know the vignetting condition of the distance measuring light beam.

However, as described above, the intensity distribution on the pupil plane is not uniform because the distance measuring light beam passes through the diffusion plate. Therefore, in order to know the vignetting condition in more detail including this intensity distribution, attention is paid to the projection surface of the aperture opening image as in FIG. 19, and the relationship between the distance measuring light beam and the photographing light beam on this surface is shown. FIG. 21 shows the result. (A) to (C) in the figure
20A to 20C correspond to FIGS. 20A to 20C. FIG. 21A shows a case where the exit pupil position is relatively close to the shooting screen, and FIG. 21 (C) shows a case where it is relatively far from the shooting screen.

In the figure, a point 180 is a position corresponding to the optical axis 169 of the photographing lens, and a point 181 represents only one of a pair of distance measuring light beams for simplicity. The size represents the effective light flux density. Circles 183 to 191 drawn on the distance measuring light beam 181 are exit pupils of the photographing lens. The innermost part is an F number 5.6, the middle part is an F number 4, and the outermost part is an F number 2.8. The actual distance measuring light beam is also distributed at a position symmetrical with respect to the axis 182, and the vignetting of the distance measuring light beam is symmetrical with respect to the axis 182.

According to FIGS. 20 and 21, it is understood that the distance measuring light beam is less likely to be vignetted as the photographing light beam is brighter and the projection surface 156 of the aperture opening image is closer to the exit pupil surface.

FIG. 22 shows a state where the distance measurement field of view is changed. Here, the light beam incident on the distance measurement field 104i shown in FIG. 15 will be described.

In FIG. 20, a chain line 169 indicates the optical axis of the photographing lens, and a broken line 156 indicates the projection plane of the aperture opening image.
Is the same as A broken line 200 is a distance measuring light beam in the distance measuring visual field, and light beams passing through the exit pupils 157, 158, and 159 of the photographing lens are indicated by solid lines 201 to 209, respectively. The innermost one is an F number 5.6, the middle one is an F number 4, and the outermost one is an F number 2.8.

As is clear from a comparison between FIG. 20 and FIG. 22, an image of the aperture is formed as the diameter of the exit pupil becomes smaller and the exit pupil position becomes smaller, regardless of the position of the distance measurement visual field. The greater the distance from the surface, the greater the vignetting ratio of the distance measuring light beam, but the smaller the distance, the closer the position of the distance measuring field to the optical axis 169 of the photographing lens.

Because of the vignetting of the distance measuring light beam as described above, a pair of images on the photoelectric conversion element has a very characteristic shape according to the image forming state, and the shapes are the exit pupil diameter, the exit pupil position, and the like. It is a function of the ranging field position and, therefore, the aberration of the re-imaging lens. The shape of the blurred image causes inconvenience to the conventional distance measurement calculation, and some correction means is required to maintain the distance measurement accuracy. Here, the reason that the aberration of the photographing lens is not taken into consideration is that since the re-imaging lens is generally constituted by a single lens, the aberration occurring there is much larger than the aberration occurring in the photographing lens. It is.

Next, the principle of correcting a blurred image in this embodiment will be described. Generally, an object image by an optical system is given as a convolution integral of a point image determined by characteristics of the optical system and a luminance distribution of the object. The point image corresponds to the impulse response of the optical system.If an adaptive inverse filter is created based on the transfer function that is the Fourier transform of the impulse response and applied to the point image, the point image is converted to the original point. Can be restored. If an inverse filter is applied to a general object image, the luminance distribution of the original object can be obtained. If this is applied to the focus detection optical system described here, the subject image formed on the photoelectric conversion element is a convolution integral of the point image and the luminance distribution of the subject, and furthermore, an appropriate inverse filter is used. That is, it is possible to return to the subject luminance distribution.

However, since the focus detection is performed by checking the relative image interval, the photoelectric conversion element is a one-dimensional line sensor. Therefore, the correction of the image data performed on the output of the sensor may be considered as the correction of the line image obtained by integrating the point image described above in the direction orthogonal to the pixel row. In addition, the correction of the blurred image required here does not necessarily require "restoration" to return the line image to a complete line, and the left-right asymmetric line image caused by the vignetting of the ranging light beam is at least symmetrically shaped. , The relative distance between the two images forming the pair can be obtained.

Next, a more detailed description will be given with reference to the drawings.
FIGS. 23 and 24 show a blurred image and its correction filter.
(A), (B), and (C) are a correction filter, image data, and corrected image data, respectively. A pair of secondary object images formed by the focus detection optical system are respectively OA
(I), OB (i), and modified image data A (i), B
(I). Since the distance measuring light beam is taken into the re-imaging optical system through the diffusion plate, the opening F of the photographing lens is increased.
Vignetting occurs depending on the number (diameter ratio), and as a result, a left-right asymmetric line image is shown in FIG. The vignetting of the two distance measuring light beams is symmetrical with respect to the optical axis of the photographing lens, so that their line images have a flipped relationship. Further, the line image has a certain extent as shown in the figure when the imaging lens is defocused, and the distance between the two images is different from that in the focused state. Image O in FIG.
It is for this reason that A (i) is drawn to the left and image OB (i) of FIG. 24 (B) is drawn to the right. Furthermore, compared to when the distance measuring beam is not vignetting,
The distance between the two images is narrow, and this must be taken into account in the calculation of the image shift sensitivity described later.

FIG. 10A shows a filter for correcting this line image. In the figure, FOF is the origin of the horizontal axis of the correction filter, and is also used for inversion of the correction filter, which will be described later. That is, the arithmetic means includes a C1 processing for performing filter processing on the first signal. Process and the second
And C2 processing for performing filter processing on the signal of the second processing. The filter used in the second processing has a characteristic obtained by inverting the filter used in the first processing. It is that you are.

AXX indicates the origin of the value of the filter component. As described above, since the two line images are in an inverted relationship, the filter shown in FIG.
The filter of (A) also has an inverted relationship. Therefore, only one of the shapes needs to be stored in the camera, and the original filter information described later can be used efficiently. The shape of the correction filter is designed to remove the asymmetry of the line image and to prevent the corrected image from coming too close to the original line. This is because a correction filter that returns to a position close to a line needs to have a characteristic of amplifying a high frequency extremely greatly, and is particularly susceptible to high-frequency components among random noise due to thermal noise of a photoelectric conversion element. That's why. If the high-frequency noise superimposed on the original image data is excessively enlarged, the correlation between the two images decreases, and the distance measurement accuracy decreases. The shape of this filter will be described with reference to FIG. 23. The right side of the filter mainly functions to tighten the hem pulled to the right side of the image OA (i), and the left side of the filter mainly functions to further extend the short hem on the left side. In order to facilitate the following description, this restoring section is a ZR section (for example, a portion on the right side of the FOF in FIG. 23A), and a diffusion section is a ZS section (for example, FIG.
(A) (left part of FOF). Z
By forming the S portion with only the positive component and the ZR portion with the positive and negative components, the left and right hem are equally drawn without restoring the image to a line as described above.

That is, the filter has only the positive component on one side and the positive and negative components on the other side with respect to the origin.

In order to perform image correction, FIGS.
Convolution integration of the line image shown in (B) and the correction filter shown in (A) is performed. The result is the corrected image data shown in (C). As shown in the figure, since each of the corrected image data is symmetrical, the relative distance between the images shown in FIGS. 23C and 24C can be calculated with high accuracy.

In the above, the correction of the line image has been described first. Next, correction of a general subject image, which is a convolution integral of a line image and a subject luminance distribution, will be described. FIG. 25 is an explanatory view showing a state in which white-black edges are selected as subjects and placed at the center of the distance measurement field of view. FIG. 7A is a diagram in which image data of a pair of secondary images is superimposed, and FIG. 7B is the corrected image data. Image data drawn with solid lines Figure 23
The image data drawn by the broken line as the superposition of the point images of FIG. 24B is formed as the superposition of the point images of FIG. Therefore, since the shapes of the point images are different, the shapes of the two images are naturally different.

FIG. (C) shows the result of applying a correction filter to such image data. The correction data shown in FIG. 23A is applied to the solid line image data, and the correction filter shown in FIG. 24B is applied to the broken line image data. As a result, the shapes of the two images become the same, and the distance between the images can be calculated accurately.

When the defocus of the photographing lens is small, image data as shown in FIG. 26 is obtained. Since the size of the point image is small, the image is sharp, and the relative distance between the two images is different from that in FIG. In this case, since the size of the blurred image on the primary imaging plane of the photographing lens changes almost in proportion to the defocus, by changing the abscissa magnification of the correction filter according to the defocus, the image shown in FIG. As in the case, the image can be corrected. Specifically, FIG.
If the filters of FIGS. 3 (A) and 24 (A) are compressed in the horizontal axis direction and used, a corrected image as shown in FIG. 26 (B) can be obtained.
This will be described again later in the description of the flowchart.

FIG. 27 is a block diagram showing an example of a camera focus detection device suitable for carrying out the present invention.

In the figure, PRS is a central processing circuit of the camera, which is a one-chip microcomputer in which a CPU, a RAM, a ROM, an ADC (A / D converter), an input / output port and the like are arranged. A series of camera control software and parameters including AF control are stored. The PRS forms a part of the calculation means.

DBUS is a data bus, and SHT receives data input via the data bus DBUS while the control signal CSHT is being input from the central processing circuit PRS. A shutter control circuit for controlling the movement of the curtain, an aperture control circuit for receiving data input via the data bus DBUS while the control signal CAPR is being input, and controlling an aperture control mechanism (not shown) based on the data. , SWS is a release switch, a continuous shooting mode switch, a switch for setting various information, and a focus detection that considers only a substantially central portion of a screen described later or a focus detection that considers both a substantially central portion and a peripheral portion of the screen. Is a group of switches such as a switch for selecting. The LCOM receives data input via the data bus DBUS while the control signal CLCOM is input, and drives the lens in synchronization with the clock signal LCK by a lens communication circuit that performs serial communication with the lens control circuit LNSU based on the data. Data DCL is transmitted to the lens control circuit, and at the same time, lens information DLC is serially input.

BSY is a signal for informing the camera that the focus adjustment lens (not shown) is moving. When this signal is generated, the serial communication is not performed. SPC is a photometric circuit, and when receiving a control signal CSPC from the control circuit, sends a photometric output SSPC to the central processing circuit PRS. The photometric output SSPC is the central processing circuit P
A / D conversion is performed by an ADC in the RS, and is used as data for controlling the above-described shutter control circuit SHT and aperture control circuit APR.

Reference numeral 220 denotes a light projecting circuit for projecting auxiliary light for focus detection, and a control signal AC from the central processing circuit PRS.
The LED is driven by T and the synchronous clock CK to emit light.

The SNS includes a pair of line sensors 108a1,
.. 108j1, 108j2, each of which is configured to receive an image at a position corresponding to each of the detection visual fields 104a to 104j on the screen. SDR is a sensor drive circuit that controls each of the light receiving circuits 1 to 5 according to signals STR and CK input from the central processing circuit PRS.
1, φ2, CL, SH to control the light receiving circuits 1 to 5,
One of the light receiving circuits 1 to 5 is selected by the selection signals SEL1 to SEL5, and the image signal SSNS obtained from the selected light receiving circuit is transmitted to the central processing circuit PRS.

FIG. 28 shows the manner in which the lens control circuit LNSU obtains the focal length information and the distance ring information of the lens. The position of the distance ring and the position of the zoom ring with the brush and the defocus amount of the focus adjustment lens are shown. The coefficients and the like are converted into 5-bit signals, and the CPs in the lens control circuit LNSU are converted.
U is calculated, and is input to the central processing circuit PRS via the lens communication circuit LCOM as lens information DLC. Note that the position information of the distance ring in this case is not used for the direct focusing operation, so that a high degree of accuracy is not required.

The operation of the camera having the above configuration will be described with reference to the flowchart of FIG.

When a power switch (not shown) is turned on, power supply to the central processing circuit PRS is started, and PRS is stored in the ROM.
The execution of the sequence program stored in is started.

FIG. 29 is a flowchart showing the entire flow of the above program. When the execution of the program is started by the above operation, the program proceeds to step 00 through step 001.
In step 2, the state of sw1 which is turned on by pressing the release button at the first stage is detected, and when sw1 is turned off, the process proceeds to step 003 to clear all control flags and variables set in the RAM in the PRS. And initialize. The above steps 002 and 003 are repeatedly executed until the sw1 is turned on or the power switch is turned off. When sw1 turns on, step 00
2 to step 011. In step 011, an interrupt operation of a timer interrupt described later is permitted. As a result, the timer interrupt operation is periodically executed during the steps after step 012, and various state detections are performed.

At step 012, a "photometry" subroutine for exposure control is executed. PRS is a photometric sensor S
The output SSPC of the PC is input to the analog input terminal, and A /
D-conversion is performed to calculate optimal shutter control values and aperture control values from the digital photometric values, and store them at predetermined addresses in the RAM. During the release operation, the shutter and the aperture are controlled based on these values.

The following step 014 is a subroutine for inputting an image signal and calculating a defocus amount. After the PRS inputs an image signal from the sensor device SNS, the PRS performs a filtering process on the image signal to obtain an image shift. A predetermined calculation based on the method is performed to output a defocus amount of the photographing lens. The specific operation of this step will be described later in detail.

Step 015 is a subroutine for driving the lens based on the detected defocus amount DEF, the details of which will be described later.

At step 016, the timer interrupt permitted at step 011 is prohibited.
In step 019, no interruption is caused. Step 01
7 is turned on by pressing the release button at the second stage.
Is detected, the process proceeds to step 018 if it is off, and proceeds to step 019 if it is on. Step 01
The AFFLG of 8,019 corresponds to steps 012 to 015 described above.
Is a flag that specifies whether the AF has been completed in a state where only sw1 is on, or has been completed in a state where sw2 is also on. Therefore, when sw2 is off, that is, when only sw1 is on and AF is completed, AFFLG is set to “0”.
Is stored, and when sw2 is on and AF is completed, “1” is stored in AFFLG. Here, first, it is assumed that sw2 is off and AFFLG is “0”.

The above steps 002 and 011-018 are AF operations in which only sw1 is in the ON state. Next, a timer interrupt operation that enters steps 012-015 in the flow will be described.

The timer interrupt at step 021 is an interrupt function for periodically monitoring sw2, the charge / feed state, and the AF history to detect whether or not to shift to the release operation. For example, 5
An interrupt is performed every msec.

Hereinafter, each step will be described.

Step 022 via step 21
Then, the state of sw2 is determined, and if it is off, an interrupt return is made in step 023. That is, only sw1 is on and A
During the repetition of F, even if the interruption is performed every 5 msec, the operation immediately returns because sw2 is off.

If sw2 is on in step 022, the flow advances to step 024 to determine the switch SWCH.
When the switch SWCH is off, the charging / feeding operation has already been performed after the release has been performed, so that the release cannot be performed yet. Therefore, the process returns in step 023 to continue the AF operation and the charging operation.

If the switch SWCH is on in step 024, the charging and feeding have been completed, and
A charge / feed stop signal is issued at and MTR1 is stopped.
Move to step 026.

At step 026, the flag AFFLG is determined. If AFFLG = 1, since the AF has been completed at least once after sw2 is turned on, the flow shifts to the release operation of step 027. AFFLG # 1 or "0"
In this case, AF has not been completed once since the switch SW2 is turned on, so that the process returns in step 023, and the release operation is not performed.

Subsequently, in step 028, AFFLG is reset to "0", and charging / feeding is started in step 029 (this is performed by rotating the MTR1).
Step 002 while starting charging and feeding
Then, the next focus detection operation is started from step 011.

FIG. 30 shows a flowchart of the "AF" subroutine.

First, in step 402, a full-field detection mode in which focus adjustment is performed in accordance with the distance measurement information of all distance measurement fields, or a specific field detection mode in which focus adjustment is performed only in the distance measurement field at the center of the screen Is determined. If the mode is the full-field detection mode, the process proceeds to step 403. If the mode is the specific-field detection mode, the process proceeds to step 405.
The setting of the detection mode is performed by the photographer using the switch SWS shown in FIG.

In step 403, the SPF is reset, and in step 404, all distance measurement fields are set as detection distance measurement fields. Also, in step 405, SP
F is set to 1, and in subsequent step 406, the distance measuring fields 104c and 104h shown in FIG. 15 are set as the detected distance measuring fields.

At step 407 following steps 404 and 406, the sensor driving circuit S shown in FIG.
The line sensor SNS is driven via the DR, and the output analog image signal of the detected distance measurement field of view is, after A / D conversion,
It is stored at a predetermined address in the RAM.

In step 408, RA in the previous step
Defocus detection is performed based on the image signal stored in M. Details of this subroutine will be described later.

In step 409, it is checked whether or not defocus detection is impossible for all the set distance measurement fields.
If so, the flow shifts to step 410 to perform focus detection disable processing such as search operation and display of detection failure. If any of the distance measurement fields can detect the focus, the process proceeds to step 412.

In the following step 412, step 403
The state of the SPF set in steps 405 and 405 is checked. If the state is zero, the process proceeds to step 413;
Move to 5.

In step 413, the distance measuring visual field in which the closest object has been detected is selected from the distance measuring visual fields that could be detected. In step 414, the defocus amount of this closest point is stored in DEF. I do. Step 41
In step 5, the average defocus of the detectable ranging field of view is calculated, the average defocus is stored in DEF in step 416, and the subroutine is returned in step 418.

FIG. 31 shows the "focus detection" subroutine.

In the defocus detection subroutine, first, it is checked whether or not the defocus calculation has been completed for all the distance measurement fields set as the detected distance measurement fields in step 502, and if it has been completed, the flow proceeds to step 503. And return to the subroutine. If there is a range-finding field for which defocus calculation has not been completed, the process proceeds to step 504.

In this step 504, a new distance measurement field of view is set as a target of the defocus calculation, and the subsequent step 50
In step 5, the defocus amount of the target distance measurement field of view set in the previous step is calculated, and the flow proceeds to step 502. Details of the defocus calculation will be described later.

FIG. 32 shows a flowchart of the "lens drive" subroutine.

When this subroutine is executed, FIG.
In step 202, two data "S" and "PTH" are input by communicating with the lens. “S” is a “coefficient of the defocus amount versus the focus adjustment lens extension amount” unique to the photographing lens. For example, in the case of a single lens of the entire extension type, “S”
= 1 ”, and in the case of a zoom lens,“ S ”changes depending on each zoom position. "PTH" is an encoder ENCF linked to the movement of the focusing lens LNS in the optical axis direction.
Is the amount of extension of the focusing lens per one output pulse.

Therefore, the defocus amount DE to be adjusted for focus
The value obtained by converting the extension amount of the focusing lens by F, S, and PTH into the number of output pulses of the encoder ENCF, so-called lens drive amount FP, is given by the following equation.

FP = DEF × S / PTH Step 203 executes the above equation as it is.

In step 204, the FP obtained in step 203 is sent to the lens FLNF to command the driving of the focus adjustment lens (or the entire photographing lens in the case of the whole extension type single lens).

In the next step 205, it is communicated with the lens FLNS to detect whether or not the driving of the lens driving amount FP commanded in the step 204 has been completed. Return the subroutine.

FIG. 33 shows the flow of the "release" subroutine.

In step 302, the click return mirror of the camera is raised. This is executed by controlling a mirror driving motor (not shown) via the motor control circuit shown in FIG.

In the next step 303, the previous step 01
The aperture control value already stored in the "photometry" subroutine of No. 2 is sent to the lens FLNS, and the lens FLNS performs aperture control.

In step 304, the previous step 302,
It is detected whether or not the mirror up and the aperture control in 303 have already been completed. The mirror-up can be confirmed by a detection switch (not shown) attached to the mirror, and the aperture control is performed by communication to determine whether the lens FLNS has been driven to a predetermined aperture value.

If any of them has not been completed, the process waits at step 304 and continues to detect the state. When both controls are confirmed, the process proceeds to step 305. At this point, the preparation for exposure is complete.

In step 305, shutter control is performed using the shutter control value already stored in the "photometry" subroutine in step 012, and the film is exposed.

When the shutter control is completed, in the next step 306, a command is sent to the lens FLNS to open the aperture, and subsequently, in step 307, the mirror is lowered. The mirror-down operation is performed by controlling a mirror driving motor (not shown), similarly to the mirror-up operation.

In the next step 308, similar to step 304, the process waits until the mirror down and aperture opening control are completed. When both the mirror down and the aperture opening control are completed, the process proceeds to step 309 and returns.

The above flow is outlined again in FIG.

First, when only sw1 is on, step 00 is executed.
Steps 2,011-017,018 are repeatedly executed. While steps 012 to 015 are being executed, an interrupt after step 021 is performed every 5 msec.
In step 22, sw2 is determined to be off, and the process immediately returns.

When sw2 is turned on during the above cycle,
Step 02 with the first timer interrupt after sw2 on
It is determined to be on at 2 and the procedure goes to step 024. Since this flow is assumed to be the first release, the charge / feed is kept in a completed state as a matter of course.
24, it is determined that the switch SWCH is on, and
5, the execution of step 025 is meaningless, and the operation goes to step 026. The flag AFFLG is “0” when the process reaches step 026 for the first time, and the process returns in step 023. Then, the AF and the timer interrupt are alternately continued, and when the AF is completed, the process proceeds from step 017 to step 019, and AFFLG “1”
Is stored and the process returns to step 002. And step 0
At the first timer interrupt immediately after moving from step 11 to step 012, the processing reaches steps 022 to 026. At step 026, it is determined that AFFLG = 1 and step 02
It is allowed to go to release 7.

After the release operation is completed, AFFLG is set to "0".
Then, the process returns to step 023 while starting charging / feeding in step 026, and moves from step 011 to 012 to start the next AF.

In a timer interrupt immediately after the on state of sw2 is continued, the switch SWCH is turned off and the AFF
Since LG is “0”, AF is repeated. When the switch SWCH is turned on first, the charging / feeding is stopped, and it is waited that the flag AFFLG is "1", that is, AF is completed once, and then the release operation is performed after the AFFLG becomes "1".

On the other hand, if the flag AFFLG is set to "1" first, the AF is repeated until the switch SWCH is turned on, and when the switch SWCH is turned on, the film feeding motor (not shown) is turned on. Is stopped and the release operation is started.

FIGS. 34 and 35 are flowcharts of the "defocus calculation" subroutine. First, the correlation calculation used in this subroutine will be described, and then the details of the processing will be described.

A signal processing method for detecting an image shift amount from an image signal output from a pair of sensor arrays is disclosed in
Japanese Patent Application Laid-Open No. 142306/1984, 107313/1985, and 60-101513 / 1985. Specifically, the number of pixels forming the sensor array is N, and the image signals from the i-th (i = 0,..., N−1) sensor array are A (i) and B (i). When

[0128]

(Equation 1) Or

[0129]

(Equation 2) Is calculated for k ₁ ≦ k ≦ k ₂ . Note that M is the number of calculation pixels represented by (M = N− | k | −1),
K is called a relative displacement, and k ₁ and k ₂ are usually −N /
2, N / 2 in many cases. Where max @ a,
The operator b｝ indicates that the larger one of a and b is extracted, and the operator min {a, b} indicates that the smaller one of a and b is extracted. Therefore, (1),
Page X ₁ (k), X ₂ (k), Y _{1 in} equation (2)
(K) and Y ₂ (k) can be considered as correlation amounts in a broad sense. Looking further at the above equations (1) and (2), X ₁
(K) and Y ₁ (k) are actually the correlation amounts according to the above definitions in the (k−1) displacement, X ₂ (k) and Y ₂
(K) represents the correlation amount at the displacement of (k + 1). Therefore, the evaluation amount X (k) which is the difference between X ₁ (k) and X ₂ (k) is equal to the image signal A at the relative displacement amount k.
(I) means the amount of change in the correlation amount of B (i).

As is clear from the above definition, the correlation amounts X ₁ (k) and X ₂ (k) become minimum when the correlation between the two images is the highest. Therefore, the change amount X (k) should be “0” when the correlation is the highest, and the slope should be negative. However, since X (k) is discrete data, in practice, X (kp) ≧ 0, X (kp + 1) <0 (3) and a relative displacement section [kp] where X (kp) − (kp + 1) is maximum・ Kp + 1], there is a peak of the correlation amount,

[0131]

(Equation 3) By performing the interpolation calculation, the image shift amount PR of a pixel unit or less can be detected.

On the other hand, the correlation amount of Y ₁ (k) and Y ₂ (k) is X ₁ when the correlation between the two images is the highest according to the above definition.
(K) and X ₂ (k), which are maximum. Therefore, the change amount Y (k) should be “0” when the correlation is the highest, and the slope should be positive. Y (k) is the same as X (k) when Y (kp) ≦ 0, Y (kp + 1)> 0 (6) and when Y (kp) −Y (kp + 1) is maximum

[0133]

(Equation 4) By performing the interpolation calculation, the image shift amount PR of a pixel unit or less can be detected.

The image shift amount can be detected by using any of the focus evaluation amounts X (k) and Y (k). As can be seen from JP-A-60-101513, | X (K
p) -X (kp + 1) |> | Y (kp + 1) -Y (k
p) |, the focus evaluation amount X (k) is given by | X (kp)-
When X (kp + 1) |> | Y (kp + 1) -Y (kp) |, it is better in terms of S / N to obtain the image shift amount PR using the focus evaluation amount Y (k).

Now, description will return to the flowcharts of FIGS. 34 and 35. Step 100 shows the A1 process in this embodiment. In steps 100 to 105,
The focus evaluation amount X (k) when the relative displacement amount k is changed within the range of “−20 to 20” is obtained. Where the relative displacement k
Is in the range of −20 to 20, because the number of pixels in the sensor row is assumed to be “40”. However, this processing target pixel range is variable according to the focal length of the photographing lens used. Is also good.

First, in step 150, the image is corrected in order to eliminate the above-mentioned vignetting of the distance measuring light beam. This "image correction" subroutine will be described in detail later,
The image data after image correction recovers the similarity of the pair of images,
The accuracy of the correlation calculation described here is extremely high.

In step 101, the number M of operation pixels is calculated by the equation M = 39− | k |. The number of calculation pixels M is variable according to the relative displacement amount k, and decreases as the absolute value of k increases. This is because the output of the corresponding sensor drops from the end as the relative displacement amount k increases.
In step 102, the sign (positive or negative) of the relative displacement k is checked, and then, according to the sign, the head pixel positions PA and PB at which the calculation of the A and B images is started are determined in step 103 or 1
Calculate with 04. In step 105, the focus evaluation amount X
The calculation of (k) is performed.

Here, the processing steps in steps 100 to 105 will be described with reference to FIG. FIG. 36A shows two image signals A (i) and B (i). FIG.
(B) shows the correspondence between the sensor rows in the correlation operation at k = −20. At this time, M = 39− | 20 | = 19.
Where PA is “0” and PB is “20”.

That is, since the relative displacement amount k is a negative value, the B image is displaced relatively to the left by k pixels (−20 pixels). The first term of the equation for calculating X (−20) is that the B image is further displaced to the left by one pixel from this correspondence, and the corresponding relation is that the A image is displaced to the left by one pixel. Computed by the above corresponds to the first term.

In the first and second terms, the A and B images are displaced one pixel at a time to the left and are operated. Therefore, when calculating the number of operation pixels M, M = 40− | k | And M = 39− |
k |. FIG. 36C shows the correspondence when k = 0. FIG. 36 (D) shows the correspondence of the correlation operation at k = 20. Contrary to FIG. 36 (B), image A is displaced to the left by 20 pixels.

The focus evaluation amount X calculated as described above
FIG. 37 shows an example in which (k) is plotted.

Returning to the flow of FIGS. 34 and 35, in step 110, the peak value kp of the image shift amount of each of the two images A and B is detected from the focus evaluation amount X (k). Hereinafter, in steps 120 to 146, the image shift amount PR of less than a pixel unit is obtained. On the basis of the peak value kp obtained in step 110, two focus evaluation amounts X (k), Recalculate Y (k). There are two reasons for this. One is that in the previous step of obtaining kp, the number of calculation pixels M is variable according to the relative displacement amount k, and the focus evaluation amount X (k) calculated in such a variable calculation range.
Is calculated to obtain the image shift amount PR of a pixel unit or less, there is a possibility that an error due to the mismatch of the number of calculated pixels M may be included. Another method is to use the combination of X (k) and Y (k) rather than to calculate the image shift amount PR using only the focus evaluation amount X (k) in the earlier application (Japanese Patent Laid-Open No. 60-101513). As described above, depending on the signal pattern of the subject, S
This is because they are superior in / N. From the above, in steps 120 to 135, the number of calculation pixels M is made constant based on kp (step 120), and the focus evaluation amounts X (k), Y
(K) is required at the same time.

Note that this image shift operation in which the number M of operation pixels is constant is referred to as a re-correlation operation.

As will be described in detail in the "image correction" subroutine, the image correction filter used in step 150 is determined as a function of the relative displacement k.
However, since the purpose of this re-correlation calculation is to obtain an image shift amount of a pixel unit or less, it is necessary to set an image correction filter more strictly. The following steps 151-1
Reference numeral 53 denotes a process for obtaining a relative displacement amount k for producing an image correction filter with a resolution of a pixel unit or less.

In step 151, kz, z ₁ and z ₂ are set from kp obtained in the previous step.
ZD = | z ₁ −z ₂ |, PR ′ = kz + |
z ₁ / ZD |

This means that the A2th process is performed using a fixed filter regardless of the relative displacement. Thus, the image shift amount PR 'is obtained.

Now, at step 120, M = 38− | k
Calculation of p | is performed to determine the value of the number M of operation pixels. Next, in steps 130 to 135, the focus evaluation amounts X (k) and Y (k) are calculated in the same manner as above at three points of k = kp-1, kp, and kp + 1 with the previously obtained kp as the center. When calculating the number M of operation pixels, M = 38− | kp |
This is because the absolute value of the three points k = kp-1, kp, and kp + 1 is fixed to the number M of operation pixels at the maximum relative displacement.

Step 130 corresponds to the A2th process.

In step 154, an image correction filter is prepared by using the previously described PR ', and the image is corrected with high accuracy. In step 130, the image passes through step 154 three times, and all three times image correction is performed using the same correction filter.

That is, in this embodiment, some of the arithmetic means include:
The first filtering process in which the filtering process is performed using the filter determined according to the relative displacement amount, and the second filtering process that is performed subsequently, using the fixed filter regardless of the relative displacement amount. A2 processing steps;
The filter in the process of (1) is determined on the basis of the image shift amount obtained in the process of A1.

Next, from the focus evaluation amounts X (k) and Y (k) obtained as described above, the image shift amounts kpx and kpy in pixel units due to the respective focus evaluation amounts are detected again (steps 140 and 141). ). At this time, each focus evaluation amount X (k), Y
XD (XD = X) which roughly represents the contrast evaluation amount of (k)
(Kpx) -X (kpx + 1)) and YD (YD = Y (k
px + 1) -Y (kpx)) is also determined. This is because when the out-of-focus amount is large, looking at the contrast evaluation amounts XD and YD, which also takes into account the information of the edge image signal, gives information as if looking at the edge image signal for each relative displacement. This is because the information is used while paying attention to the fact that the contrast evaluation amounts XD and YD are large. Therefore, the two contrast evaluation amounts XD and YD are compared in step 142, and when XD ≧ YD, the focus evaluation amount X (k) is adopted (step 143), and XD <
In the case of YD, the focus evaluation amount Y (k) is adopted (step 1).
44). In steps 145 and 146, the adopted ZD
By using (z ₁ −z ₂ ) and Kz, an interpolation calculation of PR = kz + | z ₁ / ZD | is performed to obtain an image shift amount PR in a pixel unit or less. This process is shown in FIG. In the example as shown in the figure, since the relationship of XD <YD is satisfied, the focus evaluation amount Y (k) is adopted, and when calculating the image shift amount PR of a pixel unit or less, kz
= Kpy, z ₁ = (kp), z ₂ = Y (kp + 1). In step 130, the filter processing is performed using a fixed correction filter.
The meanings of the focus evaluation amounts at x and kpx + 1 are equal.

When the image shift amount of the pixel unit or less is obtained by the above steps, the defocus amount can be calculated based on the image shift amount.
In step 155, an "image shift sensitivity operation" subroutine is executed for obtaining the image shift amount → defocus amount conversion coefficient α required for this calculation.

In step 156, the defocus conversion coefficients α and PR and the constant G determined by the re-imaging optical system are used.

[0154]

(Equation 5) DEF which is the output of this subroutine
Is determined, and in a succeeding step 157, the subroutine is returned.

Next, the "image correction" subroutine will be described. As described above, the image on the photoelectric conversion element is given as an optical convolution of a subject and a point image.
That is, the impulse response of the optical system is a point image, the input is the luminance distribution of the subject, and the output is the image. On the other hand, if some kind of digital filter is determined and convolution integration of the digital filter and the image signal is performed, it is possible to change the shape of the blurred image by software. Therefore, if this digital filter is determined according to the conditions of the optical system and the overall result of the two convolutions is made to be equivalent to one symmetric transfer function, even if the distance measuring beam is vignetting Even if the image data of the line image is not symmetric, the image data after the filter processing can be returned to the target shape.

In particular, since the similarity of the images is important in the defocus calculation, the digital filter to be selected here does not need to be an inverse filter for restoring an asymmetric line image by an optical system into lines. As described above, a filter that restores a symmetric image may be used.

FIG. 39 shows the "image correction" subroutine.
To correct the blurred image, first create a correction filter according to the state of the taking lens, then perform convolution integration of the image data and this correction filter to correct the image, and finally pass the high-pass filter and then normal It is performed in the order of conversion. Since the shape of the blurred image correction filter described in FIG. 39 changes depending on the state of the distance measuring optical system, the type of the photographing lens, the distance measuring ring position, and the zoom position, it is extremely difficult to store all of them in the ROM. is there. So, ROM
Only contains the original data of the filter,
According to the above conditions, this is processed before use.

First, in step 601, information such as the open F number of the taking lens and the position of the exit pupil are fetched.

In step 603, a "correction filter magnification" subroutine for determining parameters for creating an image correction filter is executed. The information captured in the previous step is used in this subroutine.

In the following step 604, an "execution filter creation" subroutine for creating a correction filter actually used for image correction is executed in accordance with the parameters. Steps 603 and 604 will be described later.

In step 605, the peripheral value of the image data is extended to both sides to generate a virtual image signal in order to minimize an error particularly in correcting the peripheral image. FIG. 40 is an explanatory diagram showing a state where a signal of the same level as the peripheral value of the image is added to both sides of the actual image signal at the center.

As will be described later, this processing is necessary because the image correction is performed as a convolution integral with the image correction filter. That is, the modification of the signal of a certain pixel is
Since the correction is performed under the influence of the uncorrected signal that is located at a distance corresponding to the length of the image correction filter, the end of the actual image signal cannot be strictly corrected. If only a completely corrected image is always used for focus detection, it is necessary to add a considerably large number of image correction pixels around pixels used for defocus calculation. However, when the defocus of the photographing lens is large to some extent, for example, 5 mm or more, the image correction accuracy is not so required. Therefore,
The correction processing operation can be performed on the entire image signal, and in order to minimize the error of the image correction, the peripheral values of the image data are extended aside as shown in FIG.

Conversely, when the defocus amount is several millimeters or less, the image signal needs to be strictly corrected, so that the number of pixels used for the defocus calculation is smaller by 5 pixels at both ends than the actual image signal. You have set. This is because the filter length of the correction filter that adapts to a certain defocus is
This is because the use of the property that the smaller the defocus amount is, the shorter the defocus amount is, and by providing such a correction pixel, a filter corresponding to a small defocus can be operated without including an error.

As described above, in this embodiment, the photoelectric conversion element has pixels whose outputs are used for the filtering process on the pixels outside the pixels used for the operation for obtaining the evaluation amount. I have.

Note that the image data A (i), B
Speaking of (i) indicates data in the defocus calculation range.

In the next step 606, the previous step 6
Convolution integration is performed between the image correction filter obtained in step 04 and the virtual image signal obtained in step 605. By this processing, the image signal is returned to a shape as if there was no vignetting of the distance measuring light beam, and the similarity of the two images is restored.
The "convolution integral" subroutine will be described later in detail.

In the following step 607, high-pass filter processing for removing low-frequency components contained in the image signal is performed. When a strong spotlight is incident on the taking lens, a low-frequency ghost component may be included in the image signal. Therefore, in order to increase the accuracy of the defocus calculation described with reference to FIGS. 34 and 35, it is necessary to cut low frequencies.

FIG. 41B shows a state where a low frequency component is superimposed on the image signal of FIG. 41A. HA (i) = − A (i−2) + 2 × A (i) −A (i +
2) A (i): corrected image data HA (i): image data subjected to high-pass processing i: pixel position of the sensor Filter processing is performed, and Ha (i) → A again
If replaced by (i), the image data becomes as shown in FIG. 41 (C), and it is possible to accurately detect the amount of image shift.

In the last step 608, after normalizing the image signal to 8-bit data, the subroutine is returned in step 609.

Next, referring to FIGS. 42 to 44, the filtering process is performed using a filter determined in accordance with the aperture ratio of the imaging optical system to be focus-detected. The operation performed using a filter determined according to the exit pupil position of the imaging optical system to be performed will be described.

FIG. 42 shows a "correction filter magnification" subroutine. This subroutine performs preparations for creating a correction filter actually used for image correction from original filter information of the correction filter. Specifically, the horizontal axis magnification 1FGR, the horizontal axis magnification 2FGL, and the vertical axis magnification 1 used in the next “execution filter creation” subroutine
YGR and the vertical axis magnification 2YGL are determined. The horizontal axis magnification and the vertical axis magnification are the axis magnifications when the original filter stored in the ROM is compressed and expanded to create a correction filter.

First, F-number (aperture ratio) correction coefficients 1 to 4 HVR, HVL, LV used in the following steps
R, LVL, pupil position correction coefficients 1 to 4 KVR, KVL, M
FIGS. 43 and 44 show examples of VR and MVL.
Number correction coefficients 1-4 HVR, HVL, LVR, LV
L is a function of the open F number FNO of the photographing lens, and pupil position correction coefficients are 1 to 4 KVR, KVL, MVR, M
VL is a function of the distance PD between the exit pupil of the taking lens and the expected image plane. These relationships are determined by the diffusion characteristics of the diffusion plate inserted into the optical path of the distance measuring optical system.

Now, description will be made from step 702 in FIG. In this step, step 100 (the A1st subroutine) of the “defocus calculation” subroutine shown in FIGS.
It is determined whether this subroutine has been called during the process (A) or whether it has been called during step 130 (A2 process). If the call was made in step 100, step 703 (ie, B-th
1), and if it is called in step 130, that is, in the re-correlation calculation (A2 process), the process proceeds to step 704 (B2 process).

That is, in this embodiment, the arithmetic means includes the B1 processing step in which the filter processing is performed using the filter determined in accordance with the relative displacement, and the filter processing is performed independently of the relative displacement. No. B2 made using a fixed filter
And selecting either the B1 processing step or the B2 processing step. Selection means is provided.

In step 703, the relative displacement amount k is stored in MKFS which is a parameter for determining the filter length of the correction filter. "Defocus calculation" explained earlier
The reason that the image data is shifted by changing the value of k in the subroutine is nothing less than assuming a certain defocus for each k, and here, the filter of the correction filter is based on the assumed defocus. K is used to set the length. On the other hand, in step 704, PR ′ having a resolution equal to or less than the pixel unit is stored.

This step is executed when this subroutine is called in the re-correlation. In this case, the MKF
By setting S = PR ′, the correction filter for each relative displacement amount is fixed. This is because a more accurate correction filter is applied and the weight of the correlation evaluation amount is obtained regardless of the value of the relative displacement amount k when calculating the image shift amount of less than a pixel unit in step 146 of the “defocus calculation” subroutine of FIG. Is made constant, and the accuracy of the linear interpolation performed in step 146 of the “defocus calculation” subroutine is improved.

In step 705, the horizontal axis magnification is 1F.
The parameter NV determined by the defocus of the taking lens assumed for GR and the horizontal axis magnification of 2FGL is stored. This parameter NV is MKFS-FDLT as shown in FIG.
Is a function of Here, the FDLT is a correction term for applying a certain amount of correction filter to make the image asymmetric due to aberration of the taking lens and vignetting of the light beam even when the defocus of the taking lens is zero.

That is, the filter corrects the first and second signals even when the relative displacement is zero.

In general, the parameter NV is set to a value of 1 or less to avoid a decrease in accuracy due to the creation of an execution filter by enlarging the original filter. Since the defocus of the photographing lens and the size of the blurred image are substantially proportional, the NV has a characteristic substantially proportional to the absolute value of MKFS-FDLT. However, it is not necessarily in perfect proportion due to the aberration generated especially in the re-imaging lens. In the example shown in FIG. 45, this is considered by adding a slightly quadratic term.

The FDLT used here will be described a little more. Even when the taking lens is in focus, if the distance measuring light beam is vignetting, the point image on the photoelectric conversion element does not have a left-right symmetric shape. FIG. 46 is an explanatory diagram of this state, and is an example of a spot diagram on the photoelectric conversion element in a focused state. The direction of the arrow in the figure is the direction of the pixel column, and asymmetry of the light amount distribution is recognized with respect to this direction. By setting the FDLT, this asymmetry can be removed, and even a slight image mismatch at the time of focusing can be corrected. The value of FDLT may be stored in the ROM in the taking lens, and may be appropriately taken into the RAM in the microcomputer through communication between the taking lens and the camera body for use.

At the next step 706, the horizontal axis magnification is 1FG.
R is the horizontal axis magnification of 1 FGR, distance measurement visual field position PF, and proportionality constant C
The product of V is stored, and the horizontal axis magnification 2FGL is stored in the horizontal axis magnification 2FG.
The product of L, the distance measurement visual field position PF, and the proportionality constant CV is stored.
This is because the degree of vignetting of the distance measuring light beam can be approximated as being proportional to the distance from the optical axis of the photographing lens to the distance measuring visual field.

That is, the filter processing is performed using a filter determined according to the distance from the optical axis of the imaging optical system to be focus-detected to the distance measurement field determined by the photoelectric conversion means. I have.

Particularly, in this embodiment, when the defocus amount conversion operation is performed on the image shift amount to detect the imaging state of the imaging optical system, the defocus conversion operation is performed on the image shift amount. Is a function of the distance representing the distance from the optical axis of the image forming optical system to be subjected to focus detection to the distance measurement visual field determined by the photoelectric conversion means including the plurality of pixels.

In step 707, the product of the horizontal axis magnification 1FGR, the F number correction coefficient 1HVR, and the pupil position correction coefficient 1KVR is stored in the horizontal axis magnification 1FGR, and the horizontal axis magnification 2FGL and the F number correction coefficient 2HVL are stored in the horizontal axis magnification 2FGL. , The product of the pupil position correction coefficient 2KVL.

The product of the F-number correction coefficient 3LVR and the pupil position correction coefficient 3MVR is stored in the vertical axis magnification 1YGR,
The product of the F number correction coefficient 4LVL and the pupil position correction coefficient 4MVL is stored in the vertical axis magnification 2YGR, and the final value of each correction coefficient is determined.

At the last step 708, the subroutine is returned.

Next, FIG. 47 shows a "creation of execution filter" subroutine. In this subroutine, a correction filter is created from the original filter in accordance with the axial magnification determined in the “correction filter magnification” subroutine. Two kinds of original filters are stored in ROM,
MKFS-FDLT is selectively used depending on whether it is positive or negative. This is because the shape of the line image is determined mainly by the re-imaging lens irrespective of the type of the photographing lens, and can be roughly classified into a front focus and a rear focus. In other words, since the configuration of the re-imaging lens requires only a simple single lens, the aberration generated there is considerably larger than the aberration generated in the photographing lens.

FIG. 48 is a diagram for explaining this, and is a spot diagram when the subject distance is shifted before and after the in-focus position. The primary image of the subject is set at three points on the optical axis of the photographing lens, and light rays reaching the photoelectric conversion element through the dark photographing lens are traced. If the distance measuring light beam is vignetting, the right side of the point image is lost in the front focus state, and the left side of the point image is lost in the rear focus state. In other words, this means that the manner of pulling the bottom of the point image is reversed between the front focus and the rear focus. Furthermore, since the shape of the point image is completely different between the front pin and the rear pin, it is also clear that the correction filter adapted to these does not have the same shape even if it is turned over.

[Creating an execution filter] with reference to FIG.
The subroutine will be described.

In step 802, first, the MKFS-F
Check the sign of DLT. If the value is positive or zero, the process proceeds to step 803. If the value is negative, the process proceeds to step 803.
Move to 5.

In step 803, “first original filter 1” shown in FIG. 49A is selected from the original filters and stored in the RAM as a correction filter. In the following step 804, 10 is stored in the FOF. Also, in step 805, of the original filters, FIG.
The “second original filter 2” shown in (B) is selected and stored in the RAM as a correction filter. Next step 80
In step 6, 30 is stored in the FOF. In addition, this original filter is composed of 41 8-bit data.
F is a variable indicating the origin position of the original filter data. The reason why the FOF is used as a variable is that the original filter has a shape that is longer on one side as viewed from the origin.

The following steps 807 to 810 are steps for compressing and expanding the original filter data stored in the RAM as a correction filter in the previous step to create a desired correction filter.

First, in step 807, the ZR portion of the filter data, which mainly has the effect of tightening the bottom, is multiplied by FGR in the horizontal axis direction. Next, in step 808, the ZS portion having the action of mainly extending the skirt is further moved in the horizontal axis direction by FZ.
GL times.

After the filter length is determined in this manner, in the following step 809, the data of the ZR portion is compressed YGR times in the vertical axis direction, and in step 810, the data of the ZS portion is compressed YGL times. At this time, the data of the origin FOF is not subject to compression.

In the last step 811, this subroutine is returned.

As described above, in the present embodiment, the calculating means performs the first processing for creating the filter from the first original filter information and the second processing for creating the filter from the second original filter information. And a processing step,
The first process. A selection means is provided for selecting any one of the second processing steps according to the relative displacement between the first and second signals.

Thus, a correction filter is created.

As described above, in this embodiment, the first or second original filter is used according to the relative displacement. For example, by using two types of original filters, “Filter 1” mainly corresponding to the rear focus and “Filter 2” mainly corresponding to the front focus, it is possible to make an appropriate correction filter in any focus state. .

As described in the "correction filter magnification" subroutine, the horizontal axis magnification is 1FGR and the horizontal axis magnification is 2F.
Since GL, vertical axis magnification 1YGR, and vertical axis magnification 2YGL are functions of the open F number FNO of the photographing lens, the distance PD between the exit pupil of the photographing lens and the expected imaging plane, and the distance measuring field position PF, Even if a lens is mounted, it is possible to create an appropriate correction filter for each ranging field of view.

From the change in the horizontal axis magnifications 1 and 2 and the vertical axis magnifications 1 and 2 described in the “correction filter magnification” subroutine, how the shape of the correction filter changes can be roughly described as shown in FIG. Since the F-number correction coefficient 2 and the pupil position correction coefficient 1 shown in (B) and (C) approach zero as the photographing lens becomes brighter, the value of the vertical axis of the ZR portion of the image correction filter approaches zero. On the other hand, in the ZS section, the filter length approaches zero, and the image correcting ability gradually decreases.

It should be noted that in order to create all the correction filters used for image correction using one of the two types of original filters, it is necessary to store an extremely small amount of information in the ROM. There are also features.

FIG. 50 shows the "convolution integral" subroutine. The general formula for convolution integrals that deal with discrete data is
The two arrays are D (i) and E (i)

[0203]

(Equation 6) And S (i) is its output. Where D
If (i) is image data and E (i) is a correction filter, S (i) is corrected image data.
In this subroutine, each of a pair of secondary images is set as image data, and as described above, since the two blurred images have image resilience to the optical axis of the photographing lens,
On one of the image data, the correction filter is turned upside down to obtain those corrected images.

The "convolution integral" subroutine of FIG. 50 will be described. Image data OA (i), image O
B (i).

In step 902, that is, in the C1 processing step, convolution integration of the image OA (i) and the correction filter F (i) created by the "creation of execution filter" subroutine is performed. As mentioned above,

[0206]

(Equation 7) By performing the following calculation, a corrected image A (i) is obtained.

In step 903, that is, in the process of C2, convolution integration of the image OB (i) and the correction filter F (-i) turned over with respect to the origin FOF is performed.

[0208]

(Equation 8) By performing the following calculation, a corrected image B (i) is obtained. As described above, by using the same correction filter upside down, the “creation of the execution filter” described above is performed.
In other words, one filter should have been created in the subroutine. Therefore, the calculation time for creating the correction filter is reduced by almost half. Also, RO
This also has the effect of reducing the amount of original filter information stored in M.

In step 904, this subroutine is returned.

That is, when the imaging state of the imaging optical system is detected by performing the defocus amount conversion operation on the image shift amount, the defocus conversion operation is further performed on the focus for the image shift amount. The fact that it is a function of the aperture ratio or exit pupil position of the imaging optical system to be described will be described.

FIG. 51 shows the "image shift sensitivity" subroutine. This subroutine calculates the image shift amount → defocus conversion coefficient α. Since the shape of the appropriate correction filter changes according to not only the defocus of the photographing lens but also the vignetting of the distance measuring light beam, this image shift sensitivity is also a function of the open F number, the exit pupil position, and the distance measuring field position. Becomes

That is, when the imaging state of the imaging optical system is detected by performing a defocus amount conversion operation on the image shift amount, the defocus conversion operation is further performed on the focus for the image shift amount. The aperture ratio or the exit pupil position of the imaging optical system to be formed, or a function representing the distance from the optical axis of the imaging optical system to the distance measurement visual field.

First, in step 1001, the MKF
Check the sign of S-FDLT. If it is positive or zero, the process proceeds to step 1003, and if it is negative, the process proceeds to step 1004. This can be said to be a branch due to the original filter used.

In step 1003, the F-number correction coefficients 1 to 4 HVR, HVL, LV shown in FIGS.
R, LVL, pupil position correction coefficients 1 to 4 KVR, KVL, M
Using VR, MVL, and the distance-measuring visual field position PF, the image shift sensitivity AP when the distance-measuring light beam has no vignetting is corrected.
The correction formula is: a = AP + PF × (H1 × HVL × LVL × KVL × M
VL + H2 × HVR × LVR × KVR × MVR), and H1 and H2 are correction coefficients depending on the shape of the correction filter.

Similarly, in step 1004, a correction formula a = AP + PF × (H3 × HVL × LVL × KVL × M) using correction coefficients H3 and H4 based on the shape of the correction filter.
VL + H4 × HVR × LVR × KVR × MVR) determines the image shift sensitivity.

At the last step 1005, this subroutine is returned.

As described above, the image shift sensitivity α is calculated based on the above arithmetic expression by using the open F number FN of the photographing lens.
O, the distance PD between the exit pupil of the photographing lens and the expected imaging plane, and the distance measuring field position PF are prepared for each correction filter as a function of the distance measuring field, so that no matter which photographing lens is attached, And accurate defocus calculation can be performed.

The above equation will be described. As described with reference to FIG. 23, the feature of the image correction filter is that the ZS portion has the function of extending the skirt of the image, and the ZR portion has the function of tightening the skirt of the image. Pay attention to the movement of the image,
Considering this, it can be said that both the ZS portion and the ZR portion in FIG. 23A have an effect of moving the center of gravity to the left with respect to the line image in FIG. FIG.
Both the ZS portion and the ZR portion in FIG. 4A move the barycenter of the line image in FIG. 24B to the right. Therefore, it can be seen that the center of gravity of the two images is separated from each other by such a filtering process. However, in this case, since the distance measuring light beam is vignetted from the outside, the original image interval is closer than when there is no vignetting in the distance measuring light beam. It means that you are approaching when there is no vignetting. From the properties of the filter described above, it can be seen that in order to convert the image shift amount to the defocus amount, it is necessary to further use this conversion function as a function of the correction filter. Then, ZS part of filter, ZR
Based on the above properties of the unit, the correction term of the image shift sensitivity by the filter is defined as the sum of the operation of the ZS unit and the operation of the ZR unit. More specifically, the correction is performed using an F-number correction coefficient, a pupil position correction coefficient, and a distance measurement visual field position which determine the shape of a correction filter required for correcting an image. The F-number correction coefficients 2, 4HVL, LVL, and a pupil position correction coefficient 2,4 KV
L, MVL product and F number correction coefficient 1,3 HVR,
If the image shift sensitivity AP is corrected when there is no vignetting in the distance measuring light beam, the correction filter is used for correcting the sum of the product of the LVR and the product of the pupil position correction coefficients 1, 3 KVR and MVR and the distance measuring field position PF. Image shift sensitivity suitable for the above. By such a calculation, the distance measurement light flux becomes more vignetting as the position of the distance measurement visual field becomes farther from the optical axis of the photographing lens, the vignetting becomes easier as the aperture ratio becomes larger, and the exit pupil position of the photographing lens becomes the focus detection optical system. The further away from the set pupil position, the more easily vignetting occurs. The image shift sensitivity is set in consideration of each of the following.

The description of the subroutine ends here. Now, the operation of the defocus detection in FIG. 34 will be described again.

In the optical path of the focus detecting optical system, if a diffusion plate is provided on the expected image forming plane of the photographing lens, the distance measurement accuracy is improved for a bright lens having a small F number. However, if a camera having such a focus detection device is mounted with a dark photographing lens having a large F-number, vignetting of the ranging light flux occurs. For example, image data as shown in FIG. Obtained as the output of the conversion element.

FIGS. 52 (A) to 52 (C), FIG. 53 (A),
(B) shows an image formation state of the photographing lens, and FIG.
(A), (B) → focus (c) → rear pin diagram 53 (A),
(B) is changed to show line images in each state. In the figure, OA (i) and OB (i) are line images obtained from a pair of pixel columns. Since the line image is an impulse response of a combined system of the photographing lens and the distance measuring optical system, by returning the line image to a symmetrical shape, the shape of a pair of subject images formed by the distance measuring optical system is also returned to a similar shape. As a result, it becomes possible to know the relative distance between the two images.

In the line images shown in the figures, a pair of images are in a reversed relationship to each other, and their shapes can be roughly classified into a front pin and a rear pin. Also, since the degree of spread of the line image changes almost in proportion to the amount of image shift, blur image correction filters for front focus and rear focus are prepared as shown in FIGS. The line image can be returned to the left-right symmetrical shape by adjusting and acting so as to be appropriate for the defocus at that time.

Further, a method of applying a correction filter to image data at a stage where the image shift amount is unknown is shown in FIGS.
This is the “defocus calculation” subroutine shown in FIG.
In this subroutine, the value of the relative shift amount k is sequentially changed in step 100, and the image is corrected using a correction filter that is a function of k, and the correlation evaluation amount ((1), (1), (Equation (2)) is obtained.

FIGS. 54 (A) to 54 (C), FIG. 55 (A),
(B) is a diagram showing the shape of the correction filter for various values of k, and FIGS. 54 (A) to (C), FIG. 55 (A),
(B) is an example of a filter in which the line images of FIGS. 52 and 53 are each appropriately corrected. The correlation evaluation amount is obtained while sequentially applying such a filter to a certain line image. When the relative displacement amount k is different from the actual image shift amount, the image correction becomes inappropriate. However, when the actual image shift amount gradually approaches with the sequential change of k, the image correction also approaches an appropriate state. Go. In this way, the best image correction state is obtained at k closest to the actual image shift amount. Therefore, finding the relative displacement amount k at which the correlation evaluation amount becomes zero is nothing less than finding an optimal image correction filter. When the appropriate correction filter is activated, the corrected line image is shown in FIG.
(A) to (C), as shown in FIGS. 57 (A) and (B),
Each image is sufficiently symmetrical. As a result, the pair of images has a similar shape, and the relative interval between the two images can be accurately obtained. As described above, if the line image is corrected in this manner, the general subject image is similarly corrected to two images having similar shapes.

On the other hand, in the sequential change of the relative displacement k,
FIG. 58 shows a state in which an inappropriate correction filter has acted on the line image shown in FIG. 53 (B). Fig. 58
(A) and (C) are examples of correction filters different from the correction filter of FIG. 55 (B), and FIGS. 58 (B) and (D) are correction images by these correction filters, respectively. As described above, when the relative displacement amount k is significantly different from the shift amount between the two images, the similarity of the corrected image is extremely reduced, and the change in the correlation evaluation amount accompanying the relative displacement of the images is further emphasized.

In step 130 of the "defocus calculation" subroutine of FIGS. 34 and 35, a more accurate correction filter is created using the relative displacement PR 'of a pixel unit or less obtained by the interpolation processing. The final image shift amount is calculated by performing the maximum correlation operation while keeping the shape of.

In order to obtain a correction filter suitable for various conditions for determining the vignetting condition of the distance measuring light beam, it is necessary to correct the F number obtained as a function of the open F number, the exit pupil position, and the distance measuring visual field position. The horizontal axis and the vertical axis of the original filter are enlarged or reduced based on the coefficients 1 to 4 and the pupil position correction coefficients 1 to 4 and FIGS. 43 and 44 (“create execution filter” subroutine). For example, FIG. 59 is a diagram showing a correction filter corresponding to a bright photographing lens having a small F-number when the photographing lens is attached. It is assumed that there is still some vignetting and the relative displacement is the same as that in FIG. 57 (B). It was done. This filter is obtained by making the filter length of the original filter correspond to the relative displacement, and then expanding the ZR portion in the horizontal axis direction, reducing the vertical axis direction, and reducing the ZS portion in the horizontal axis direction.

In the above description, the defocus detection method using image correction has been described by taking a distance measuring optical system having a diffusion plate in the optical path as an example. However, the present invention is not limited to this, but can be applied to various other ranging optical systems.

For example, in the conventional distance measuring optical system in which the diffusion plate 42 is removed from the distance measuring optical system shown in FIG. 1, even when a considerably dark photographing lens is attached and the distance measuring light beam is vignetted, The defocus detection shown here is effective.
Therefore, a distance measuring optical system is configured so that a distance measuring light beam is not vignetted for a photographing lens having an open F-number brighter than 5.6, and a conventional method is used for a photographing lens with F5.6 or brighter. It is also conceivable to perform distance measurement and, on the other hand, if a photographing lens darker than F5.6 is attached, perform image correction and detect defocus as described above.

Further, when such a conventional distance measuring optical system has a diaphragm as shown in FIG.
In a strict sense, the shapes of the blurred images do not match in a strict sense even if the distance measuring light beam is not vignetting. This is because the aperture 48
Due to the fact that the shapes of a and 48b do not overlap with one another when the opening is moved in parallel, the blurred image behaves as if the outside of the rectangular opening is vignetting on an arc. is there. Therefore, the image can be corrected by substantially the same filter processing as described above. However, the only difference is that the correction filter is not a function of the F-number of the photographing lens, the exit pupil position, and the distance measurement visual field position.

[0231]

As described above, according to the present invention, (a) even in a dark imaging lens having a large F-number with a vignetting of the distance measuring light beam, the defocus amount conversion operation can be accurately performed for the image shift amount. Now you can do it. Therefore,
The restriction on the aperture ratio of the imaging optical system in the conventional focus detection device is relaxed, and the focus can be detected regardless of the aperture ratio of the imaging lens.

In particular, in a focus detection device having an improved distance measuring optical system in which a diffusion plate is arranged in the optical path of the distance measuring optical system, the distance measuring characteristic which is a feature of a bright imaging lens is described. At the same time as the improvement in accuracy has been achieved, it has become possible to detect a focus even on a dark imaging lens which could not be used because of a vignetting distance measuring light beam. (B) Since the distance can be measured even if the distance measuring light beam is vignetting as described above, the distance measuring visual field can be set at any position in the shooting screen without darkening the setting of the light beam to be taken into the focus detection optical system. It became possible to arrange. (C) Accurate defocus detection is possible irrespective of the aperture ratio of the imaging optical system, and the number of lens drive operations until focusing can be reduced.

[Brief description of the drawings]

FIG. 1 is a sectional view of a focus detection optical system according to the present invention.

FIG. 2 is a diagram showing diffusion characteristics of a diffusion plate 42 in FIG. 1;

FIG. 3 is a plan view of the diaphragm 48 of FIG. 1;

FIG. 4 is a diagram showing an image of an aperture on an exit pupil of the taking lens of FIG. 1;

FIG. 5 is a diagram for explaining the principle of distance measurement of the focus detection device.

FIG. 6 is a cross-sectional view of a single-lens reflex camera incorporating a focus detection device.

FIG. 7 is a cross-sectional view in the short side direction of a finder visual field in which a focus detection optical system is developed.

FIG. 8 is a cross-sectional view of a viewfinder with a focus detection optical system developed in a long side direction.

FIG. 9 is a plan view of a condenser lens 85;

FIG. 10 is an explanatory view of a focus plate 107.

FIG. 11 is a plan view showing the shape of the re-imaging lens 90 viewed from the light incident surface.

FIG. 12 is a plan view of an aperture 93;

FIG. 13 is a view for explaining a projected image of an aperture opening.

FIG. 14 is a plan view of a light shielding mask 89.

FIG. 15 is a diagram showing a positional relationship between a shooting screen and a distance measurement field of view.

FIG. 16 illustrates a pixel arrangement of a photoelectric conversion element.

FIG. 17 is a partially enlarged view of the light guide prism of FIG. 6;

FIG. 18 is a perspective view of the light guide prism of FIG. 6;

FIG. 19 is a perspective view for explaining the vignetting of the distance measuring light beam.

20 is a plan view of the optical path to 104j in the distance measurement visual field shown in FIG. 15 as viewed from above.

FIG. 21 is a diagram showing a range through which a photographing light beam passes on a surface of a distance measuring optical system on which an aperture image is formed.

FIG. 22 is a plan view of the optical path to 104j in the ranging field of view shown in FIG. 15 as viewed from above;

FIG. 23 shows a blurred image, its correction filter, image data,
Explanatory drawing of corrected image data

FIG. 24 shows a blurred image and its correction filter, image data,
Explanatory drawing of corrected image data

FIG. 25 is an explanatory diagram for correcting a general subject image.

FIG. 26 is an explanatory diagram for correcting a general subject image.

FIG. 27 is a block diagram showing a circuit of a camera according to the present invention.

FIG. 28 is a diagram showing a position detection configuration of a distance ring and a zoom ring of a photographing lens.

FIG. 29 is a flowchart illustrating a sequence of a camera.

FIG. 30 is a flowchart of an “AF” subroutine.

FIG. 31 is a flowchart of a “focus detection” subroutine.

FIG. 32 is a flowchart of a “lens drive” subroutine.

FIG. 33 is a flowchart of a “release” subroutine.

FIG. 34 is a flowchart of a “defocus calculation” subroutine.

FIG. 35 is a flowchart of a “defocus calculation” subroutine.

FIG. 36 is a diagram for explaining a correspondence between a relative displacement amount and a calculation pixel;

FIG. 37 is a diagram illustrating an example of a relative displacement amount and a focus evaluation amount.

FIG. 38 is a diagram for explaining a method of calculating an image shift amount in pixel units or less.

FIG. 39 is a flowchart of an “image correction” subroutine.

FIG. 40 is a view for explaining how the peripheral values of the image data are extended.

FIG. 41 is a diagram showing an image signal without low-frequency noise, an image signal with low-frequency noise superimposed thereon, and an image signal subjected to high-pass filtering, respectively.

FIG. 42 is a flowchart of a “correction filter magnification” subroutine.

FIG. 43 is a characteristic diagram of an F number correction coefficient and a pupil position correction coefficient.

FIG. 44 is a characteristic diagram of an F-number correction coefficient and a pupil position correction coefficient.

FIG. 45 is a diagram showing the relationship between MKFS-FDLT and NV.

FIG. 46: Spot diagram of a point image

FIG. 47 is a flowchart of a “create execution filter” subroutine.

FIG. 48 is a spot diagram for explaining a change in a point image depending on an image forming state.

FIG. 49 is a view showing an original filter.

FIG. 50 is a flowchart of a “convolution integral” subroutine.

FIG. 51 is a flowchart of an “image shift sensitivity” subroutine;

FIG. 52 is a diagram showing image data of a line image according to the image forming state of the photographing lens.

FIG. 53 is a diagram showing image data of a line image according to the image forming state of the photographing lens.

FIG. 54 is a view showing a blur correction filter;

FIG. 55 is a diagram showing a blurred image correction filter.

FIG. 56 is a view showing image data of a corrected line image.

FIG. 57 is a diagram showing image data of a corrected line image.

FIG. 58 is a diagram showing a blur image correction filter and image data of a corrected line image.

FIG. 59 is a diagram illustrating an example of a corresponding correction filter when a bright photographing lens having a small F-number is attached.

[Explanation of symbols]

 40 subject surface 41, 81 photographing lens 42, 84 diffuser 43, 85 field lens 44, 90 re-imaging lens 46, 47, 108 photoelectric conversion element array 48, 93 aperture 88 eyepiece 89 light-shielding mask

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Akira Akashi 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (56) References JP-A-3-214133 (JP, A) JP-A-61 -18911 (JP, A) JP-A-60-189721 (JP, A) JP-A-60-86517 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G02B ^7/ 28- 7/40

Claims (3)

(57) [Claims] 1. A re-imaging optical system that forms first and second two object images whose relative positional relationship changes according to an imaging state of an imaging optical system to be focus-detected, The first and second
Photoelectric conversion means comprising a plurality of pixels for respectively outputting first and second signals corresponding to the object image of the object, and sequentially displacing the first and second signals in an arithmetic operation to obtain relative displacement positions In calculating the evaluation amount indicating the degree of coincidence between the first and second signals in the above, a filter for each relative displacement position determined in advance in accordance with the relative displacement position with respect to the first and second signals. Performing a filtering process with the use of the image data, calculating an image shift amount between the first and second object images based on the change in the evaluation amount, and performing a defocus amount conversion operation on the image shift amount. focus detection apparatus der detecting the imaging state of the imaging optical system by
I, the defocus amount conversion operation on said image shift amount,
Further, the aperture ratio of the imaging optical system to be focus-detected is set to one.
A focus detection device which calculates as a function . 2. A re-imaging optical system for forming first and second object images whose relative positional relationship changes in accordance with an imaging state of an imaging optical system to be focus-detected, The first and second
Photoelectric conversion means comprising a plurality of pixels for respectively outputting first and second signals corresponding to the object image of the object, and sequentially displacing the first and second signals in an arithmetic operation to obtain relative displacement positions When calculating the evaluation amount representing the degree of coincidence between the first and second signals in the above, a filter for each relative displacement position determined in advance according to the relative displacement position with respect to the first and second signals is used. Filter processing, calculating an image shift amount between the first and second object images based on the change in the evaluation amount, and performing a defocus amount conversion operation on the image shift amount to obtain the image forming optical system. met focus detection device for detecting the imaging state of the system
In addition, the defocus amount conversion calculation for the image shift amount further sets one exit pupil position of the imaging optical system to be subjected to focus detection.
A focus detection device , wherein the calculation is performed as a function of: 3. A re-imaging optical system for forming a pair of object images whose relative positional relationship changes according to an imaging state of an imaging optical system to be focus-detected. It has a plurality of photoelectric conversion means comprising a plurality of pixels for outputting corresponding first and second signals, respectively, and relatively displaces the first and second signals sequentially in operation, and at each relative displacement position When calculating an evaluation quantity representing the degree of coincidence between the first and second signals,
Filter processing is performed on the first and second signals in advance using a filter for each relative displacement position determined according to the relative displacement position, and the first and second signals are determined based on the change in the evaluation amount. determine the image shift amount of the object image, a focus detection device for detecting the imaging state of the imaging optical system by performing a defocus amount conversion operation on said image shift amount, said image shift amount to In the defocus amount conversion calculation, the distance representing the distance from the optical axis of the imaging optical system to be subjected to focus detection to the distance measurement field determined by the photoelectric conversion means including the plurality of pixels is calculated as one function. A focus detection device characterized by the above-mentioned.