JPH0624517B2 - Focus detection device of fundus observation device - Google Patents

Description

TECHNICAL FIELD OF THE INVENTION The present invention relates to a focus detection device for a fundus camera.

(Background of the Invention) In an apparatus for observing the fundus, it is difficult to focus due to insufficient light amount due to low reflectance of the fundus and poor contrast. Therefore, as disclosed in Japanese Patent Publication No. 57-13294, there is proposed a device that projects an index on the fundus and focuses on the index image.

This device divides an index image with a split prism or the like to form an image on the fundus of the eye, observes lateral displacement of the index image, and focuses the image. Here, the observation optical system is only for observing the state of the lateral shift, and it is the index projection optical system that directly focuses on the fundus. Then, the index and the observation image plane (and, if necessary, the imaging image plane) are made to cooperate by a mechanical interlocking device, and the index is conjugated with the fundus to indirectly focus on the fundus.

However, in such an indirect structure, a mechanical error of the interlocking device between the index projection optical system and the observation optical system or an error due to the difference between the optical systems (the index projection optical system is
Is large, and since the peripheral light flux is used, the aberration is large, and the observation optical system has a small NA, a deep depth of focus, and the aberration is relatively small.)
In addition, the interlocking device itself becomes complicated, which leads to an increase in size and cost of the device. In addition, when the magnification is changed, insufficient light intensity may occur, and the reflectance of the fundus may vary depending on the site, resulting in insufficient light intensity at some regions or excessive light intensity at certain regions, resulting in saturation of the light receiving element (eg, CCD). However, there is a drawback that signal detection becomes difficult due to insufficient light amount.

(Object of the Invention) An object of the present invention is to solve these drawbacks, and to provide a focus detection device capable of performing accurate signal detection regardless of the difference in the reflectance of the fundus and the magnification change.

(Summary of the invention) Therefore, the present invention has an index projection optical system for projecting a luminescent index on the fundus of the eye to be examined and a photoelectric converter for receiving the index of the fundus of the eye to be examined, and the output of the photoelectric converter causes focusing. In the focus detection device of the fundus oculi observation device including the focus information detection optical system that detects information, a selection unit that selects whether or not to project the index, and from the photoelectric converter at the time of projection of the index. A subtraction unit that obtains a difference signal between the obtained first signal and the second signal obtained from the photoelectric converter when the index is not projected, and outputs a focusing signal from the difference signal; When the signal obtained from the photoelectric converter is equal to or higher than a predetermined level when the index is projected, a control unit that reduces the light amount of the index so as to be equal to or lower than the level is provided. And

(Embodiment) FIG. 1 is a diagram showing the overall configuration of a fundus camera having a focus detection device according to an embodiment of the present invention.

Normally, the illumination system includes an observation illumination light source 1, a light source relay lens 2, a reflecting mirror 3, and a strobe tube 4 and a ring slit 5, which are arranged conjugate with the light source 1 by the light source relay lens 2.
(The plan view is shown in FIG. 3), the ring slit relay lenses 6 and 8, the perforated mirror 9, and the objective lens 11. As is well known, the ring slit 5 is adjusted by the relay lenses 6 and 8 for the ring slit and the objective lens 11 so that the working distance between the fundus camera and the eye 12 is adjusted so as to be substantially conjugate with the cornea of the eye 12. To be done. An index projection light source 16 that can be turned on and off by an index projection light source driving unit described below passes through an index projection relay lens 17a and a slit-formed index plate 18 (a plan view is shown in FIG. 4) from behind. Illuminate. The light source 16 forms an image after passing through the slit of the index plate 18, and the index projection relay lens 17b, the dichroic prism 7, and the ring slit relay lens 8 are formed.
Is imaged again in the vicinity of the perforated mirror 9 and further in the vicinity of the pupil of the eye 12 to be examined by the objective lens 11 to illuminate the fundus. These rays are drawn by the chain double-dashed line in FIG.

By the way, in the case of the optical system shown in FIG.
Reference numeral 16 is, for example, an infrared light emitting diode, and the dichroic prism 7 has characteristics of infrared light reflection and visible light transmission. Therefore, the dichroic prism 7 is the observation light source 1,
Visible light from the strobe tube 4 passes, but infrared light emitted from these light sources 1 and 4 is reflected and goes out of the illumination optical system, and the fundus is eventually reflected by the infrared light from the index projection light source 16. Will be illuminated. When the dichroic mirror 14 is made to have a characteristic of reflecting infrared light and transmitting visible light, the light reaching the array sensor 23 described later is only the transmitted light of the index plate 18 illuminated by the index projection light source 16, and the index image ( The slit image) has a relatively good contrast.

On the other hand, as shown by the dotted line in FIG. 1, after the slit of the index plate 18 is imaged near the dichroic prism 7 by the index projection system relay lens 17b, the relay lens 8,
An image is formed at the rear focal position of the objective lens 11 by the perforated mirror 9. Therefore, the light transmitted through the slit that has passed through the objective lens 11 becomes substantially parallel light and enters the eye 12 to be inspected, and if the eye 12 to be inspected is an emmetropic eye, the slit forms an image on the fundus. However, the slit image may be defocused depending on the diopter of the eye 12 to be inspected. However, the light source that actually illuminates the index image 18 forms a conjugate image of the eye 12 near the pupil with respect to the subject's eye 12, and since the size of the light source image is small, the numerical aperture (NA) of the light flux that substantially exits the slit is It becomes smaller and the depth of focus becomes deeper. Therefore, in the substantial range of the eye to be inspected (about ± 15 diopters), the blur of the image due to defocus is not so large, and the contrast that withstands the focus detection is guaranteed.

FIG. 2 shows only the focus information detection optical system in FIG. In the present embodiment, the focus information is obtained even for zooming and zooming, but for the sake of explanation, an operation at a certain magnification will be described, and then a device for zooming and zooming will be described.

As shown by the solid line in FIG. 2, the slit image projected on the fundus serves as a secondary light source and is once imaged by the objective lens 11, and then the aperture of the perforated mirror 9, the aperture stop 10, and the focusing. After passing through the relay lens 13 and being reflected by a dichroic mirror 14 that transmits infrared light and visible light, an image is formed in the vicinity of the field stop 19. The field stop 19 is conjugated with the array sensor 23 by the re-imaging lens 21, and the array sensor 23 is conjugated with the imaging surface (observation image surface is not shown) 15. Therefore, the slit image from the fundus is imaged in the vicinity of the array sensor 23 by the re-imaging lens 21, and at that time, the pupil division roof type prism 22 (a stereoscopic view is shown in FIG. 5 ... FIGS. 1 and 2). In the figure, the ridge is drawn perpendicular to the paper surface, but for convenience of expression, it is actually rotated by 90 ° around the optical axis and is arranged so that the ridge is parallel to the paper surface.) The light flux is divided into two by each, and each light flux is coupled to a different part of the array sensor 23 (arranged so that the array direction of the cells of the array sensor 23 is orthogonal to the paper surface according to the ridge direction of the prism 22). To be imaged.

Here, as shown by the dotted line in FIG. 2, the ridge of the prism of the pupil division prism 22 is almost conjugate with the aperture stop 10 by the focusing relay lens 13 and the field lens 20 behind the field stop 19.

Since this aperture stop 10 corresponds to the exit pupil of the focusing optical system,
The image on the array sensor 23 has two different pupils of this optical system.
An image is formed by the light fluxes from the two parts, and the two images thus obtained are laterally offset from each other by the front focus and the rear focus, as is well known. Therefore, the focus state can be detected by measuring the distance between the two images on the array sensor 23. That is, if the distance between the two images is L during focusing as shown by the solid line in FIG. 6, if the focal point is behind the field stop 19 (dotted line in FIG. 6), the image distance is as shown in FIG. As indicated by the dotted line in Fig. 2, the focus is narrower than L and the focal point is the field stop 19
In the case of being further forward, the image interval is wider than L.

When measuring the distance between the images, the distance between the two image positions can be easily determined if the image itself is in a good image-forming state with good contrast. However, in reality, an image is often formed in a defocused state on the array sensor 23, and a good image is not always obtained.

Therefore, even in such a bad condition, the method proposed in Japanese Patent Laid-Open No. 54-68667 is used to accurately extract the lateral displacement of the image.

This is a method in which the image on the array sensor 23 is Fourier-transformed, the phase component thereof is extracted, and the lateral shift is obtained by the transition theorem of Fourier transform.

The image intensity on the array sensor 23 is defined as a function f (x) of the position X.
Then, the Fourier transform F ₁ of the function f (x) before the image is displaced can be expressed as in equation (1) where S is the spatial frequency.

Fourier transform F ₂ when this image is laterally displaced by α ₂
Becomes the formula (2) according to the transition theorem.

F ₂ (s) ＝ e ^{−2 π ias} F ₁ (s) …… (2) When comparing equations (1) and (2), the magnitude of the Fourier component does not change due to the displacement, and only the phase shifts by 2πas. If the shifted phase is Δθ, the displacement α can be calculated by knowing Δθ.
Obtained as in (3).

To specifically realize this, the discrete Fourier transform (DFT) is performed with the spatial frequency as the reciprocal (1 / l) of the sample length l on the array sensor 23.

That is, as shown in FIG. 7, for one of the two slit images on the array sensor 23, an electric signal corresponding to the amount of light incident on each cell on the array sensor 23 is I _i , and the number of samples of the cell. Is N _i and the length of the array sensor 23 at the sampled portion is 1, and the following electric quantities Ix and Iy are obtained.

These Ix and Iy are obtained by multiplying the real part (Ix) and the imaginary part (Iy) of the Fourier component of the spatial frequency 1 / l in the image on the array sensor 23 by N / l. Therefore, the phase deviation angle θ of the Fourier component is obtained from Ix and Iy as shown in (6).

If the image is laterally displaced, this phase deviation angle θ is displaced, and the amount of change in θ obtained by the calculation can be corrected to the lateral displacement according to the equation (3).

Therefore, the following distance d is defined by the phase deviation angle θ as in Expression (7).

Focusing information can be obtained by calculating the distance d for each of the two slit images on the array sensor 23, calculating the distance between the two slit images, and comparing this with the distance at the time of focusing. . As shown in Figure 1,
From the array sensor 23, each cell is sequentially driven by the signal from the driving means 24, and photoelectric conversion signals corresponding to each cell are sequentially output. Based on this output, the calculation means 25 performs the above calculation, and based on the result, the motor drive means
The motor M is driven by 26, and the focusing relay lens 13 is moved in the optical axis direction for focusing.

An electric processing system for performing the above-mentioned processing is shown in the block diagram of FIG. 8, and its operation is shown in FIG. 9 and a flow chart of the microcomputer 252 in FIG. 10 (a) and FIG. 10 (b).
Explained by. In the block diagram of FIG. 8, the same members as those in FIG. 1 are designated by the same reference numerals, but the array sensor 23 is called, for example, a charge coupled device (CCD), and has a start signal (see Figure 9
Each cell is driven by the clock pulse (FIG. 9 (C)) following (b)), that is, after the start signal is generated,
The array sensor 23 outputs an electric signal corresponding to the amount of light incident on the _first cell C ₁ of the array sensor 23 (FIG. 9A) with the first pulse P ₁ . Further, for example, an electric signal corresponding to the amount of light incident on the _seventh cell C ₇ by the seventh pulse P ₇ is output.

By the way, looking at the electric signal output from the array sensor 23, various noise components are included.

In other words, the light flux incident on the array sensor 23 includes the light flux S from the slit image projected on the fundus as a signal component to be obtained and the observation N ₁ and stray light N ₂ as unnecessary noise components. Further, there is a noise component N ₃ due to the dark current of the array sensor 23 itself. Therefore, the electric signal output from the array sensor 23 is S + N ₁ + N ₂ + N ₃ (FIG. 12 (a)). The above is the index projection light source drive circuit 253.
When the index projection light source 16 is turned on by, but consider next when the index projection light source 16 is turned off,
The electric signal output from the array sensor 23 is a combined signal N ₁ + N ₂ + N ₃ of the observation light N ₁ , stray light N ₂ , and dark current N ₃ as unnecessary noise components (FIG. 12 (b)). Therefore, the output of the array sensor 23 when the light source 16 for index projection is turned on (
By taking the difference between the output of the array sensor 23 when the light source 16 for index projection is turned off (FIG. 12 (b)),
Only the signal component S can be obtained (Fig. 12 (c)).

A method for obtaining the difference between the outputs of the array sensor 23 when the index projection light source 16 is turned on and when it is turned off will be described.
The electric signal photoelectrically converted by the array sensor 23 is
It is output at every storage time Tint of the array sensor 23. Therefore, the index projection light source 16 is turned off in advance and the period T during which it is turned on for the first time is set to To (step 1 in FIG. 10 (a)).
00) In synchronization with the N-th start pulse (FIG. 13 (a)) after at least one storage time Tint has elapsed (step 101 in FIG. 10 (a)), the index projection light source 16 is turned on (FIG. 10).
(Step 102 of (a)) The output of the array sensor 23 output here is a value photoelectrically converted when the light source 16 is turned off.
It becomes the noise component in Fig. 12 (b). The signal output from the array sensor 23 is converted into a digital signal by the A / D converter 251 via the amplifier 250 and the like. This digital signal is input to the microcomputer 252. That is, the microcomputer 252 synchronizes with the scanning of the array sensor 23 and the A / D
The output I _i of the converter 251 is read and the output corresponding to each cell is stored in the built-in memories M _1-1 and M _2-1 . (No.
10 Step 103 of FIG. 10A). Here, the memory M _1-1 is the ninth
As shown in FIG. 3A, the signals from the cells of the length l _{1 in} the left half of the array sensor 23 are stored in the memory M _2-1.
Also has a length l ₂ that is approximately in the right half of the array sensor 23.
The signal from the cell group will be stored. The microcomputer 252 detects that a predetermined period T has elapsed since the light source 16 was turned on (step 10 in FIG. 10 (a)).
4), turn off the light source 16 (step 10 in FIG. 10 (a))
5). Next, the output of the array sensor 23 synchronized with the (N + 2) th start pulse is a value photoelectrically converted when the light source 16 is turned off, and becomes the signal component + noise component of FIG. 12 (a) (first
0 step 106 of FIG. The output signal is stored in the memories M _1-2 and M _2-2 in the same manner as the processing for the noise component only (step 107 in FIG. 10 (a)). Fig. 10 (a)
The timing of turning off the index projection light source 16 in step 105 will be described later. Next, the microcomputer 252 determines that the stored values of the memories M _1-1 , M _1-2 , M _2-1 and M _2-2 . performs determination of whether more than a predetermined value (a step 108 of FIG. 10 (a)), a memory M takes the difference between the memory M _1-1 from the memory N _2-1 among the stored value is equal to or less than a predetermined value Remember in ₁ ,
The difference between the memory M _2-2 and the memory M _2-1 is _calculated and stored in the memory M ₂ (step 109 in FIG. 10 (a)). And memory M
_The period T in which the light source 16 is next turned on is determined from the magnitudes of the values stored in ₁ and M ₂ (step 11 in FIG. 10 (a)).
0).

Next, the microcomputer 252 sequentially performs the operations of the formula (4), the formula (5), the formula (6) and the formula (7) according to the stored value of the memory M ₁ , and stores the obtained distance d ₁ in the memory M ₃ . (Fig. 10 (b)
Step 111). Similarly, the distance d ₂ obtained from the stored value of the memory M ₂ is stored in the memory M ₄ (step 112 in FIG. 10 (b)).

Next, the microcomputer 252 performs an operation (l ₁ −d ₁ + d ₂ ) between the stored values d ₁ and d ₂ of the memories M ₃ and M ₄ and the distance l ₁ (step 113 in FIG. 10 (b)). ). And step 11
The value (l−d ₁ + d ₂ ) obtained in step 3 ... This value corresponds to the interval between the two slit images, and is compared with the in-focus interval to find the difference, and this difference is used as in-focus information. Can be used as

The motor driving means 26 inputs a pulse signal of a different frequency to the motor M according to the focus information output from the microcomputer 252. At this time, the frequency of the signal input to the motor M is large when the focusing relay lens 13 largely deviates from the in-focus position. Therefore, the motor M rapidly rotates, and when the focusing relay lens 13 approaches the in-focus position. Since the pulse signal is small, the motor M rotates at a low speed and is controlled so that the motor M stops immediately at the in-focus position.

Hereinafter, steps 108 and 110 of the above-mentioned flowchart
Focusing control according to zooming will be described in detail. In step 108, when the amount of light incident on the array sensor 23 becomes large due to blinking or the like, it is determined that focus control is impossible and the measurement value is re-acquired. I am trying.

By the way, the ocular measurement has a considerably different reflectance depending on the site. For example, the reflectance of the papilla is about 10 times higher than that of the macula. Therefore, if the output of the light source 16 is not controlled so that the maximum value of the memory values of the memories M ₁ and M ₂ obtained by the focus detection system becomes a constant level, the light amount is insufficient in the macula or the light amount in the nipple. The array sensor 23 may become saturated due to excess. Therefore, in order to make the above level constant in the focus detection system, it is possible to some extent by controlling the storage time by the array sensor 23, but if the storage time is controlled, the array sensor 23 will be affected by the dark current. Will reduce the dynamic range of. Therefore, in step 110, a constant level signal is obtained by controlling the lighting time T of the light source 16 via the index projection light source drive circuit 253 while keeping the accumulation time constant, that is, within one accumulation time. As already mentioned, Fig. 13 (a) shows the start pulse, Fig. 13
(b) shows the relationship between the light-off period and the light-on period of the light source 16. For example, during the period T ₁ , the light source 16 when the signal level is high in the focus detection system
Shows the lighting time of. Further, the period T ₂ shows the lighting time of the light source 16 when the level is small. However, in order to actually realize this, there is a limit to the brightness of the light source that can enter the human eye, so that the amplifier 250 that amplifies the electrical signal output from the array sensor 23 is a programmable amplifier capable of gaining, By controlling the gain of the amplifier by the microcomputer 252, a very small signal to a large signal may be realized with a small number of parts in combination with the control of the brightness of the light source and the lighting time.

Next, zooming and zooming will be described. The following ideas were made in the optical design so that the focus can be detected even during zooming and zooming.

When the magnification is changed and the zoom is performed, the focus position on the array sensor 23 and the size of the index image can be known. FIG. 11 shows the movement of the image on the array sensor 23 when the magnification is changed. (Inspected eye O
D) Considering the size of the array sensor 23 sufficiently, the movement due to the diopter of the eye (horizontal displacement of the image on the array sensor 23) at the time of high magnification is larger than that at the time of low magnification. The magnification (determined by the field lens 20 and the re-imaging lens 21) and the size of the index (slit) of the focus detection optical system are determined so as to satisfy the degree (± 15D). It should be noted here that the two slit images divided into the pupils should not be bulky.

The operation of the focus detection with the variable magnification and the smoothness thus configured is as follows.

For example, let us take an example of a case where the variable magnification is 3 times. In that case, the turret and the like
The focusing relay lens 13 shown in the figure is replaced by another focusing relay lens 13
You can replace it with a.

Focusing intervals A, B, and C corresponding to a-times (low-magnification), b-times (medium-times), and c-times (high-magnification) are stored in a microcomputer, and information of magnification change is received from a fundus observation device ( Fig. 10
(Steps 114 and 115 in (b)) The calculation result obtained in Step 113 is compared with the focusing interval at each magnification (steps 116 and 11 in FIG. 10B).
7, 118), and the focus information may be obtained (step 119 in FIG. 10 (b)). Similarly, in the case of zooming, the variation of the focus position at each magnification on the array sensor 23 at the time of zooming is put on a certain curve, which is stored in the microcomputer, and the zoom information is stored. The position at the time of the magnification is obtained from the eye fundus observation device, and the same processing as that described in steps 114 to 119 may be performed.

Here, the method of removing the various noise components N ₁ , N ₂ , and N ₃ was described above, but in the previous method, at least 2 accumulation times were required to perform one processing, so a method of processing in a shorter time there I will describe.

It is an array sensor driving means that can change the accumulation time in two steps, that is, the accumulation time when the light source 16 is turned on.
The tinton is set to the same accumulation time as the 13th (a), that is, T ₂ . In addition, Tintoff is the accumulation time when the light is off, and
We will set it so that intoff <Tinton.
In this way, even if only the signal component is obtained by the above-described method, the accurate signal component cannot be obtained because the length of the accumulation time differs depending on whether the light is on or off. That is, the photoelectric output of the array sensor 23 is proportional to not only the amount of incident light but also the accumulation time. Therefore, the dark output value of the array sensor 23 is measured in advance by the accumulation time Tinton and Tintoff, and the value is stored in the memory as Ninton and Nintoff (output of the array sensor 23 = photoelectric conversion component depending on incident light amount + Dark output component).

In addition, the accumulation time Ti for the same incident light amount of the array sensor 23
The proportional constant K of photoelectric conversion between nton and Tintoff is measured and stored in the memory.

In this way, (Sintoff-Nintoff) × K is calculated for the array sensor data Sintoff read when reading the signal components accumulated in the array sensor when the light is turned off. Lighting data for Sinton (Sinton-Ninton)
And then (Sinton-Ninton)-(Sintoff-Nintoff)
Only the signal component can be obtained by obtaining xK.

If the dark output value can be ignored for the signal component
It is sufficient to obtain (Sinton-Sintoff) × K.
Then, in synchronization with the array sensor driving means, the reference power source Vref of the A / D converter is set at the ratio of the proportional constant K of photoelectric conversion between Tinton and Tintoff, that is, Vrefon and Vrefoff (Vrefon
= Vrefoff × K) or by using two A / D converters with Vrefon and Vrefoff as the reference power supply,
If the / D converter is switched, it is not necessary to multiply the proportional constant K after reading the array sensor output accumulated when the light source 16 is turned off, and further shortening the accumulation time reduces the output value. It is also possible to avoid lowering the dynamic range.

In the above description, the index plate 18 is illuminated from behind by the index projection light source 16, and the slit of the index plate 18 is used as a light emitting index.However, a rectangular light emitter is provided instead of the slit, and this light emitter is used. Of course, it may be used as a light emission index.

Further, in order to change the light quantity of the index, it is possible to use other various light quantity adjusting means, for example, a diaphragm or an ND filter.

(Effects of the Invention) By implementing the present invention as described above, optical noise (noise due to observation light, stray light, etc.) and electrical noise can be removed, and by controlling the light amount of an index, it can be anywhere on the fundus. Accurate focus detection is now possible

[Brief description of drawings]

FIG. 1 is a diagram showing an optical system and an electric block of a fundus camera having an embodiment of the present invention, and FIG. 2 shows only a focus information detection optical system and an observation optical system in FIG. Fig. 3, Fig. 3 is an enlarged plan view of a ring slit, Fig. 4 is an enlarged plan view of an index having an index, Fig. 5 is an enlarged perspective view of a split pupil roof prism, and Fig. 6 is an in-focus state. An enlarged view of the main part of the information detection optical system, FIG. 7 is a view showing the intensity distribution of the optical image on the array sensor, FIG. 8 is a detailed view of the electric block of FIG. 1, and FIG. 9 is an array sensor. 10 (a) and 10 (b) are flowcharts for explaining the operation of the microcomputer used in FIG. 8, and FIG. 11 is a flowchart for explaining the operation of the microcomputer used in FIG. The figure which shows the movement of the slit image (eye OD to be inspected), FIG. 12 (a), (b), (c) shows the signal component + unnecessary noise component.
FIG. 13 (a), unnecessary noise component (b), and signal component (c), respectively, and FIGS. 13 (a) and 13 (b) show the relationship between the start pulse (a), the turn-off period and the turn-on period of the light source 16, respectively. FIG. 14 (b) is a diagram, and FIGS. 14 (a) and 14 (b) are diagrams showing a relationship (b) between the turn-off period and the turn-on period of the light source 16 of the start pulse (a). (Explanation of symbols of main parts) 7 ... dichroic prism, 8 ... relay lens for ring slit, 9 ... perforated mirror, 10 ... aperture stop, 11 ... objective lens, 13 ... relay lens, 13a ... …
Variable magnification relay lens, 14 …… dichroic mirror, 16…
Light source for index projection, 17a, 17b ... Relay lens for index projection system, 18 ... Index plate, 19 ... Field stop, 20 ... Field lens, 21 ... Reimaging lens, 22 ... Split pupil roof Shape prism, 23 ... Array sensor, 24 ... Driving means, 25 ...
… Computing means, 26 …… Motor driving means, M …… Motor, 252 …… Microcomputer, 253 …… Light source driving circuit for index projection.

Claims (1)

[Claims] 1. An index projection optical system for projecting a luminescent index onto the fundus of the eye to be inspected and a photoelectric converter for receiving the index of the fundus to be inspected, the focus information being detected by the output of the photoelectric converter. In a focus detection device of a fundus oculi observation device including a focus information detection optical system, selection means for selecting whether or not to project the index, and a first signal obtained from the photoelectric converter when projecting the index. And a second subtraction signal obtained from the photoelectric converter when the index is not projected, subtraction means for outputting a focus signal from the difference signal, and the selection means projecting the index. Sometimes, when the signal obtained from the photoelectric converter is above a predetermined level, the focus detection device is provided with a control means for reducing the light amount of the index so as to be below the level. .