JPH02204715A - Image pickup device having optical low pass filter - Google Patents

Abstract

PURPOSE:To effectively prevent the generation of color moire in a color single plate solid state element having offset sampling structure by attaining optical low pass filter characteristics by three crystal plates whose light dividing directions are respectively +45 deg., +90 deg. and -45 deg. directions against a scanning direction. CONSTITUTION:An optical low pass filter 10 is constituted of 1st to 3rd optical members 11 to 13 each of which consists of a crystal double reflection plate and the light dividing directions of respective members are 45 deg., 90 deg. and 135 deg. in the counterclockwise direction from the scanning direction. An incident light beam 14 is successively divided into 45 deg., 90 deg. and 135 deg. components in the counterclockwise direction by the optical members 11 to 13 and eight light beams are finally projected, so that required space frequency characteristics are obtained. The ideal shape of the optical low pass filter for offset sampling is a hexagon, which constitutes a low pass filter system for setting up its frequency characteristics to '0' on a line constituted of a straight line in the multiple direction of 45 deg. on the frequency plane.

Description

Detailed Description of the Invention (Industrial Field of Application) The present invention relates to an imaging device having an optical low-pass filter, and particularly to a video camera using a color solid-state imaging device,
The present invention relates to an imaging device having an optical low-pass filter suitable for processing images two-dimensionally and discretely, such as in an electronic still camera.

(Prior Art) In general, an imaging device using a solid-state imaging device having a discrete pixel structure spatially samples an optically formed image to obtain an output image.

FIG. 3 shows the configuration of a stripe-shaped color filter for obtaining color signals in a conventional single-chip color video camera. Each color filter corresponds to a cell of the solid-state image sensor, and the three colors of RGB are separated by this.

A problem with discretely sampling an image is aliasing error when the subject contains frequency components higher than the sampling spatial frequency. This error causes moire fringes and false colors in the captured image.
This will cause the image to become distorted. In particular, in a single-chip camera, spatial sampling in the horizontal direction is performed once every three pixels, resulting in a coarse sampling pitch, which tends to cause color moiré.

The processing system after the image sensor is originally configured on the premise that there is a certain limit on the spatial frequency band of the image.

For this reason, conventionally, an optical low-pass filter has been inserted into the imaging system to limit the high-frequency spatial frequency band of the subject in accordance with the system after the imaging device. Optical low-pass filters that utilize birefringence, such as quartz plates, are often used.

For example, in a single-panel color video camera, a ray of light passes through one crystal plate and is split into two by birefringence, and the horizontal direction is 1.5.
The optical system is configured so that the light enters the position shifted from the pixel. The fact that the same images overlap in this way with a 1.5 pixel horizontal shift means that the horizontal frequency characteristic of this crystal low-pass filter becomes a cosine-type frequency characteristic as shown in Figure 4. There is. If the pitch of one pixel is d, if there is a sine wave chart with a pitch of 3d, it will be 1.5
Due to pixel misalignment, the peak and valley exactly overlap, resulting in a contrast of O.

The sampling frequency fs per pixel is fXi = 1/d
For this reason, the characteristic of this crystal low-pass filter becomes 0 at fs/3, as shown in FIG. In the case of a single-chip color solid-state image sensor, sampling is performed once every three pixels, so color moiré is reflected from this point. Therefore, such a filter had a great effect on suppressing moiré.

(Problems to be Solved by the Invention) On the other hand, in recent years, solid-state imaging devices in which each pixel is configured in an offset sampling configuration and RGB three-color filters are arranged in each pixel have been attracting attention.

What is shown in FIG. 5 is an example of a color solid-state image sensor having a conventional offset sampling structure.

However, in this case, the color filter arrangement is two-dimensional rather than a simple one-dimensional arrangement like the conventional stripe-shaped one, so the conventional one was simply replaced with a crystal low-pass filter to prevent color moiré. A problem has arisen in that the desired effect cannot be obtained even when used.

The present invention relates to a new crystal low-pass filter that can be applied to the above-mentioned problems specific to solid-state imaging devices with an offset sampling configuration, and relates to an imaging device having an optical low-pass filter using the crystal filter.

(Means for solving the problem) Therefore, the present invention focuses on the two-dimensional spatial frequency structure of the signal when the color filter arrangement of offset sampling as shown in Fig. 5 is adopted, and the present invention is optimized to prevent color moiré. By adding an optical low-pass filter to the imaging system,
It is characterized by solving the above problem. In order to explain the conditions of the low-pass filter, we will first analyze the spatial frequency structure in the case of offset sampling.

FIG. 6 shows an example of offset sampling configured such that the horizontal pitch is PH, the vertical pitch is Pv, and the horizontal offset is PH/2. It is known that the sampling structure in this case is as shown in FIG. 7 on a two-dimensional spatial frequency plane. That is, the sampling structure is such that the pitch in the horizontal direction is 2/PH, the pitch in the vertical direction is 1/PV, and the offset in the horizontal direction is 17P)l. The derivation of this relationship can be found, for example, in "Digital Signal Processing of Images" (1985) by Takahiko Fukisaka, published by Nikkan Kogyo Shimbun, p.
317.

Therefore, if a luminance signal is obtained by the so-called switch-Y method, in which the color signal output from the color filter of each pixel is equivalently regarded as a luminance signal in a sensor as shown in FIG. Recopy the diagram and it will look like the circle mark (O mark) in Figure 2. Looking at this for each color signal, the horizontal pitch force 3 P
It can be seen that M has an offset sampling structure in which the pitch in the vertical direction is Pv and the offset in the horizontal direction is -PH. Therefore, the sampling structure of the color signal is reduced by 1/3 only in the horizontal direction in the spatial frequency plane, resulting in something like the square (0 mark) in FIG. 2.

The optimal optical low-pass filter for preventing color moiré is one in which the spatial frequency spectrum of the original signal has as little influence as possible on adjacent regions of the sampling structure. In Figure 2, this means that it is preferable to have a low-pass filter characteristic in which the frequency characteristic becomes 0 or a minimum near the spatial frequency part on the side of the hexagon formed by connecting the center points of adjacent sampling structures with solid lines. It means that.

Note that the reason why the two horizontal lines (Pv), fX=±-·□, are used to prevent the brightness from folding back in the vertical direction is to be avoided.

(Embodiment) FIGS. 1A and 1B are concrete configuration diagrams of an embodiment of an optical low-pass filter according to the present invention.

In the figure, 10 is an optical low-pass filter 11°12.
Reference numerals 13 denote optical members 1, 2, and 3, which are all made of quartz birefringent plates. The beam splitting direction is 45°900.135° counterclockwise with respect to the scanning direction.

FIG. 1(A) is an explanatory diagram showing the light rays incident on the optical low-pass filter as seen from a plane perpendicular to the scanning direction, and FIG. 1(B) is an explanatory diagram showing the direction in which the light rays are shifted as seen from the incident plane.

The incident light beam 14 is sequentially
, ■, and 0, and are emitted as eight light rays. As a result, desired spatial frequency characteristics are obtained.

The ideal optical low-pass filter for offset sampling is a hexagonal shape as shown in Figure 2, but in actual crystal low-pass filters, the direction in which the light rays are shifted is 45 degrees.
If it is a multiple of , the system configuration is easy and easy to realize.

This is due to the polarization characteristics of the crystal, and setting the angle to a multiple of 45° is a matter of balancing the light intensity ratio of the light beams divided by the crystal.

Therefore, the present invention is characterized by configuring a low-pass filter system in which the frequency characteristic becomes O on a line formed by straight lines in directions that are multiples of 450 on the frequency plane.

The first embodiment is shown by a solid line in FIG. 1(C).
This is an example of a configuration of a crystal plate low-pass filter in which a line extended in the 45° direction and fXf=wavenumber characteristics are 0 or minimal.

Satisfying such characteristics can be easily achieved by controlling the amount of shift by the crystal plate.

Since the hexagon is arranged symmetrically with respect to the origin in the spatial frequency plane, and the hexagon is composed of a group of straight lines with three directions of o0 ± 45° with respect to the fX axis,
The low-pass filter indicated by the solid line in FIG. 1(C) can be realized by a configuration of a total of three crystal filters corresponding to the directions of the sides of the hexagon.

The conditions for the crystal filter are 6 points from the origin on the spatial frequency plane.
It can be easily determined by calculating the length of the perpendicular line drawn to each side of the rectangle. For example, in order to realize a crystal low-pass filter whose frequency characteristics become 0 on such a straight line, it is sufficient to construct a crystal plate in real space. This method of determination is the same as that for conventional stripe types. Similarly, the corresponding low-pass filter may be one that is shifted by Pv in the vertical direction in real space. Therefore, the low-pass filter shown by the solid line in Figure 1 is as shown in Figure 8 (^), (
This can be realized by configuring two crystal filters in the 45° direction in real space as shown in B) and (C).

In this low-pass filter, if PH and Pv are equal, the spatial frequency spread in the horizontal direction is larger than in the vertical direction, as shown in FIG. 1(C). If the vertical shift amount is made slightly smaller than Pv, the vertical resolution will improve and the difference between the two will become smaller. However, since it deviates from the desired low-pass filter characteristics, it has the side effect of increasing color moiré. A small amount of shift also has some effect on brightness. The design of the low-pass filter should basically be determined by looking at the entire system and making a trade-off between resolution and color correction.

In this embodiment, the image element has an offset sampling structure in which the horizontal pitch PH1, the vertical pitch PV, and the horizontal offset amount are PH/2. 4 clockwise
A first optical member having a function of dividing a light beam incident in a direction of 50 into two trees and a direction of 90° clockwise or counterclockwise with respect to the scanning direction of the image sensor or a direction opposite to the scanning direction. A second optical member has a function of splitting the light ray into two, and the light ray that is incident in a direction of 90° clockwise or counterclockwise with respect to the light ray splitting direction of the first optical member is split into two. An optical low-pass filter composed of a third optical member having a function, the first optical member constituting the optical low-pass filter has a beam splitting width of 21, and the third optical member has a beam splitting width of 21. When set to P3,
We try to satisfy the following conditions.

Also, the vertical pitch Pv is Pv ≧ P.H. When ・・・・・・・・・・・・-(3) ・・・・・・・・・・・・-(4) We try to satisfy the following conditions.

If the upper limits of these conditional expressions are exceeded, moiré will increase, which is not good, and if the lower limits are exceeded, the resolution will decrease, which is not good.

(Other Embodiments) In the second embodiment, the frequency characteristics are 0 or minimum on a hexagon as shown by the dotted line in FIG. 1(C). When the ideal shape shown in FIG. 2 is approximated by straight lines of ±45° and 00, this corresponds to choosing a different point from the first embodiment for the same sampling structure. There is a point on the spatial frequency plane through which a straight line of ±45° passes, and the lengths of perpendiculars drawn from the origin to the four sides of ±45° are all equal. This can be achieved by combining two crystal filters.

In the second embodiment, the overall resolution is improved compared to the first embodiment in which the area of the hexagon is larger in terms of spatial frequency, but as a side effect, color moire increases.

Therefore, in practice, an intermediate combination between the first and second embodiments may be selected by making a trade-off between the two. Since the amount of shift of the light beam in the crystal is proportional to the thickness of the crystal plate, the plate thickness in this case exists in a region exactly between the first and second embodiments.

The third example is a case where the first example and the second example match. This means that in the spatial frequency domain, for a straight line where the length of the perpendicular drawn to the side of a straight line of ±45° is equal or 0°,
This means that the same thing as in the first embodiment holds true. To summarize the above equation, this simply means: 2PV=3PM. Therefore, if the horizontal pitch and vertical pitch of the pixels of the sensor are set to 2=3, the diagonal line portion showing the ideal low-pass filter for preventing color moiré shown in FIG. 2 will be exactly ±450. As a result, an optimal optical low-pass filter for preventing color moiré can be realized under the condition of ±45°, which is preferable when a crystal plate is used, and a systematic fusion of the two can be achieved. In this case as well, it is sufficient to use a three-piece crystal filter that is shifted.

The present invention is based on the analysis of two-dimensional sampling structures. Therefore, the present invention applies to R,G.

In addition to the three-color arrangement of B, there is also Cy.

A combination of M, Y, etc. can be said to be an equally effective means.

(Effects of the Invention) As explained above, in the present invention, based on characteristic analysis in the spatial frequency range of a color single-plate solid-state element with an offset sampling structure, a crystal low-pass filter that can effectively prevent color moiré has been constructed. Successful.

In the present invention, optical low-pass filter characteristics are realized using three crystal plates that are shifted by predetermined amounts in the ±450, vertical, and −45° directions, respectively, depending on the amount determined from the sampling pitch.

If the filter of the present invention is applied to a photographing system, it will be possible to deal with color moiré in a color solid-state image sensor with an offset sampling structure, which could not be solved with conventional configurations.

Furthermore, since the filter of the present invention has a simple configuration of three crystal plates, it can be easily replaced with a conventional system. Further, the configuration of the image sensor according to the analysis of the present invention is highly effective in the sense that an ideal low-pass filter can be realized.

[Brief explanation of the drawing]

Figures 1 (A) and (B) are schematic diagrams of the main parts of the optical low-pass filter according to the present invention, and Figure 1 (C) shows the two-dimensional spatial frequency characteristics of the optical low-pass filter made of crystal according to the present invention. Explanatory diagram: The solid line is the first example, the dotted line is the second example, FIG. 2 is an explanatory diagram showing the two-dimensional spatial frequency characteristics of an ideal crystal low-pass filter, and FIG. 3 is the conventional RGB
Schematic diagram showing the configuration of the stripe color filter, No. 4
The figure is an explanatory diagram showing the spatial frequency characteristics of a conventional crystal low-pass filter used in the RGB stripe color filter of Fig. 3, and Fig. 5 is an explanatory diagram showing the spatial frequency characteristics of a conventional crystal low-pass filter used in the RGB stripe color filter of Fig. 3. 6 is a diagram showing the offset sampling structure in real space, FIG. 7 is a diagram showing the sampling structure of offset sampling in the spatial frequency domain, and FIG. 8 is a diagram showing the crystal low-pass filter of the first embodiment. It is an explanatory view showing a shift of each light ray in three constituent crystal plates. In the figure, 10 is an optical low-pass filter 11.12.1
3 is number 3! , 2nd. This is the third optical member.

Claims (6)

[Claims] (1) In an imaging device having a color solid-state imaging device with an offset sampling structure, an imaging device equipped with an optical low-pass filter characterized by being equipped with a three-piece crystal low-pass filter that shifts light beams in different directions. Device. (2) The color solid-state image sensor with the offset sampling structure has a pitch in the horizontal direction of P_H, an offset of P_H/2, a pitch in the vertical direction of P_V, and a spatial frequency plane (f_X, f_
2. The frequency characteristic according to claim 1, wherein the frequency characteristic is zero or minimum on the side of a hexagon surrounded by four straight lines in the ±45° direction and two straight lines parallel to the f_X axis. An imaging device with an optical low-pass filter. (3) The coordinates of the vertices formed by the sides in the ±45° direction of the hexagon exist symmetrically on the f_X axis with respect to the f_Y axis, and on the negative side of f_X from -2/3P_H to -(
1/3・1/P_H+1/2・1/P_V), f
On the positive side of _X (1/3・1/P_H+1/2・1/
3. The imaging device according to claim 2, further comprising a crystal low-pass filter that is present between P_V) and 2/3P_H. (4) The imaging device having an optical low-pass filter according to claim 3, wherein 2P_V=3P_H holds true in the color solid-state imaging device having the offset sampling structure. (5) Horizontal pitch P_H, vertical pitch P_
V, has an offset sampling structure in which the horizontal offset amount is P_H/2, and P_V≦2/3P_
H, and a first image sensor having a function of dividing into two light beams incident in a direction of 45° clockwise or counterclockwise with respect to the scanning direction of the image sensor or a direction opposite to the scanning direction.
a second optical member having a function of dividing into two light beams incident in a clockwise or counterclockwise direction at 90 degrees with respect to the scanning direction of the image pickup device or the direction opposite to the scanning direction; The optical low-pass filter is composed of a third optical member that has the function of splitting a light beam incident in a direction of 90 degrees clockwise or counterclockwise with respect to the light beam splitting direction of the first optical member into two beams. Then, the beam splitting width of the first optical member constituting the optical low-pass filter is P_1.
, when the beam splitting width of the third optical member is P_3, 3P_H/2√2≦P_1≦(3√2P_H・P_V)
/(3P_H+2P_V)3P_H/2√2≦P_3≦
An imaging device having an optical low-pass filter that satisfies the following condition: (3√2P_H·P_V)/(3P_H+2P_V). (6) Horizontal pitch P_H, vertical pitch P_
V, an image sensor having an offset sampling structure in which the offset amount in the horizontal direction is P_H/2, and P_V=>3/2P_H, and a clockwise direction or a direction opposite to the scanning direction of the image sensor. A first optical member having a function of dividing a light beam incident in a direction of 45 degrees counterclockwise into two beams and a direction 90 degrees clockwise or counterclockwise with respect to the scanning direction of the image sensor or a direction opposite to the scanning direction. A second optical member has a function of splitting a light beam incident in the direction of 90 degrees into two beams, and the first optical member splits the light beam incident in the direction of 90 degrees clockwise or counterclockwise with respect to the light beam splitting direction of the first optical member. An optical low-pass filter composed of a third optical member having a function of dividing into two parts,
When the beam dividing width of the first optical member constituting the optical low-pass filter is P_1, and the beam dividing width of the third optical member is P_3, (3√2P_H・P_V)/(3P_H+2P_V)≦
P_1≦3P_H/2√2(3√2P_H・P_V)/
An imaging device having an optical low-pass filter that satisfies the following condition: (3P_H+2P_V)≦P_3≦3P_H/2√2.