Imaging device and imaging method

Abstract

To provide an imaging system capable of improving the parallax performance of a stereo camera using HDR technology.SOLUTION: The imaging system has multiple cameras. Each camera has an imaging device with unit pixels arranged periodically including a high-sensitivity subpixel that outputs a first output value for a certain amount of exposure and a low-sensitivity subpixel that outputs a second output value lower than the first output value for a certain amount of exposure: a synthesis unit that is configured so as to, when the exposure amount is smaller than a predetermined exposure value, select and output an output of the high sensitivity subpixel, when the exposure amount is greater than a predetermined exposure value, select and output an output of the low sensitivity subpixel; and an amplification unit that amplifies the output of the synthesis unit and outputs the same. In the imaging system, the predetermined exposure value is the same among the multiple cameras and the amplification unit is configured to correct image signals for the high-sensitivity subpixel and the low-sensitivity subpixel so that in the same predetermined exposure value, the image signals of the high-sensitivity subpixel and the low-sensitivity subpixel are the same between multiple cameras.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an imaging apparatus and imaging method having a function of expanding a dynamic range by combining imaging signals with different exposure amounts or sensitivities. [Background technology]

[0002] Stereo cameras are known as a type of device for measuring the distance to an object. Stereo cameras are devices that measure the distance to an object based on trigonometry, utilizing the difference in the position of object images (parallax) in images captured by multiple cameras placed at different positions. These stereo cameras are installed in automobiles and other vehicles and are used in in-vehicle sensing technology to detect the positions of obstacles around the vehicle. In-vehicle sensing technology is required to have imaging characteristics with a wide dynamic range, as it is used under a variety of ambient lighting conditions, such as tunnel entrances and exits, stop lights of preceding vehicles at night, and LED-type signs.

[0003] Also, in the technical field of imaging devices, high dynamic range (hereinafter referred to as "HDR") technology is known that combines pixel signals from multiple captured pixels to expand the dynamic range of a captured image. For example, Non-Patent Document 1 discloses a technology for acquiring an HDR image by combining signals from multiple sub-pixels with different sensitivities into a single pixel signal. In the HDR combination in Non-Patent Document 1, a predetermined threshold (combining point output value) is set for the output value of the high-sensitivity sub-pixel, and when the output value of the high-sensitivity sub-pixel is smaller than the combination point output value, the output signal of the high-sensitivity sub-pixel is selected, and when the output value is greater than the combination point, the output signal of the low-sensitivity sub-pixel is amplified to generate an HDR signal.

[0004] Here, the maximum number of gradations of the generated HDR image signal will be larger than the maximum number of gradations of the individual sub-pixel signals before synthesis, and if this continues, it will be impossible to perform image calculations at the desired processing speed or to display the image on an image display device with a fixed maximum input gradation value.As a method for solving this problem, nonlinear compression is known to achieve an appropriate maximum number of gradations. [Prior art documents] [Non-patent literature]

[0005] [Non-Patent Document 1] IEDM 2018,A 0.68e-rms Random-Noise 121dB Dynamic-Range Sub-pixel architecture CMOS Image Sensor with LED Flicker Mitigation Summary of the Invention [Problem to be solved by the invention]

[0006] In a stereo camera, to calculate disparity from the difference in the position of an object image between images captured by multiple cameras, the object image (or part of the object image) of interest is compared between the images, and if they match, the difference in the position of the object image (or part of the object image) between the images is calculated in pixel units. Therefore, the images (or parts of the object) of the object captured by multiple cameras must be identical except for the pixel position. Therefore, lenses with uniform resolution characteristics are used for each camera, geometric correction is performed to correct geometric distortion caused by the lenses, and sensitivity correction is performed to align the sensitivity characteristics of the imaging devices.

[0007] However, with HDR technology that uses subpixels with different sensitivities, the output value of the synthesis point that switches between subpixels with different sensitivities shifts when correcting the sensitivity between multiple cameras, making it impossible to completely match the sensitivity characteristics between multiple cameras, resulting in a problem of reduced parallax performance.

[0008] Therefore, an object of the present invention is to provide an imaging device and an imaging method that can match the sensitivity characteristics of all cameras without causing misalignment of the synthesis points between cameras, thereby achieving good parallax performance. [Means for solving the problem]

[0009] In order to solve the above problem, the imaging device of the present invention is an imaging device having a plurality of cameras, each of which has an imaging device in which unit pixels each consisting of a high-sensitivity sub-pixel that outputs a first output value for a certain exposure amount and a low-sensitivity sub-pixel that outputs a second output value lower than the first output value for the certain exposure amount are periodically arranged, a synthesis unit that selects and outputs the output of the high-sensitivity sub-pixel when the exposure amount is smaller than a predetermined exposure value, and selects and outputs the output of the low-sensitivity sub-pixel when the exposure amount is greater than the predetermined exposure value, and an amplifier that amplifies and outputs the output of the synthesis unit, wherein the predetermined exposure value is the same for the plurality of cameras, and the amplifier corrects the image signals of the high-sensitivity sub-pixel and the low-sensitivity sub-pixel so that the image signals of the high-sensitivity sub-pixel and the low-sensitivity sub-pixel are the same for the same predetermined exposure value between the plurality of cameras. [Effects of the Invention]

[0010] According to the imaging device or imaging method of the present invention, it is possible to match the sensitivity characteristics of all cameras without causing misalignment of the synthesis points between the cameras, and to achieve good parallax performance. [Brief explanation of the drawings]

[0011] [Figure 1] 3A and 3B are schematic diagrams illustrating pixel structures of the conventional imaging device and the imaging device according to the first embodiment. [Figure 2] 1A and 1B are diagrams for explaining generation of an HDR pixel signal from signals of two sub-pixels with different sensitivities. [Figure 3] 1 shows the configuration of an imaging device according to a first embodiment. [Figure 4] 1 shows an HDR composition and sensitivity correction method performed by the imaging device of the first embodiment. [Figure 5] HDR compositing and sensitivity correction method using a conventional imaging device. [Figure 6] 1 is a flowchart of HDR composition and sensitivity correction according to the first embodiment. [Figure 7] HDR composition and sensitivity correction method using the imaging device of Example 2 [Figure 8] 10 is a flowchart of HDR composition and sensitivity correction according to the second embodiment. [Figure 9] HDR composition and sensitivity correction method using the imaging device of Example 3 DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, an embodiment of an imaging device according to the present invention will be described with reference to the drawings. [Example]

[0013] First, an image capturing apparatus 100 according to a first embodiment of the present invention will be outlined with reference to Fig. 3. As shown in the figure, the image capturing apparatus 100 of this embodiment is a stereo camera that outputs parallax images calculated based on a left image PL captured by a left camera 1L and a right image PR captured by a right camera 1R. Note that in Fig. 3, the reference numerals of the components of the left camera 1L are suffixed with L and the reference numerals of the components of the right camera 1R are suffixed with R, but since the functions of the corresponding components are basically the same, hereinafter, unless otherwise necessary, the suffixes L and R will be omitted and generalized reference numerals will be used to describe the details of this embodiment.

[0014] For imaging with the left camera 1L and the right camera 1R, solid-state imaging devices such as CCD (Charge Coupled Device) and CMOS (Complementary Metal Oxide Semiconductor) (hereinafter referred to as "imaging device 11") are used. Fig. 1(a) is a schematic diagram of the pixel structure of a conventional imaging device 11', and Fig. 1(b) is a schematic diagram of the pixel structure of the imaging device 11 of this embodiment. In any of the imaging devices, four unit pixels of red (R), green (G1, G2), and blue (B) are periodically arranged. However, in the imaging device 11 of this embodiment, each unit pixel is composed of two types of sub-pixels: a high-sensitivity sub-pixel 11a with a large area and a low-sensitivity sub-pixel 11b with a small area. Therefore, in the imaging apparatus 100 of this embodiment, HDR can be realized by synthesizing the output signals from the two types of sub-pixels with low sensitivity and high sensitivity into one as the output signal of the unit pixel.

[0015] <Principle of HDR Synthesis> Fig. 二 shows the principle of HDR synthesis of signals from two types of sub-pixels with low sensitivity / high sensitivity. The graphs in Figs. 2(a) to (c) all show the sensitivity characteristics with the exposure amount on the horizontal axis and the output value of the pixel signal or sub-pixel signal on the vertical axis.

[0016] As shown in Figure 2(a), the output characteristics of the subpixels before blending are a straight line with a steep slope for the high-sensitivity subpixel 11a and a shallow slope for the low-sensitivity subpixel 11b. The maximum output value (saturation output value) of the subpixels is determined by the characteristics of the photodiode and AD converter of the imaging device 11. For example, in a CMOS imaging device using a 12-bit AD converter, the saturated output value is 4095 gradations. In HDR blending, a blending point is determined on the sensitivity line of the high-sensitivity subpixel 11a, and the signal of the low-sensitivity pixel is multiplied by a predetermined gain so that the output of the high-sensitivity subpixel 11a and the output of the low-sensitivity subpixel 11b form a smooth straight line. The blending point is determined on the sensitivity line of the high-sensitivity subpixel 11a, and the gain is multiplied only on the low-sensitivity pixel signal because the gain value applied to the subpixel signal in the subsequent stage is greater than 1. With the synthesis point as the boundary, the high-sensitivity sub-pixel 11a is responsible for the subject with a small exposure amount to the left of the synthesis point, and the low-sensitivity sub-pixel 11b is responsible for the subject with a large exposure amount to the right of the synthesis point, so that the dynamic range of the imaging device 11 is essentially expanded.

[0017] In this case, as shown in Figure 2(b), the maximum number of gradations in the combined HDR image signal will be larger than the maximum number of gradations in each of the individual subpixel signals before combination, which means that image calculations cannot be performed at the desired processing speed or the image cannot be displayed on an image display device with a fixed maximum input gradation value. Therefore, as shown in Figure 2(c), the combined output is nonlinearly compressed to an appropriate maximum number of gradations before being used in subsequent processing (such as distance measurement).

[0018] <Details of the Imaging Device 100> Here, referring again to Fig. 3, the imaging device 100 according to this embodiment will be described in detail. As shown in the figure, the imaging device 100 is a device including a left camera 1L, a right camera 1R, and a parallax image generation unit 50. The left camera 1L also includes an imaging unit 10L, a lens 20L, a geometric correction unit 30L, and a register 40L. Similarly, the right camera 1R also includes an imaging unit 10R, a lens 20R, a geometric correction unit 30R, and a register 40R. Each component will be described in detail below.

[0019] The parallax image generating unit 50 takes in the left image PL and the right image PR captured synchronously from the left camera 1L and the right camera 1R, and generates and outputs parallax images through stereo matching processing in the stereo matching unit 51.

[0020] The imaging units 10 of the left and right cameras 1 have an imaging device 11, a gain amplifier 12, an HDR synthesis unit 13, a gain amplifier 14, and a signal compression unit 15.

[0021] The imaging device 11 has a pixel structure consisting of high-sensitivity sub-pixels 11a and low-sensitivity sub-pixels 11b (see FIG. 1(b)), and therefore generates a high-sensitivity sub-pixel signal Sa and a low-sensitivity sub-pixel signal Sb for each unit pixel from an object imaged on the light-receiving surface via the lens 20.

[0022] As described in FIG. 2, the HDR synthesis unit 13 determines the level of the high-sensitivity sub-pixel signal Sa and performs HDR synthesis. Specifically, if the level of the high-sensitivity sub-pixel signal Sa is lower than a predetermined synthesis point (see FIG. 2(a)) provided by the register 40, the HDR synthesis unit 13 outputs the high-sensitivity sub-pixel signal Sa as is as a synthesis signal. On the other hand, if the level of the high-sensitivity sub-pixel signal Sa is higher than the synthesis point, the HDR synthesis unit 13 outputs a signal obtained by multiplying the low-sensitivity sub-pixel signal Sb by a predetermined sub-pixel synthesis gain using the gain amplifier 12 as a synthesis signal (see FIG. 2(b)). The sub-pixel synthesis gain is a gain corresponding to the sensitivity ratio between the low-sensitivity sub-pixel 11b and the high-sensitivity sub-pixel 11a, and is registered in advance in the register 40.

[0023] The composite signals output by the HDR composition units 13 of the left and right cameras are corrected by the respective gain amplifiers 14 to correct for the difference in sensitivity between the left and right cameras. The correction gain values ​​used here are also pre-registered in the register 40. Then, the signal compression unit 15 performs signal compression to suppress the maximum number of gradations (see FIG. 2(c)).

[0024] The signals output from the left and right imaging units 10 are input to the left and right geometric correction units 30, respectively. Generally, each camera has its own inherent distortion. This distortion can be caused by various factors, such as misalignment between the imaging device 11 and the lens optical axis or distortion of the lens 20. In triangulation using a stereo camera, it is important that the left and right images are precisely aligned parallel to each other in order to check the degree of match between small image regions (matching blocks) in the left and right camera images. Therefore, the geometric correction unit 30 not only corrects the distortion but also parallelizes the left and right images using geometric correction information provided by the register 40. The left and right image signals geometrically corrected by the geometric correction unit 30 are input to the parallax image generation unit 50, which generates parallax images as described above.

[0025] Next, the HDR compositing of sub-pixel signals and the sensitivity correction of the left and right cameras according to this embodiment will be described using Fig. 4. The graphs in Fig. 4(a) to (c) all show sensitivity characteristics with the exposure amount on the horizontal axis and the output value of the pixel signal or sub-pixel signal on the vertical axis, and show (a) the state before compositing high-sensitivity and low-sensitivity sub-pixel signals, (b) the state after compositing, and (c) the state after sensitivity correction of the left and right cameras.

[0026] 4(a) illustrates a case where the subpixel sensitivity characteristics of the imaging device 11L of the left camera 1L are higher than the subpixel sensitivity characteristics of the imaging device 11R of the right camera 1R. As shown here, in this embodiment, the synthesis points of the high-sensitivity and low-sensitivity subpixels are first determined so that the exposure amount is the same for the left and right cameras. In this example, since the synthesis point output values ​​are not the same when the exposure amounts of the left camera 1L and the right camera 1R are the same, the synthesis point output values ​​are determined so that they differ by the difference in sensitivity between the left and right high-sensitivity subpixels 11a.

[0027] Furthermore, in FIG. 4(b) after composition by the HDR composition unit 13 of each camera, the deviation of the composition points on the left and right composition lines appears as a deviation in the output values ​​corresponding to the difference in sensitivity between the left and right. In contrast, as shown in FIG. 4(c) after amplification by the gain amplifier 14R of the right camera 1R, when sensitivity correction is performed on the composition line of the right camera 1R, not only do the left and right sensitivity lines match, but the left and right composition points also match. As a result, the types of subpixels selected by the left and right cameras always match, allowing the parallax image generation unit 50 to generate appropriate parallax images.

[0028] For comparison, a conventional method for performing HDR blending of subpixel signals based on output will be described with reference to FIG. 5. As shown in FIG. 5(a), in the conventional method, the blending point is determined not by the exposure amount but so that the left and right cameras have the same output value. In this case, as shown in FIG. 5(b) after blending, a difference in the blending point exposure value on the left and right blending lines appears as a difference in the exposure amount. Even if sensitivity correction is performed on FIG. 5(b) as shown in FIG. 5(c), the left and right sensitivity lines will match, but the left and right blending points will be misaligned. As a result, in the section between the left and right blending points, the types of subpixels selected by the left and right cameras will differ, increasing the possibility of a deterioration in the accuracy of the parallax image generated by the parallax image generation unit 50.

[0029] <How to register parameters to register 40> Next, a method for registering various parameters in the register 40 of this embodiment will be described with reference to the flowchart of Fig. 6. The subject of each step described below is a parameter registration operator or a parameter registration operation device.

[0030] First, in step S11, the sensitivity characteristics of the low-sensitivity and high-sensitivity subpixels of the left camera 1L are acquired. To acquire the sensitivity characteristics, a white light source with strictly controlled brightness is used, and the exposure amount is changed by changing the shutter value on the camera to acquire the gradation value of the captured image. From the changed exposure amount and the acquired gradation value, the sensitivity characteristics of the low-sensitivity subpixel 11b and the high-sensitivity subpixel 11a can be acquired.

[0031] Next, in step S12, a gain value for combining the sub-pixels into one straight line is calculated from the sensitivity ratio between the low-sensitivity sub-pixel 11b and the high-sensitivity sub-pixel 11a.

[0032] In step S13, the parameter (synthetic gain value) calculated in step S12 is registered in the register 40L of the left camera 1L.

[0033] In steps S14 to S16, the same operations as in steps S11 to S13 are performed for the right camera 1R. Note that the order of steps S11 to S13 and steps S14 to S16 may be reversed.

[0034] In step S17, the exposure amount for combining the sub-pixels is determined. This value may be determined in advance before the sub-pixel sensitivity characteristics are calculated, or may be determined from the sensitivity characteristics of the left and right sub-pixels after their sensitivity characteristics are obtained.

[0035] In step S18, it is determined what output value to use to composite the sub-pixels of the left camera 1L for the composite point exposure amount determined in step S17 (calculation of the sub-pixel switching point).

[0036] In step S19, the parameter (subpixel switching point) determined in step S18 is registered in the register 40L of the left camera 1L.

[0037] In steps S1a and S1b, the same operations as in steps S18 and S19 are performed for the right camera 1R. Note that the order of steps S18 and S19 and steps S1a and S1b may be interchanged.

[0038] In step S1c, a sensitivity correction value gain for correcting the difference in sensitivity between the left and right cameras is calculated. This operation may be performed after obtaining the left and right sub-pixel sensitivities.

[0039] Finally, in steps S1d and S1e, the sensitivity correction gains calculated in step S1c are registered in the registers 40 of the left and right cameras.

[0040] The above explanation has been given for two cases of sub-pixels, low and high sensitivity, but in cases such as low, medium and high sensitivity, if the sensitivity is divided into low and medium sensitivity, or medium and high sensitivity, the principle is the same, and there is no limit to the number of sub-pixels.

[0041] According to the imaging device of this embodiment described above, it is possible to match the sensitivity characteristics of all the cameras without causing misalignment of the synthesis points between the cameras, and to achieve good parallax performance. [Example]

[0042] Second Embodiment An imaging device 100 according to a second embodiment of the present invention will be described below with reference to Fig. 7. Note that a duplicated description of points common to the first embodiment will be omitted.

[0043] The graphs in Figures 7(a) to (c) show the high-sensitivity and low-sensitivity subpixel signals before and after synthesis, and after sensitivity correction. Figure 7 illustrates a case where the sensitivity characteristics of the imaging devices 11 used are not uniform and vary from the lowest sensitivity to the highest sensitivity. In this embodiment, the sensitivity variation range of the imaging devices 11 is estimated in advance, and the high-sensitivity subpixel 11a of the highest sensitivity device or a slightly higher sensitivity characteristic is set as the target synthesis line for all cameras. Next, the synthesis point exposure amount is determined so that the low-sensitivity and high-sensitivity subpixels of all cameras are synthesized with the same exposure amount.

[0044] As a result, as shown in Figure 7(b) after compositing, the output values ​​of the composite points from all cameras are not the same, but they are composited with the same exposure, and the composite output value appears to be shifted by just the difference in sensitivity.In contrast to this, as shown in Figure 7(c), if you perform sensitivity correction by applying gain so that the sensitivity characteristics of all cameras become the target composite line, all composite points can be made the same point.

[0045] <How to register parameters to register 40> Next, a method for registering various parameters in the register 40 of this embodiment will be described with reference to the flowchart of Fig. 8. The subject of each step described below is a parameter registration operator or a parameter registration operation device.

[0046] First, in step S21, a target composite line is determined, which is a common target value for all cameras.

[0047] Next, in step S22, the exposure amount at the synthesis point is determined. This exposure amount at the synthesis point is also a common value for all cameras.

[0048] Steps S11 to S13 and S18 to S19 are the same as those in Example 1. As a result, the sub-pixel synthesis gain value and the sub-pixel switching point are registered in the register 40L of the left camera 1L.

[0049] In step S23, a correction gain for correcting the sensitivity difference with respect to the target composite straight line determined in step S21 is calculated for the left camera 1L. This operation may be performed after the sub-pixel sensitivity is acquired.

[0050] In step S1d, the correction gain calculated in step S23 is registered in the register 40L of the left camera 1L.

[0051] Steps S14 to S16 and S1a to S1b are the same as those in Example 1. As a result, the sub-pixel synthesis gain value and the sub-pixel switching point are registered in the register 40R of the right camera 1R.

[0052] In step S24, a correction gain for correcting the sensitivity difference with respect to the target composite straight line determined in step S21 is calculated for the right camera 1R. This operation may be performed after the sub-pixel sensitivities are obtained.

[0053] In step S1d, the correction gain calculated in step S24 is registered in the register 40R of the right camera 1R.

[0054] According to the imaging device of this embodiment described above, even if the sensitivity characteristics of the imaging devices 11 of the cameras vary from the lowest sensitivity to the highest sensitivity, it is possible to match the sensitivity characteristics of all the cameras without causing any misalignment of the synthesis points between the cameras, thereby achieving good parallax performance. [Example]

[0055] An imaging device 100 according to a third embodiment of the present invention will be described below with reference to Fig. 9. Note that a duplicated description of points common to the above-mentioned embodiments will be omitted.

[0056] The solid line in FIG. 9(a1) shows the sensitivity line after the sensitivity correction of Example 1 (see FIG. 4(c)), and the solid line in FIG. 9(b1) shows the sensitivity line after the sensitivity correction of the conventional example (see FIG. 5(c)).

[0057] As shown in both figures, the HDR image signal after subpixel synthesis and left / right sensitivity correction has a much larger maximum number of gradations than the original subpixel signal, and if left as is, it will be impossible to perform image calculations at the desired processing speed or display on the desired image display device.For this reason, the output after synthesis and sensitivity correction is nonlinearly compressed in signal compression unit 15 to an appropriate maximum number of gradations, as shown in Figures 9(a2) and (b2).

[0058] As mentioned above, when combining subpixels, a large gain is applied to the low-sensitivity subpixel 11b so that the low-sensitivity and high-sensitivity subpixels are on the same sensitivity line. As a result, in areas where the low-sensitivity subpixel 11b is used, a gain is also applied to noise, increasing it. As a result, as shown in Figures 9(a3) and (b3), the noise characteristics increase sharply at the combining point.

[0059] For this reason, in the conventional example in which the synthesis points are different between the left and right cameras, the noise characteristics also vary between the synthesis points of the left and right cameras, as shown in the lower graph of Fig. 9(b). On the other hand, in this embodiment, as shown in the lower graph of Fig. 9(a), the synthesis points of the left and right cameras are matched after synthesis and sensitivity correction, so there is no section between the synthesis points of the left and right cameras, and the noise characteristics of the left and right cameras match regardless of the exposure amount.

[0060] Incidentally, nonlinear compression that reduces the maximum number of gradations has the effect of compressing noise, and the higher the compression rate of the compression curve, the greater the noise suppression effect. Therefore, determining the compression curve determines the noise compression effect. In this embodiment, which allows the combination point to be constant, the second compression rate is set higher in areas where the exposure level is higher than the combination point and where noise is increased, compared to the first compression rate in areas where the exposure level is lower than the combination point and where noise is not increased. This allows for efficient noise suppression. [Explanation of symbols]

[0061] 100 Imaging device 1 camera 10. Imaging unit 11 Imaging devices 11a High-sensitivity subpixel 11b Low sensitivity subpixel 12 Gain Amplifier 13 HDR synthesis section 14 Gain Amplifier 15 Signal Compression Section 20 lenses 30 Geometric correction section 40 registers 50 Parallax Image Generation Unit 51 Stereo matching section Sa High-sensitivity sub-pixel signal Sb Low sensitivity sub-pixel signal

Claims

1. An imaging device having a plurality of cameras, Each camera is an imaging device in which unit pixels are periodically arranged, each unit pixel being a high-sensitivity sub-pixel that outputs a first output value for a certain amount of exposure and a low-sensitivity sub-pixel that outputs a second output value that is lower than the first output value for the certain amount of exposure; a synthesis unit that selects and outputs the output of the high-sensitivity sub-pixel when the exposure amount is smaller than a predetermined exposure value, and selects and outputs the output of the low-sensitivity sub-pixel when the exposure amount is larger than the predetermined exposure value; an amplifier that amplifies and outputs the output of the combiner; It has the predetermined exposure value is the same among the plurality of cameras, an amplifier for amplifying the image signals of the high-sensitivity subpixel and the low-sensitivity subpixel so that the image signals of the high-sensitivity subpixel and the low-sensitivity subpixel are the same across the multiple cameras at the same predetermined exposure value;

2. 2. The imaging device according to claim 1, wherein a gain is set in said amplifier section to match the output characteristics of the synthesis section of one camera with the output characteristics of the synthesis section of another camera.

3. 2. The imaging device according to claim 1, wherein a gain is set in said amplifier section so that the output characteristics of the combining section of each camera become predetermined target characteristics.

4. Each camera further includes a signal compression unit that compresses the signal output from the amplifier unit; 2. The imaging device of claim 1, wherein the signal compression unit compresses the signal at a first compression rate when the exposure amount is smaller than the predetermined exposure value, and compresses the signal at a second compression rate higher than the first compression rate when the exposure amount is larger than the predetermined exposure value.

5. The imaging device according to claim 1 , further comprising a parallax image generating unit that generates a parallax image using outputs from the plurality of cameras.

6. An imaging method using a plurality of cameras in which unit pixels are made up of a plurality of types of sub-pixels with different output characteristics relative to exposure amounts, the plurality of types of sub-pixels include high-sensitivity sub-pixels that output a first output value for a certain amount of exposure, and low-sensitivity sub-pixels that output a second output value that is lower than the first output value for the same amount of exposure, a step of switching an output from the high-sensitivity sub-pixel and an output from the low-sensitivity sub-pixel around a predetermined exposure value with respect to the exposure amount, and setting the output as an output of the unit pixel; and combining the image signals from the high-sensitivity sub-pixels and the low-sensitivity sub-pixels so that the image signals from the high-sensitivity sub-pixels and the low-sensitivity sub-pixels are the same at a predetermined exposure value between the plurality of cameras, an imaging method, wherein the predetermined exposure value is the same among the plurality of cameras;

Images