JP7599688B2 - Imaging method and imaging device - Google Patents

Description

The present invention relates to an imaging method and imaging device for reconstructing a two-dimensional or three-dimensional image from imaging data acquired by an image sensor.

Conventionally, single pixel imaging (SPI) and high-speed camera technology using CCD/CMOS are known as technologies for reconstructing two-dimensional or three-dimensional images from imaging data acquired by an image sensor.

SPI is a technology that reconstructs an image of an object by repeatedly capturing images of the object by focusing light onto an image sensor that has a special projection pattern superimposed on it, and then performing statistical processing, linear transformation, optimization calculations, etc. based on the multiple captured image data. In addition, because SPI uses a photodiode, which is a unit element of the image sensor, it has a higher degree of freedom in the wavelength band that can be captured compared to high-speed camera technology using CCD/CMOS, and in principle, high-speed imaging based on the high-speed response of the photodiode is possible.

The imaging device of Patent Document 1 reconstructs a three-dimensional image using SPI, and includes a spatial light modulator provided between a light source that irradiates illumination light and an object to be imaged, a lens arranged so that the illumination light patterned by the spatial light modulator is imaged on the object to be imaged, and an imaging element that acquires imaging data carried by the reflected or transmitted light of the illumination light from the object to be imaged. The imaging device changes the multi-value two-dimensional projection pattern displayed on the spatial light modulator over time, captures the relationship between the projection pattern displayed on the spatial light modulator and the imaging data acquired for each unit element of the imaging element when the projection pattern is displayed, reconstructs a two-dimensional element image for each unit element based on the imaging data captured for each projection pattern, and reconstructs a three-dimensional image of the object to be imaged based on the reconstructed two-dimensional element image for each unit element, thereby providing excellent resolution and resolution in the depth direction. The imaging element used in Patent Document 1 has unit elements arranged two-dimensionally.

JP 2020-136837 A (pages 5 to 6, Figure 1)

However, in order to obtain a high-resolution three-dimensional image using the imaging device of Patent Document 1, it is necessary to reconstruct a high-resolution two-dimensional elemental image by capturing imaging data based on many projection patterns for each unit element, which results in a problem of long imaging times.

The present invention was made in response to these problems, and aims to provide an imaging method and imaging device that can obtain reconstructed images quickly and with high resolution.

In order to solve the above problems, the imaging method of the present invention comprises:
forming an image of the light, in which the image of the object to be imaged and the plurality of projection patterns are superimposed, on a linear image sensor having at least two unit elements arranged one-dimensionally;
and performing an optimization calculation for the light collection direction of the linear image sensor based on a plurality of one-dimensional imaging data acquired by the linear image sensor.
According to this feature, the one-dimensional imaging data acquired by the linear image sensor is continuous data without duplication or loss in the direction perpendicular to the light collection direction, and the data is not compressed, so that an optimization calculation is performed in the light collection direction based on the multiple one-dimensional imaging data, and the result is combined with the data in the direction perpendicular to the light collection direction to obtain a reconstructed image at high speed and with high resolution with a small number of projection patterns. In addition, the imaging method of the present invention does not require mechanical scanning by forming an image of the imaging target and light in which multiple projection patterns are superimposed on the linear image sensor, and therefore can be applied to the imaging of various imaging targets.

The method is characterized by comprising a step of performing an optimization calculation in a light collecting direction of the linear image sensor, and then performing an optimization calculation in a direction perpendicular to the light collecting direction.
According to this feature, first, an optimization calculation is performed in the light collecting direction of the linear image sensor, which has many overlaps and defects, and then a two-stage optimization calculation is performed in the direction perpendicular to the light collecting direction, thereby making it possible to obtain a reconstructed image with higher resolution.

The optimization calculation of the focusing direction of the linear image sensor is characterized by performing a calculation based on Equation 1 on the basis of matrix A representing the irradiation of the multiple projection patterns, matrix B for determining the total variation, vector X^ representing one line in the focusing direction of the image to be reconstructed, and vector y representing the shooting data, thereby minimizing the squared error of one line in the focusing direction between the shooting data and the reconstructed image, and the total variation in one line in the reconstructed image.
According to this feature, one line is reconstructed by performing an optimization calculation for each line in the focusing direction, and a similar optimization calculation is repeated a predetermined number of times according to the number of pixels of the object to be imaged, thereby obtaining a reconstructed image at high speed and with high resolution.

The optimization calculation of the light collecting direction of the linear image sensor and the optimization calculation of the direction perpendicular to the light collecting direction are characterized in that the same mathematical formula is used.
According to this feature, a reconstructed image can be obtained quickly and with high resolution.

The matrix A is characterized in that it includes matrix conditions that represent a physical photographing process in which a plurality of the projection patterns are irradiated and one-dimensionally collected by an optical system.
According to this feature, a reconstructed image can be obtained quickly and with high resolution.

The imaging device of the present invention comprises:
a spatial light modulator that outputs a plurality of projection patterns;
a linear image sensor having at least two unit elements arranged one-dimensionally;
an imaging means for forming an image of an object to be imaged and light in which the plurality of projection patterns are superimposed on the linear image sensor;
and an analysis means for performing an optimization calculation for the light collecting direction of the linear image sensor based on a plurality of one-dimensional imaging data acquired by the linear image sensor.
According to this feature, the one-dimensional imaging data acquired by the linear image sensor is continuous data without duplication or loss in the direction perpendicular to the light collection direction, and the data is not compressed, so that an optimization calculation is performed in the light collection direction based on multiple one-dimensional imaging data, and the result is combined with data in the direction perpendicular to the light collection direction to obtain a reconstructed image at high speed and with high resolution using a small number of projection patterns. In addition, the imaging device of the present invention does not require mechanical scanning by forming an image of the imaging target and light in which multiple projection patterns are superimposed on the linear image sensor, and therefore can be applied to the imaging of various imaging targets.

The analyzing means performs an optimization calculation in a light collecting direction of the linear image sensor, and then performs an optimization calculation in a direction perpendicular to the light collecting direction.
According to this feature, by performing two-stage optimization calculations in the light collecting direction of the linear image sensor and in a direction perpendicular to the light collecting direction, a reconstructed image with higher resolution can be obtained.

The spatial light modulator is characterized in that it is a digital micromirror device.
According to this feature, it becomes possible to output a detailed projection pattern at high speed, so that a reconstructed image can be obtained at higher speed and with higher resolution.

The imaging means is characterized in that it is a cylindrical lens.
According to this feature, it is possible to one-dimensionally form an image on the linear image sensor of the light in which the image of the object to be imaged and a plurality of projection patterns are superimposed, using a simple configuration.

FIG. 1 is a schematic diagram illustrating an imaging device according to an embodiment of the present invention. FIG. 1 is a diagram showing the procedure of an imaging method in an embodiment. 1A is an imaging target, and FIG. 1B is a reconstructed image of the imaging target reconstructed by an imaging method using the imaging apparatus according to the embodiment.

The inventors have discovered that, similar to single-pixel imaging (SPI), light with a projection pattern superimposed on the image of the subject is focused on a linear image sensor with unit elements arranged one-dimensionally to capture the image, and that by repeating this capture while changing the projection pattern and performing optimization calculations on the light collection direction of the linear image sensor based on multiple one-dimensional image data acquired, it is possible to reconstruct an image with improved resolution while significantly reducing the number of captures compared to SPI, and that it is possible to achieve a frame rate that is significantly higher than high-speed camera technology using CCD/CMOS, and to reconstruct an image with improved resolution.

Image reconstruction involves performing optimization calculations for each line in the focusing direction to reconstruct one line at a time, and this process can be expressed as follows using the formula in Equation 2:

Here, X is a vector representing one line in the light collection direction of the object to be found, A is a matrix representing the irradiation of multiple projection patterns and the light collection by the optical system, and B is a vector representing the intensity data when multiple projection patterns are irradiated. For example, if the number of pixels of the imaging target is NxN pixels and the number of times the projection pattern is irradiated is M, the row vector of matrix A represents a certain projection pattern (N-dimensional vector), and if the number of times the projection pattern is irradiated is M, M different row vectors will be lined up in the column direction, and matrix A will be an MxN matrix. Furthermore, the size of vector X is Nx1, and the size of vector B is Mx1.

If the inverse matrix of matrix A could be found, vector X could be found easily, but since there is no inverse matrix for this matrix A, vector X^ is estimated by performing the following optimization using the formula 1 mentioned above, with the condition that the total variation of vector X^ is sparse.

The present invention reconstructs a two-dimensional or three-dimensional image at high speed and with high resolution by performing optimization calculations in the light collection direction of the linear image sensor based on multiple one-dimensional imaging data acquired by one-dimensionally focusing each projection pattern on the linear image sensor. Preferably, in addition to the optimization calculations in the light collection direction, optimization calculations are performed in a direction perpendicular to the light collection direction to reconstruct a two-dimensional or three-dimensional image at high speed and with higher resolution. The following formula 1 is used in this two-stage optimization calculation.

Here, A is a matrix representing the irradiation of multiple projection patterns, B is a matrix for finding the total variation, X^ is a vector representing one line in the focusing direction of the image to be reconstructed, and y is a vector representing the shooting data. The optimization calculation using the above formula minimizes the squared error between the shooting data and one line in the focusing direction of the image to be reconstructed, and the total variation in one line in the image to be reconstructed. Furthermore, matrix A includes the conditions of the matrix that represents the physical shooting process of irradiating multiple projection patterns and focusing light one-dimensionally by an optical system. Note that various parameters that reflect errors and the like according to the shooting conditions may be added to the above formula.

Preferably, the matrix B in the formula (1) is a matrix for calculating the total variation, and is a matrix for performing an operation to obtain the difference between adjacent pixels of the vector X^. This optimization operation reconstructs one line in the focusing direction of the imaging target, and by repeating this operation a total of N times in the direction perpendicular to the focusing direction, an image of N x N pixels is reconstructed. In other words, this optimization operation reconstructs an image compressed in the focusing direction, making it possible to reconstruct an image of N x N pixels. Preferably, a similar optimization is performed in the direction perpendicular to the focusing direction to reconstruct an image of N x N pixels with noise removed.

In addition, in the formula (1) used in the optimization calculation, matrix A includes matrix conditions that represent the physical imaging process of irradiating multiple projection patterns and focusing the light one-dimensionally by an optical system, thereby improving the analysis accuracy of the one-dimensional imaging data acquired by the linear image sensor, and enabling a reconstructed image to be obtained quickly and with high resolution.

The following describes the imaging method and imaging device according to the present invention based on the examples.

(Imaging device)
As shown in FIG. 1, the imaging device 1 in this embodiment is mainly composed of a spatial light modulator 2 capable of outputting a plurality of projection patterns onto a modulation surface 2a, a linear image sensor 3 in which unit elements are arranged one-dimensionally, a lens 4 that forms an image of an object to be imaged 10 on the modulation surface 2a of the spatial light modulator 2, a cylindrical lens 5 as an imaging means that one-dimensionally forms an image on the linear image sensor 3 of light in which the image of the object to be imaged 10 and the projection pattern output to the modulation surface 2a of the spatial light modulator 2 are superimposed, an A/D converter 6 that converts the one-dimensional imaging data acquired by the linear image sensor 3 from analog data to digital data, and a personal computer 7 as an analysis means that performs optimization calculations based on the plurality of digitally converted one-dimensional imaging data and the projection pattern data of the spatial light modulator 2.

The personal computer 7 includes a memory unit that stores one-dimensional imaging data corresponding to each projection pattern output to the modulation surface 2a of the spatial light modulator 2, an analysis unit that performs two-stage optimization calculations based on the multiple one-dimensional imaging data stored in the memory unit, and a display unit that inputs the image data reconstructed by the analysis unit and displays the reconstructed image.

The spatial light modulator 2 is a digital micromirror device (DMD) and has a modulation surface 2a on which multiple pixel regions capable of modulating the light intensity of output light relative to input light are arranged, and it is possible to perform two or more value light intensity modulation in each of the multiple pixel regions of the modulation surface 2a based on a set projection pattern. By using a DMD as the spatial light modulator 2, it is possible to output a detailed projection pattern at high speed.

In this embodiment, the spatial light modulator 2 is described as outputting a black and white binary two-dimensional random pattern (see FIG. 1) as the projection pattern, but it is not limited to this. It is also possible to output a two-dimensional random pattern consisting of three or more values, including values having gray levels in addition to black and white, to reduce the amount of one-dimensional imaging data required for image reconstruction and make analysis easier. In addition, the projection pattern is not limited to a random pattern, and it may be set to a pattern using a Hadamard transform or a discrete cosine transform (DCT) to obtain a reconstructed image more quickly.

The linear image sensor 3 is a Linea HS 32k (16384 pixels, 300 KHz/line) manufactured by Teledyne DALSA.

In theory, the linear image sensor in the imaging method of the present invention may have two unit elements arranged one-dimensionally.

Lens 4 is a biconvex lens that forms an image of the imaging target 10 on the modulation surface 2a of the spatial light modulator 2.

The imaging means is a plano-convex cylindrical lens 5, which collects the image of the imaging target 10 formed on the modulation surface 2a of the spatial light modulator 2 by the lens 4 and the light that is a superimposed projection pattern output to the modulation surface 2a of the spatial light modulator 2, and forms a one-dimensional image on the linear image sensor 3.

In this embodiment, the light collecting direction of the linear image sensor 3 is the direction in which light is collected by the cylindrical lens 5, i.e., the short side direction of the linear image sensor 3 (see Figure 1), and the direction perpendicular to the light collecting direction of the linear image sensor 3 is the direction in which light is not collected by the cylindrical lens 5, i.e., the long side direction of the linear image sensor 3 (see Figure 1), in other words, the direction in which the unit elements that make up the linear image sensor 3 are arranged one-dimensionally.

Imaging Methods
Next, an imaging method using the imaging device 1 in this embodiment will be described with reference to the flowchart in Fig. 2. First, an image of the imaging target 10 is formed on the modulation surface 2a of the spatial light modulator 2 by the lens 4 (step S01), a preset projection pattern is output to the modulation surface 2a of the spatial light modulator 2 (step S02), and light in which the image of the imaging target 10 and the projection pattern output to the modulation surface 2a of the spatial light modulator 2 are superimposed is one-dimensionally formed on the linear image sensor 3 by the cylindrical lens 5 (step S03).

Next, when one-dimensional imaging data is acquired by the linear image sensor 3 (step S04), the one-dimensional imaging data is converted from analog data to digital data by the A/D converter 6 (step S05), and the digitally converted one-dimensional imaging data is linked to the projection pattern data output to the modulation surface 2a of the spatial light modulator 2 during shooting and stored in the memory unit of the personal computer 7 (step S06).

Next, the personal computer 7 determines whether the number of one-dimensional imaging data stored in the storage unit has reached the same number as the preset number of times of shooting (step S07). Note that steps S02 to S07 are repeated while changing the projection pattern output to the modulation surface 2a of the spatial light modulator 2 until the number of one-dimensional imaging data stored in the storage unit reaches the preset number of times of shooting.

Next, when the number of one-dimensional imaging data stored in the memory unit reaches a preset number of shots, the analysis unit uses the above-mentioned formula to perform an optimization calculation (step S08) in the light-gathering direction of the linear image sensor 3 based on the multiple one-dimensional imaging data and the projection pattern data, and then uses the same formula to perform an optimization calculation (step S09) in the direction perpendicular to the light-gathering direction of the linear image sensor 3.

Next, in the analysis section, image reconstruction (step S10) is performed by combining the results of the optimization calculations in steps S08 and S09 , and the reconstructed image is displayed on the display section (step S11).

The spatial light modulator 2 is connected to the personal computer 7, and information such as the preset number of shots and the projection pattern output to the modulation surface 2a of the spatial light modulator 2 for each shot is shared.

In this way, the imaging method using the imaging device 1 in this embodiment involves concentrating the light, which is a superimposed image of the image of the object 10 and multiple projection patterns output to the modulation surface 2a of the spatial light modulator 2, using the cylindrical lens 5 to form a one-dimensional image on the linear image sensor 3.The multiple one-dimensional imaging data acquired by the linear image sensor 3 are compressed in the focusing direction of the linear image sensor 3, but are not compressed in the direction perpendicular to the focusing direction, resulting in continuous data without duplication or loss.Therefore, an optimization calculation is performed in the focusing direction of the linear image sensor 3 based on the multiple one-dimensional imaging data acquired for each projection pattern, and then an optimization calculation is performed in the direction perpendicular to the focusing direction of the linear image sensor 3.By combining these results, a reconstructed image can be obtained quickly and with high resolution using a small number of projection patterns.

The imaging method using the imaging device 1 in this embodiment significantly reduces the number of times images are taken compared to conventional SPI, while enabling image reconstruction with increased resolution, and has a significantly higher frame rate compared to high-speed camera technology using CCD/CMOS. The reason why this enables image reconstruction with increased resolution is that, compared to SPI, which focuses light in which a projection pattern is superimposed on an image of an object to be captured on a point sensor consisting of a single unit element, the linear image sensor 3 has a larger number of unit elements due to the unit elements being arranged one-dimensionally, so that the amount of information that can be acquired in one capture is greater. In more detail, since the unit elements of the linear image sensor 3 are arranged one-dimensionally, the spatial resolution is higher in the longitudinal direction of the linear image sensor 3, i.e., in the direction perpendicular to the light collecting direction of the linear image sensor 3. Therefore, the one-dimensional imaging data obtained by collecting light with a projection pattern superimposed on the image of the imaging target 10 using a cylindrical lens 5 and forming a one-dimensional image on the linear image sensor 3 is compressed in the collecting direction and can be considered as low-resolution point data similar to SPI, but is not compressed in the direction perpendicular to the collecting direction and can be considered as high-resolution continuous data without duplication or loss.

In this way, the one-dimensional imaging data acquired by the linear image sensor 3 is continuous high-resolution data with no overlaps or missing parts in the direction perpendicular to the light collection direction of the linear image sensor 3, so a large amount of information can be acquired in one shooting, and the correlation between the data is reasonably clear even without performing optimization calculations based on the multiple one-dimensional imaging data acquired for each projection pattern. Therefore, in order to reconstruct an image from the multiple one-dimensional imaging data acquired for each projection pattern, it is sufficient to determine the correlation between the data in the light collection direction by performing optimization calculations based on the data in the light collection direction of the linear image sensor 3, which can be considered as at least low-resolution point data, and as a result, it is possible to reduce the number of shootings required for image reconstruction and reconstruct an image with increased resolution.

In addition, in the imaging method using the imaging device 1 of this embodiment, two stages of optimization calculations are performed, one in the light collection direction of the linear image sensor 3 and the other in a direction perpendicular to the light collection direction, and the same formula is used for these optimization calculations, so that a reconstructed image can be obtained faster and with higher resolution.

Here, the results of a comparison between the imaging method using the imaging device 1 of this embodiment and high-speed camera technology using CMOS will be described. As the image sensor used in the high-speed camera technology, a 120MXS CMOS sensor (13272 x 9176 pixels) manufactured by Canon was used. The CMOS sensor has a frame rate of 9.4 fps, and requires 11.3 Gbps for data transfer. In contrast, the linear image sensor 3 constituting the imaging device 1 of this embodiment is a Linea HS 32k (16384 pixels, 300 KHz/line) manufactured by Teledyne DALSA. The inventors have verified that in the imaging method using the imaging device 1 of this embodiment, a sufficiently high-resolution reconstructed image (16384 x 16384 pixels) can be obtained with a number of projection patterns that is 50% of one line.

When estimating the imaging performance of the imaging method using the imaging device 1 of this embodiment under these conditions, the number of projection patterns is 8192, which means that an image of 16384 x 16384 pixels can be acquired at a frame rate of 37 fps (≒300 KHz/8192). In this way, it has been confirmed that the imaging method using the imaging device 1 of this embodiment enables image reconstruction with improved resolution at a significantly higher frame rate than high-speed CMOS camera technology.

In addition, data transfer using the imaging method using the imaging device 1 of this embodiment requires 1 Gbps. This indicates that the imaging method using the imaging device 1 of this embodiment implicitly has a data compression function.

The results of a comparison between the imaging method using the imaging device 1 of this embodiment and SPI will be described. Considering reconstructing an image of 16384 x 16384 pixels with SPI, when using an optimization method (compressed sensing), about 10% of the number of pixels of the reconstructed image, i.e., about 26.8 million projection patterns, are required, and when using deep learning, about 2% of the number of pixels of the reconstructed image, i.e., 5.4 million projection patterns are required. With the imaging method using the imaging device 1 of this embodiment, the number of projection patterns required to reconstruct an image of 16384 x 16384 pixels is 8192, so it has been confirmed that it is possible to reconstruct an image with improved resolution while significantly reducing the number of shots compared to SPI.

The linear image sensor 3 is assumed to be a single line sensor, but by using a multi-line sensor, the number of shots can be further reduced and image reconstruction can be performed. In this case, since the light receiving angle changes for each line of the linear image sensor, the matrix A in the optimization calculation formula is set for each line.

In addition, the imaging method using the imaging device 1 in this embodiment can be applied to imaging of various imaging targets because it does not require mechanical scanning, by one-dimensionally focusing the image of the imaging target 10 and light in which multiple projection patterns are superimposed on the linear image sensor 3 using the cylindrical lens 5.

In this embodiment, a two-dimensional image is reconstructed from one-dimensional imaging data using one linear image sensor 3. However, the present invention is not limited to this. For example, a three-dimensional image can be reconstructed by applying the principle of triangulation by arranging multiple linear image sensors and acquiring one-dimensional imaging data from different angles.

Although the embodiments of the present invention have been described above with reference to the drawings, the specific configuration is not limited to these embodiments, and the present invention also includes modifications and additions that do not deviate from the gist of the present invention.

For example, in the above embodiment, the same formula is used in two stages of optimization calculation, but this is not limiting, and for example, a different formula may be used in the optimization calculation in the direction perpendicular to the focusing direction.

In addition, in the above embodiment, image reconstruction is performed by analyzing one-dimensional imaging data using an optimization calculation formula, but image reconstruction may also be performed by deep learning.

In addition, the linear image sensor is not limited to a single unit element arranged in a continuous line, but may be a single point sensor arranged at a predetermined interval to be used in the SPI.

In addition, the spatial light modulator is not limited to a DMD, but can be anything that can perform modulation with high precision and high speed.

In the above embodiment, the image of the object to be imaged is formed on the modulation surface of the spatial light modulator and superimposed on the projection pattern. However, the present invention is not limited to this. The projection pattern output to the modulation surface of the spatial light modulator may be irradiated onto the object to be imaged and superimposed on the image of the object to be imaged.

The imaging means is not limited to a cylindrical lens, but may be anything that can one-dimensionally image the light in which the image of the object to be imaged and the projection pattern are superimposed on the linear image sensor.

In addition, the light that superimposes the image of the subject and the projection pattern only needs to be of a wavelength to which the linear image sensor is sensitive, and the image of the subject may be obtained by illumination light from a light source set for shooting, or may be obtained by natural light such as sunlight.

1 Imaging device 2 Spatial light modulator 2a Modulation surface 3 Linear image sensor 4 Lens 5 Cylindrical lens (imaging means)
6 A/D converter 7 Personal computer (analysis means)
10. Imaging target

Claims (9)

forming an image of the light, in which the image of the object to be imaged and the plurality of projection patterns are superimposed, on a linear image sensor having at least two unit elements arranged one-dimensionally;
and performing an optimization calculation for a light collection direction of the linear image sensor based on a plurality of one-dimensional imaging data acquired by the linear image sensor. The imaging method according to claim 1, further comprising a step of performing an optimization calculation in a direction perpendicular to the light collection direction of the linear image sensor after performing the optimization calculation in the light collection direction of the linear image sensor. The imaging method according to claim 1 or 2, characterized in that the optimization calculation of the focusing direction of the linear image sensor is performed based on a matrix A representing the irradiation of the multiple projection patterns, a matrix B for determining the total variation, a vector X^ representing one line in the focusing direction of the image to be reconstructed, and a vector y representing the shooting data, by performing a calculation based on Equation 1, thereby minimizing the squared error of one line in the focusing direction between the shooting data and the reconstructed image, and the total variation in one line in the reconstructed image.

The imaging method according to claim 3, characterized in that the same formula is used for the optimization calculation of the light collection direction of the linear image sensor and the optimization calculation of the direction perpendicular to the light collection direction. The imaging method according to claim 3 or 4, characterized in that the matrix A includes matrix conditions that represent the physical imaging process of irradiating the multiple projection patterns and one-dimensionally focusing the light by an optical system. a spatial light modulator that outputs a plurality of projection patterns;
a linear image sensor having at least two unit elements arranged one-dimensionally;
an imaging means for forming an image of an object to be imaged and light in which the plurality of projection patterns are superimposed on the linear image sensor;
and an analysis means for performing an optimization calculation for the light collecting direction of the linear image sensor based on a plurality of one-dimensional imaging data acquired by the linear image sensor. The imaging device according to claim 6, characterized in that the analysis means performs optimization calculations in the light collecting direction of the linear image sensor, and then performs optimization calculations in a direction perpendicular to the light collecting direction. The imaging device of claim 6, wherein the spatial light modulator is a digital micromirror device. The imaging device according to claim 6, characterized in that the imaging means is a cylindrical lens.

Images