A dual-optical-arm dynamic single-pixel imaging system and method based on radon spectrum

Abstract

The present application relates to a kind of dual-optical arm dynamic single-pixel imaging system and method based on Radon spectrum, the imaging system includes beam splitter, dove prism, plano-convex cylindrical lens, spatial light modulator, single-pixel detector and imaging lens.One optical path in dual-optical arm single-pixel imaging system is used for positioning, and the other optical path is used for single-pixel imaging.Positioning light path is completed by optical element dove prism, plano-convex cylindrical lens and linear array detector for fast positioning, and imaging light path is compensated according to the positioning result to complete the reconstruction of moving target for modulated speckle of light field.Adopting the imaging system and method can avoid the problem of increasing sampling rate caused by traditional positioning method, and also can avoid the problem that positioning speckle is vulnerable to noise, provides an effective way for moving single-pixel imaging.

Description

Technical Field

[0001] This invention relates to single-pixel imaging, and more particularly to a dual-arm dynamic single-pixel imaging system and method based on Radon spectrum. Background Technology

[0002] Single-pixel imaging is a novel imaging technique that obtains multi-dimensional information about a target object through light field modulation and multiple sampling reconstruction. During sampling, the relative motion between the object and the imaging system leads to degradation in the quality of the reconstructed image. To address this problem, researchers have proposed various solutions, including developing high refresh rate light sources, improving imaging strategies, developing advanced real-time algorithms, and utilizing various prior information, such as velocity priors, position priors, and sparse priors. Imaging methods utilizing prior information are commonly used in the field of moving single-pixel imaging. Shi Dongfeng's research group proposed a positioning method based on geometric moments, which can locate the position of a moving target using only three speckle images. However, this method is susceptible to noise and increases the number of samples, which is not conducive to the practical application of single-pixel imaging. Dynamic single-pixel imaging utilizing prior information has shown broad application prospects, but a low-sampling-rate and interference-resistant dynamic single-pixel imaging mechanism still needs further exploration. Summary of the Invention

[0003] To achieve the above objectives, the present invention adopts the following technical solution:

[0004] A dual-arm dynamic single-pixel imaging system based on Radon spectrum is disclosed for rapid localization and imaging of moving targets. The imaging system includes a beam splitter, a Dowell prism, a plano-convex cylindrical mirror, a linear array detector, a spatial light modulator (DMD), a single-pixel detector, and an imaging lens. The optical path of the system includes a localization optical path and a single-pixel imaging optical path. The localization optical path consists of the Dowell prism, the plano-convex cylindrical mirror, and the linear array detector; the single-pixel imaging optical path consists of the DMD, the imaging lens, and the single-pixel detector. The image of the moving target is split into two paths after passing through the beam splitter. One path passes through the localization optical path for localization, and the other path passes through the single-pixel imaging optical path for image reconstruction.

[0005] Meanwhile, this invention proposes a dual-arm dynamic single-pixel imaging method based on Radon spectrum, the method comprising the following steps:

[0006] Step 1: The image of the moving target is split into two paths after passing through the beam splitter. One path is used to complete the positioning optical path, and the other path is used to complete the reconstruction of the moving target image through the single-pixel imaging optical path.

[0007] Step 2: In the positioning optical path, the image is rotated by a certain angle after passing through the Dowell prism;

[0008] Step 3: The rotated image is compressed into a one-dimensional signal by passing through a plano-convex cylindrical mirror;

[0009] Step 4: The compressed one-dimensional signal is detected and collected by a linear array detector;

[0010] Step 5: Rotate the Dowell prism by a certain angle and repeat steps 2, 3, and 4 to obtain the Radon spectrum information of the moving target image at that angle. Rotate the Dowell prism multiple times to obtain Radon spectrum information at different angles. Use the multi-angle Radon spectrum information to determine the position of the moving target at the image plane of the spatial light modulator DMD.

[0011] Step 6: The single-pixel imaging optical path needs to modulate the light field of the moving target. This modulation is achieved using a spatial light modulator (DMD), with the modulated speckle pattern based on the Hadamard basis. Assume the DMD modulation frequency is... The target's movement speed is The pixel size at the moving target is The number of modulations of the moving target by the DMD within the time it takes to complete one positioning operation. The following relationship must be satisfied:

[0012]

[0013] Repeat steps 5 and 6, assuming the image size of the moving target is... pixels, total number of pixels ,in Then the computer needs to generate open The Hadamard modulated speckle pattern. The single-pixel imaging optical path will follow the order of the Hadamard speckle pattern according to 1. , , , The target image is cyclically modulated in sequence until the target stops moving or goes out of the field of view;

[0014] Step 7: Finally, based on the position information of the moving target at the image plane of the spatial light modulator DMD obtained from the positioning optical path, speckle is compensated, and the moving target image is reconstructed using the single-pixel detector value and the compensated speckle.

[0015] The dynamic single-pixel imaging method based on the Radon spectrum of this invention can achieve rapid and accurate positioning without increasing the number of samples, avoiding the problems of increased sampling number and susceptibility to noise caused by positioning speckle, and providing an effective approach for dynamic single-pixel imaging. Attached Figure Description

[0016] Figure 1 A schematic diagram of the optical path structure for dynamic single-pixel imaging with dual optical arms based on the Radon spectrum;

[0017] Figure 2This is a schematic diagram of the optical path structure of a Dove prism and a plano-convex cylindrical mirror. Figure 2 (a) is a schematic diagram of the optical path structure of the Dowell prism. Figure 2 (b) is a schematic diagram of the optical path structure for imaging with a plano-convex cylindrical mirror;

[0018] Figure 3 For multiple objectives , , , The direction of the spectrum;

[0019] Figure 4 The results are simulation results of dynamic single-pixel imaging with dual optical arms based on the Radon spectrum.

[0020] Figure 5 This is a flowchart of dynamic single-pixel imaging using dual optical arms based on the Radon spectrum. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0022] Figure 1 This is a schematic diagram of the optical path structure for dynamic single-pixel imaging using a dual-arm optical arm based on the Radon spectrum. Figure 1 As shown, the image is split into two paths after passing through the beam splitter: one for target localization and the other for single-pixel imaging. The localization optical path consists of a Dowell prism, a plano-convex cylindrical mirror, and a linear array detector. Figure 2 (a) is a schematic diagram of the optical path structure of the Dowell prism. Figure 2 Image (b) is a schematic diagram of the optical path structure for imaging with a plano-convex cylindrical mirror. The Dove prism has the function of transmitting and rotating signal images; therefore, different rotation angles can be obtained through the Dove prism. The image is a plano-convex cylindrical mirror, which can compress the rotated image into a one-dimensional signal, namely the angle. The Radon spectrum is displayed, and this one-dimensional signal is acquired by an n×1 dimensional linear array detector. Each rotation of the Dove prism results in one acquisition by the linear array detector. The rotation angle needs to be determined based on the number of moving targets; for single-target and dual-target objects, only 0° is required. and 90 Directional spectral mapping is sufficient for localization; however, for three or more targets, at least three angular spectral mappings are required for a single localization operation, with angles ranging from 0 to 90 degrees. The image is divided into equal parts. A single-pixel imaging optical route consists of a spatial light modulator (DMD, Digital Micro-mirror Device), a single-pixel detector, and an imaging lens. Single-pixel imaging requires multiple modulations of the target object, followed by a reconstruction algorithm to obtain the moving target image. The spatial light modulator (DMD) is responsible for loading modulated speckle and modulating the light field of the moving target image. The modulated total light intensity signal is then detected and collected by the single-pixel detector, which lacks spatial resolution, through the imaging lens.

[0023] The aforementioned modulated speckle patterns include, but are not limited to, Fourier basis, discrete cosine basis, Hadamard basis, and Gaussian random speckle patterns. Here, we will use Hadamard speckle patterns composed of +1 and -1 elements as an example. Hadamard speckle patterns can be easily implemented in optical systems using a DMD (Digital Modulator-Diffraction Model). The generation formula is as follows: First, the formula for generating a second-order Hadamard speckle pattern is expressed as follows:

[0024]

[0025] Based on the second-order Hadamard speckle pattern, we can deduce The Hadamard speckle pattern of order 1 is generated by the following formula:

[0026]

[0027] In the formula, This represents the Kronecker product. When generating modulated speckle, it is necessary to... The row or column vectors in the image are changed to n×n dimensions, where n×n is the same size as the moving target image.

[0028] In the Radon spectrum-based dual-arm dynamic single-pixel imaging system, the positioning optical path completes one target localization, and the DMD modulates the moving target image. Second-rate. The value setting can be determined based on the moving target's speed, and the moving target and imaging system are relatively stationary during the acquisition process of single-pixel detectors and linear array detectors. Assume the DMD modulation frequency is... The object's speed is The pixel size at the moving target is Then the number of modulations of the moving target by the DMD The following relationship must be satisfied:

[0029]

[0030] In single-pixel imaging, the DMD encodes and modulates the moving target multiple times in a binary manner, and uses a single-pixel detector to collect the total light intensity of the encoded moving target. When the moving target moves relative to the imaging system, it causes image blurring. Therefore, we need to compensate for speckle based on position prior to eliminate the image blurring caused by motion. The single-pixel imaging reconstruction formula for moving targets based on position prior is as follows:

[0031]

[0032] In the formula, Represents the reconstructed image of the moving target. Indicates the number of pixels of the moving target. This represents the spatial coordinates in the Cartesian coordinate system of DMD modulated speckle. This represents the value of a single pixel detector. and These represent the lateral and longitudinal displacements of the moving target on the DMD, respectively.

[0033] The number of moving targets includes single targets, two targets, and three or more targets. When the number of moving targets is single or two, localization only requires... and The Ladon spectrum; when there are three or more moving targets, at least three angles of Ladon spectrum are required to complete one localization operation. Here, the angles are... The equal division. For example... Figure 3 As shown, Figure 3 Arrows (a)-(d) indicate the integral projection direction of the moving target, and arrows (e)-(h) indicate the directions of the integral projection direction of the moving target, respectively. The integral projection curves of the three moving targets in the direction of the target are called the Radon spectrum, which is the value of the one-dimensional linear array detector in the positioning optical path. From Figure 3 It can be seen from this that, In the direction, the circle and the pentagon overlap; in The direction, the circle, and the hexagon overlap. And... and The directions of all three objects can be clearly identified. Based on this, it can be inferred that by increasing the positioning angle of the Radon spectrum, precise positioning of multiple targets can be achieved. Compared to traditional discrete cosine speckle localization, this method reduces the sampling rate by a factor of p, where p represents the number of discrete cosine speckles required to locate a single direction.

[0034] Figure 4 The results are from a simulation of dynamic single-pixel imaging of multiple targets. Figure 4 (a) represents the original image, and the dashed arrow indicates the direction of the object's movement. Figure 4 In the middle (b), the moving target reconstruction result without speckle compensation is shown, with a peak signal-to-noise ratio of 11.17 dB. Figure 4In Figure (c), the reconstructed moving target with speckle compensation is shown, with a peak signal-to-noise ratio (PSNR) of 25.51 dB. The experimental results demonstrate that speckle compensation effectively eliminates the influence of target motion on the reconstruction results. The formula for calculating the peak signal-to-noise ratio is as follows:

[0035]

[0036] In the formula, , represents the maximum pixel value of the image. The number of bits per sampling point; if it is 8 bits, then... . and Indicates image size, Represents the grayscale value of a real image pixel. This represents the grayscale value of a pixel in the reconstructed image.

[0037] Figure 5 This is a flowchart of a dual-arm dynamic single-pixel imaging process based on the pull spectrum. It specifically includes the following steps:

[0038] Step 1: After passing through the beam splitter, the image is divided into two paths, one for positioning and the other for single-pixel imaging.

[0039] Step 2: In the positioning optical path, the object image obtains a certain rotation angle after passing through the Dowell prism;

[0040] Step 3: The rotated image is compressed into a one-dimensional signal by passing through a plano-convex cylindrical mirror;

[0041] Step 4: The compressed one-dimensional signal is detected and collected by a linear array detector;

[0042] Step 5: Repeat steps 2, 3, and 4 to obtain multi-angle Radon spectrum information, determine the location of the moving target, and save it;

[0043] Step 6: In the imaging optical path, when the moving target is stationary relative to the imaging system, the target positioning is completed. At the same time, the spatial light modulator (DMD) modulates the light field and the total light intensity is collected by a single-pixel detector.

[0044] Repeat steps 5 and 6 until the target stops moving or goes out of the field of view;

[0045] Step 7: Use the obtained prior position information to compensate for the modulated speckle and reconstruct the moving target.

[0046] The following specific embodiments of the present invention provide a dual-arm dynamic single-pixel imaging method based on Radon spectrum, which specifically includes the following steps:

[0047] Step 1: The image of the moving target is split into two paths after passing through the beam splitter. One path is used to complete the positioning optical path, and the other path is used to complete the reconstruction of the moving target image through the single-pixel imaging optical path.

[0048] Step 2: The image in the positioning optical path is rotated by a certain angle through the Dowell prism;

[0049] Step 3: The rotated image is compressed into a one-dimensional signal by passing through a plano-convex cylindrical mirror;

[0050] Step 4: The compressed one-dimensional signal is detected and collected by a linear array detector;

[0051] Step 5: Rotate the Dowell prism by a certain angle and repeat steps 2, 3, and 4 to obtain the Radon spectrum information of the moving target image at that angle. Rotate the Dowell prism multiple times to obtain Radon spectrum information at different angles. Use the multi-angle Radon spectrum information to determine the position of the moving target at the image plane of the spatial light modulator DMD.

[0052] Step 6: The single-pixel imaging optical path modulates the light field of the moving target. This modulation is achieved using a spatial light modulator (DMD), with the modulated speckle pattern based on the Hadamard basis. Assume the DMD's flip frequency is... The moving target's speed is The pixel size at the moving target is The number of times the DMD flips within the time it takes to complete one positioning cycle is then determined. The following relationship must be satisfied:

[0053]

[0054] Repeat steps 5 and 6, assuming the image size of the moving target is... pixels, total number of pixels ,in Then the computer needs to generate open The Hadamard modulated speckle pattern, the single-pixel imaging optical path will follow the order of the Hadamard speckle pattern according to 1 , , , The target image is cyclically modulated in sequence until the target stops moving or goes out of the field of view;

[0055] Step 7: Finally, based on the position information of the moving target at the image plane of the spatial light modulator DMD obtained from the positioning optical path, speckle is compensated, and the moving target image is reconstructed using the single-pixel detector value and the compensated speckle.

[0056] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dual-arm dynamic single-pixel imaging method based on the Radon spectrum, which employs a dual-arm dynamic single-pixel imaging system based on the Radon spectrum for locating and imaging moving targets. The imaging system includes a beam splitter, a Dowell prism, a plano-convex cylindrical mirror, a linear array detector, a spatial light modulator (DMD), a single-pixel detector, and an imaging lens. The optical path of the system includes a positioning optical path and a single-pixel imaging optical path. The positioning optical path consists of the Dowell prism, the plano-convex cylindrical mirror, and the linear array detector. The single-pixel imaging optical path consists of the spatial light modulator (DMD), the imaging lens, and the single-pixel detector. The image of the moving target is split into two paths after passing through the beam splitter; one path passes through the positioning optical path for positioning, and the other path passes through the single-pixel imaging optical path for image reconstruction. The method is characterized in that... The method includes the following steps: Step 1: The image of the moving target is split into two paths after passing through the beam splitter. One path is used to complete the positioning optical path, and the other path is used to complete the reconstruction of the moving target image through the single-pixel imaging optical path. Step 2: In the positioning optical path, the image is rotated by a certain angle after passing through the Dowell prism; Step 3: The rotated image is compressed into a one-dimensional signal by passing through a plano-convex cylindrical mirror; Step 4: The compressed one-dimensional signal is detected and collected by a linear array detector; Step 5: Rotate the Dowell prism by a certain angle and repeat steps 2, 3, and 4 to obtain the Radon spectrum information of the moving target image at that angle. Rotate the Dowell prism multiple times to obtain the Radon spectrum information of the moving target at different angles. Use the multi-angle Radon spectrum to determine the position of the moving target on the DMD. Step 6: The single-pixel imaging optical path needs to modulate the light field of the moving target. This modulation is achieved using a spatial light modulator (DMD), with the modulated speckle pattern based on the Hadamard basis. Assume the DMD modulation frequency is... The target's movement speed is The pixel size at the moving target is The number of modulations of the moving target by the DMD within the time it takes to complete one positioning operation. The following relationship must be satisfied: Repeat steps 5 and 6, assuming the image size of the moving target is... pixels, total number of pixels ,in Then the computer needs to generate open The Hadamard modulated speckle pattern; the single-pixel imaging optical path will follow the order of the Hadamard speckle pattern according to 1 , , , The target image is cyclically modulated in sequence until the target stops moving or goes out of the field of view; Step 7: Finally, based on the position information of the moving target at the DMD image plane obtained from the positioning optical path, speckle is compensated, and the moving target image is reconstructed using the single-pixel detector value and the compensated speckle.

2. The method for dynamic single-pixel imaging based on Radon spectrum with dual optical arms according to claim 1, characterized in that, The moving targets include single targets, dual targets, and three or more targets. When the moving target is a single or dual target, the Dowell prism needs to rotate twice to determine the position of the moving target at the image plane of the spatial light modulator (DMD). When the moving target is three or more targets, the Dowell prism needs to rotate at least three times to determine the position of the moving target at the image plane of the spatial light modulator (DMD).

3. The method for dynamic single-pixel imaging based on Radon spectrum using dual optical arms according to claim 1, characterized in that, When the number of moving targets is single or dual, it is necessary to acquire and The directional spectral density; when there are three or more moving targets, obtaining the position of a single moving target requires at least three angles of spectral density information, where the angles are... Divide into equal parts.

4. The method for dynamic single-pixel imaging based on Radon spectrum using dual optical arms according to claim 1, characterized in that, In step 7, the compensated speckle reconstruction algorithm is as follows: In the formula, Represents the reconstructed image of the moving target. This represents the total number of pixels in the image of the moving target. This represents the spatial coordinates of the modulated speckle in the Cartesian coordinate system. This represents the value of a single pixel detector. and These represent the lateral and longitudinal displacements of the moving target at the image plane of the spatial light modulator (DMD), respectively.

Images