Hybrid coding parallel single-pixel imaging method

Abstract

The application discloses a kind of hybrid codec parallel single-pixel imaging method, it is related to the field of computational imaging technology.The specific includes: configure multiple photoelectric detectors to receive light signal from target scene, and the target scene is regarded as by multiple sub-regions;Control spatial light modulator to project a series of encoding patterns to target scene;Utilize multiple photoelectric detectors and collect light intensity signal in parallel, wherein each photoelectric detector collected light intensity signal contains the mixed light intensity component from multiple sub-regions;Obtain a unmixing matrix, the unmixing matrix represents the spatial response weight distribution of each photoelectric detector to each sub-region light signal;Based on unmixing matrix, the light intensity signal collected is unmixing operation to separate out respectively corresponding to each sub-region independent response signal;According to independent response signal and encoding pattern, reconstruct the image of each sub-region.The purpose is to solve the problem of traditional single-pixel imaging slow and array detector high cost, there is signal crosstalk.

Description

Technical Field

[0001] This invention relates to the field of computational imaging technology, and in particular to a hybrid encoding and decoding parallel single-pixel imaging method. Background Technology

[0002] Single-pixel imaging is an innovative computational imaging technique that uses a single-point detector to encode the spatial information of an object into a one-dimensional light signal through spatial light modulation, and then uses computational methods to reconstruct the image. Compared with traditional imaging techniques, single-pixel imaging does not require a high-resolution detector and can adapt to various complex scenarios such as imaging in non-visible light bands, low-light environments, and murky media.

[0003] Traditional single-pixel imaging reconstructs images by measuring the correlation between a scene and a series of illumination patterns. The image quality depends on the number of illumination patterns. For example, in Hadamard and Fourier single-pixel imaging, reconstructing an image at a certain resolution requires a large number of coded patterns, and the higher the resolution, the more coded patterns are needed, leading to longer projection times and limiting its practicality. To improve imaging speed, existing technologies have proposed single-pixel imaging schemes based on array detectors. This involves dividing the imaging area into different regions and using a customized array detector aligned with each region for parallel acquisition. However, this array detector-based imaging technique has significant drawbacks. First, physical alignment of the imaging area with the array detector is very difficult; even slight deviations can affect image quality. Second, custom-designed array detectors are expensive. Third, closely packed array detectors are prone to signal interference (crosstalk) between different regions, and high-density integration can easily lead to overheating and system crashes. Essentially, existing technologies have not yet adequately solved the problem of overcoming signal crosstalk and achieving fast, high-resolution image reconstruction in a low-cost and simple manner while reducing the number of modulation patterns.

[0004] Therefore, how to achieve fast and high-resolution single-pixel imaging with lower hardware costs and fewer modulation patterns, and effectively solve the problem of crosstalk in multiple regions, has become an urgent technical challenge. Summary of the Invention

[0005] The main objective of this invention is to provide a hybrid encoding and decoding parallel single-pixel imaging method, which aims to achieve fast and high-resolution single-pixel imaging with lower hardware costs and fewer modulation patterns, and effectively solve the problem of signal crosstalk in multiple regions.

[0006] To achieve the above objectives, this invention proposes a hybrid encoding / decoding parallel single-pixel imaging method, comprising: Multiple photodetectors are configured to receive light signals from the target scene, and the target scene is considered to be composed of multiple sub-regions; A spatial light modulator is controlled to project a series of hybrid coded patterns onto the target scene, wherein the coded patterns present corresponding modulation codes on the multiple sub-regions respectively; The light intensity signal is acquired in parallel using the multiple photodetectors, wherein the light intensity signal acquired by each photodetector contains a mixed light intensity component from multiple sub-regions; Obtain a demixing matrix, which characterizes the spatial response weight distribution of each photodetector to the optical signal in each sub-region; Based on the demixing matrix, the light intensity signals collected by the multiple photodetectors are demixed to separate the independent response signals corresponding to each sub-region. Based on the independent response signal and the modulation coding of the coding pattern in the corresponding sub-region, the image of each sub-region is reconstructed.

[0007] Preferably, the unmixing matrix is ​​constructed using the following system of linear equations:

[0008] in, Indicates the first The first photodetector collected the data of the first... Secondary light intensity signal Indicates the first Independent response signals for each sub-region Indicates the first The photodetector pairs with the first The spatial response weighting coefficients of each sub-region, the unmixing matrix being derived from... The coefficient matrix formed.

[0009] Preferably, the step of obtaining a demixing matrix includes: The spatial light modulator is controlled to project a specific coded pattern, which is used to illuminate individual sub-regions individually or in combination; The corresponding light intensity response values ​​are collected using the multiple photodetectors; Calculate the weight coefficients of each element in the coefficient matrix based on the collected light intensity response values. The value of .

[0010] Preferably, the step of performing demixing operations on the light intensity signals collected by the plurality of photodetectors based on the demixing matrix includes: Calculate the inverse of the unmixing matrix; The vector formed by the light intensity signals collected by the multiple photodetectors is multiplied by the inverse matrix on the left to calculate the independent response signal. .

[0011] Preferably, the encoding pattern is composed of a combination of different types of modulation codes; The first sub-region presents a Hadamaj pattern, and the second sub-region presents a Fourier pattern. The Hadamaj pattern and the Fourier pattern are combined to form the coded pattern.

[0012] Preferably, the encoding pattern is composed of a combination of modulation codes of the same type; The first sub-region presents a first Hadhamaki pattern, the second sub-region presents a second Hadhamaki pattern, and the first Hadhamaki pattern and the second Hadhamaki pattern are combined to form the coded pattern.

[0013] Preferably, the modulation coding is Fourier coding, and the generation expression of the Fourier coding is:

[0014] in, The average light intensity For contrast, and For spatial frequency, For phase; Furthermore, the Fourier encoding employs a four-step phase-shift method, with phase... Take respectively , , and .

[0015] Preferably, the number of the plurality of photodetectors is two, namely a first photodetector and a second photodetector; the number of the plurality of sub-regions is two, namely a first sub-region and a second sub-region; Wherein, the spatial response weighting coefficient of the first photodetector to the first sub-region is greater than its spatial response weighting coefficient to the second sub-region, and the spatial response weighting coefficient of the second photodetector to the second sub-region is greater than its spatial response weighting coefficient to the first sub-region. The unmixing matrix is: A matrix is ​​used to separate the independent response signals of the first sub-region and the second sub-region.

[0016] Preferably, the spatial light modulator is a digital micromirror device (DMD), a projector, or a liquid crystal display (LCD).

[0017] Preferred options also include: The reconstructed images of each sub-region are stitched together to form a complete high-resolution image of the target scene.

[0018] The above technical solution has the following advantages: This invention configures multiple photodetectors and treats the target scene as composed of multiple sub-regions. It utilizes a parallel imaging demixing model to obtain a demixing matrix characterizing the spatial response weight distribution of each photodetector to the light signals in each sub-region. Then, it performs demixing operations on the parallel-acquired mixed light intensity signals to separate the independent response signals of each sub-region and reconstruct the image. This method not only significantly reduces the number of required projection coding patterns, greatly improves imaging efficiency, and achieves higher-resolution image reconstruction, but also replaces physical isolation with algorithmic demixing, breaking the dependence on expensive custom array detectors, flexibly arranging the detectors spatially, and effectively solving problems such as signal crosstalk, hardware alignment difficulties, and equipment heat dissipation that are difficult to avoid in traditional array imaging. This provides strong technical support for the cost reduction and practical application of single-pixel imaging systems. Attached Figure Description

[0019] The present invention will now be described in detail with reference to specific embodiments and accompanying drawings, wherein: Figure 1 A schematic diagram illustrating the system structure and principle of the hybrid encoding and decoding parallel single-pixel imaging method provided in this embodiment of the invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the following specific embodiments are only used to explain the invention and do not constitute a limitation thereof.

[0021] Example 1 This embodiment provides a hybrid encoding and decoding parallel single-pixel imaging method. Based on the biomimetic principle of compound eye structure, this method proposes a parallel imaging demixing model algorithm, aiming to solve the technical problems in traditional single-pixel imaging technology, such as the sharp increase in the number of projected patterns, long imaging time, high hardware cost of array detectors, and severe signal crosstalk, as resolution increases. The core of this embodiment lies in using multiple non-array photodetectors in conjunction with a computational imaging algorithm to achieve efficient and high-resolution image reconstruction without the need for strict physical alignment of the imaging area.

[0022] The method described in this embodiment mainly relies on a single-pixel imaging system. This system primarily includes an illumination module, a spatial light modulation module, a detection module, and a computational control module. The illumination module provides a light source; the light is modulated by the spatial light modulation module and projected onto the target scene, or the spatial light modulation module modulates the light reflected from the target scene. In this embodiment, the spatial light modulation module specifically employs a digital micromirror device (DMD). Of course, in other feasible implementations, a projector or liquid crystal display (LCD) can also be used as a spatial light modulator. The detection module includes multiple independent photodetectors, specifically single-point detectors, such as photodiodes or photomultiplier tubes. These photodetectors are spatially independent, offering flexibility in their placement and eliminating the need for precise physical alignment and customization required by traditional array detectors. The computational control module controls the flipping of the spatial light modulator, acquires signals from the photodetectors, and executes subsequent demixing and reconstruction algorithms.

[0023] The method in this embodiment specifically includes the following steps: First, multiple photodetectors are configured to receive light signals from the target scene, which is logically considered to be composed of multiple sub-regions. Inspired by the compound eye vision mechanism, each individual photodetector is treated as a small eye, or microscopic unit, within the compound eye. Due to the spatial differences in the photodetectors, even when facing the same large target scene, the response sensitivity (i.e., the weight of the received light intensity) of each detector to different regions of the scene varies. We divide the entire target scene to be imaged into multiple sub-regions corresponding to the number of detectors. For example, if 10 photodetectors are used, the target scene is divided into 10 sub-regions. Each sub-region corresponds to a specific photodetector as the master detector, while the other detectors act as auxiliary detectors relative to that sub-region. This division is a logical division based on a computational model, rather than a rigid physical cut.

[0024] Secondly, an unmixing matrix is ​​obtained, which characterizes the spatial response weight distribution of each photodetector to the optical signal in each sub-region. This is a key step in achieving parallel imaging and signal separation. Due to the difference in detector positions, for the optical signal reflected from the i-th sub-region, the intensity received by the i-th detector depends not only on the light intensity of that sub-region but also on the spatial response weight of that detector for that sub-region. To quantify this relationship, this embodiment constructs a linear equation model. Let... Indicates the first The photodetector in the first The mixed voltage or current value collected during the second sampling is the light intensity signal. Indicates the first The sub-regions in the first The independent response signal at each sampling is the light intensity signal that the sub-region should have contributed. Indicates the first The photodetector pairs with the first The spatial response weighting coefficients of each sub-region. The signal received by a detector can then be expressed as a weighted sum of the signals from each sub-region, as shown in the following expression:

[0025] Based on the above relationships, in order to simulate the imaging effect of multidimensional encoding, the following coefficient matrix relationship is constructed:

[0026] In the above formula, the left-hand vector represents the actual measured values ​​of the detector, the middle matrix is ​​the demixing matrix, and the right-hand vector represents the independent signals of each sub-region to be solved. To obtain the weighting coefficients in this demixing matrix... The value needs to be pre-calibrated or initialized. Specifically, this involves controlling the spatial light modulator to project a specific calibration pattern. For example, each sub-region in the target scene can be illuminated sequentially; when only the first sub-region is illuminated... When the number of sub-regions is reached, the response values ​​of all photodetectors are recorded. At this time, the number of... The readings of each detector reflect the weighting coefficient. The size of the unmixing matrix is ​​determined by traversing all sub-regions. Since the detector's spatial response is relatively stable once its position is fixed, this unmixing matrix typically remains unchanged throughout an imaging mission.

[0027] Next, the formal data acquisition phase begins. The spatial light modulator projects a series of coded patterns onto the target scene. These coded patterns represent corresponding modulation codes on the various sub-regions. To achieve efficient parallel sampling, this embodiment employs a hybrid coding strategy or a parallel coding strategy. That is, in a single projection, the entire working surface of the spatial light modulator is simultaneously loaded with coded information covering all sub-regions. At this time, the multiple photodetectors are used to acquire light intensity signals in parallel. Due to the superposition characteristics of light signals, the light intensity signal acquired by each photodetector... In reality, it contains mixed intensity components from multiple sub-regions, which is known as signal crosstalk. In conventional array imaging, this crosstalk is considered harmful noise, but in this embodiment, through the parallel imaging demixing model constructed above, this crosstalk is regarded as a linear combination containing rich spatial information.

[0028] Subsequently, based on the unmixing matrix, the light intensity signals collected by the multiple photodetectors are unmixed to separate the independent response signals corresponding to each sub-region. Specifically, the unmixing process involves solving the aforementioned system of linear equations. The process involves vectors. Mathematically, this can be achieved by calculating the inverse of the unmixing matrix. That is:

[0029] By multiplying the vector of light intensity signals collected by multiple photodetectors by the inverse matrix on the left, the signal can be obtained in each sampling. In the middle, from mixed signals The independent response signals of each sub-region are calculated. This step effectively eliminates signal interference between different imaging regions, achieving digital separation of signals.

[0030] Finally, based on the separated independent response signals and the modulation coding of the coded pattern in the corresponding sub-regions, the images of each sub-region are reconstructed. This step follows the basic principles of single-pixel imaging. For the first... For each sub-region, we have obtained a series of unmixed measurements. Meanwhile, we know the encoded pattern sequence applied to this sub-region during these sub-sampling. The image of this sub-region can be reconstructed using correlation algorithms or compressed sensing algorithms. The mathematical description of single-pixel imaging is:

[0031] in This represents the voltage or current value obtained by the detector. It represents the ratio between light intensity and voltage or current value. and These represent the distribution of the object to be reconstructed and the coded pattern, respectively. and These represent the number of pixels horizontally and vertically, respectively. This is achieved through measurements... and coded patterns By performing relevant calculations, such as inverse Fourier transform in Fourier single-pixel imaging or inverse Hadamard transform in Hadamard single-pixel imaging, the sub-region image can be recovered.

[0032] After reconstructing the images of all sub-regions, the reconstructed images of each sub-region are stitched together to form a complete high-resolution image of the target scene. Since spatial response weights have been considered in the demixing model, the reconstructed sub-region images have a clear correspondence in brightness and spatial location, and the final large field of view or high-resolution image can be synthesized through a seamless stitching algorithm.

[0033] The method in this embodiment significantly improves imaging efficiency. Compared to traditional single-pixel imaging technology, this invention uses only a limited number of pixels. Image reconstruction can be achieved using modulation patterns, where the target resolution is... , For the number of detectors. To reconstruct Taking a pixel-based image as an example, if four photodetectors are used to divide the image into four sub-regions, traditional single-pixel imaging requires full-field sampling, such as Fourier or Hadamard sampling, which requires 131,072 patterns. However, using the method in this embodiment, each sub-region only needs to undertake one-quarter of the sampling task, and these four regions are processed in parallel. Theoretically, the number of coded patterns required for full sampling is reduced to one-quarter, or 32,768. This means that the imaging speed is increased by four times, and as the number of detectors increases, the number of modulation patterns required for imaging will further decrease, further improving the imaging speed. Simultaneously, this method uses algorithmic demixing to replace physical isolation in hardware, reducing the stringent requirements on detector hardware and avoiding the problems of overheating and high cost associated with array detectors.

[0034] Example 2 This embodiment, based on Embodiment 1, further details the specific form of the encoding pattern and the hybrid encoding strategy to support the technical solutions regarding different types of modulation and coding combinations in the claims.

[0035] In this embodiment, the coding pattern projected by the spatial light modulator can adopt various forms of substrate patterns, mainly including Hadamard coding patterns and Fourier coding patterns. For the multiple sub-regions, this invention proposes a flexible coding allocation mechanism.

[0036] One approach is to use a hybrid coding method. This involves presenting a Hadamard pattern in the first sub-region and a Fourier pattern in the second sub-region, with the Hadamard and Fourier patterns combined to form the coded pattern. For example, in a target scene divided into left and right sub-regions, the left region can be modulated using a binary pattern generated by a Hadamard matrix, while the right region can be modulated using a Fourier stripe pattern with varying grayscale values. The advantage of this hybrid coding method is that it leverages the different sensitivities of different coding methods to various texture features. For instance, Hadamard coding is sensitive to point features, while Fourier coding is sensitive to periodic textures, thus acquiring richer frequency or spatial information in a single imaging session.

[0037] Another approach is to use the same type of modulation and coding combination. That is, the first sub-region presents a first hadamard pattern, the second sub-region presents a second hadamard pattern, and the first hadamard pattern and the second hadamard pattern are combined to form the coded pattern. In this case, although both are hadamard patterns, their corresponding spatial frequencies or sequences may be different, or they may simply be a splicing of spatial positions.

[0038] When Fourier coding is used as the modulation code, its generation expression follows the mathematical model as follows:

[0039] in, and These represent average light intensity and contrast ratio, respectively, and are generally set to 0.5. and Spatial frequency; For phase. To eliminate the DC component and obtain accurate Fourier spectrum coefficients, the Fourier encoding employs a four-step phase-shift method. Specifically, for each spatial frequency... and It is necessary to project four patterns with different phases, and their phases Set to 0 respectively , and .

[0040] The Fourier spectrum of the target object can be calculated using the four-step phase-shift method. :

[0041] in It is the imaginary unit. The measured intensity corresponds to different phases. Finally, an inverse Fourier transform is performed. Reconstructed Image :

[0042] This phase-shift-based differential measurement can effectively suppress ambient light noise. Combined with the parallel demixing strategy of this invention, the acquisition time can be significantly shortened while maintaining a high signal-to-noise ratio.

[0043] Example 3 This embodiment provides a specific dual-detector parallel imaging scheme to illustrate the technical configuration of the two photodetectors and two sub-regions in the claims.

[0044] In this embodiment, the resolution of the target scene to be imaged is set to 1. Pixels. To improve imaging speed, the scene is divided into two parts, each with a size of [missing information]. The pixel is divided into a first sub-region and a second sub-region. The system is configured with two photodetectors, designated as the first photodetector. Second photodetector These two detectors are placed independently in space and do not require strict optical alignment with the sub-region.

[0045] Due to differences in spatial location, the first photodetector Because it is closer to the first sub-region or at a more positive angle, the photodetector receives a stronger optical signal from the first sub-region. This is reflected in the spatial response weighting coefficient of the first photodetector to the first sub-region. It is relatively large; at the same time, it can also receive scattered or reflected light from the second sub-region, but the intensity is weaker, i.e., the spatial response weighting coefficient of the second sub-region. Smaller. Similarly, the second photodetector Spatial response weighting coefficients for the second sub-region Greater than its spatial response weighting coefficient for the first sub-region .

[0046] At this point, the unmixing matrix is ​​a 2x2 matrix, and its specific matrix equation form is as follows:

[0047] This matrix is ​​used to separate the independent response signals of the first sub-region and the second sub-region. In practice, this can be achieved by pre-illuminating the first and second sub-regions and measuring the output voltages of the two detectors. , as well as , The specific value is then calculated by inverting the matrix. and .

[0048] During imaging, the coded pattern projected by the spatial light modulator consists of two... The pixels are formed by stitching together the base patterns. For example, it could be two different Hadamard patterns, or a combination of a Hadamard pattern and a Fourier basis pattern. During the acquisition process, the detector... and Simultaneously, data is collected at the same sampling rate. Assuming full sampling reconstruction is required, for... For images like this, traditional methods require 16,384 acquisitions (and even more for differential measurements). In this embodiment, because the two regions are processed in parallel, the total number of projections depends only on the number of patterns required for a single sub-region, not the entire region. The required quantity for the region.

[0049] Comparative data shows that, for For natural images, traditional single-pixel imaging requires 131,072 modulation patterns. When using the method of this embodiment and expanding to four detectors, full sampling requires only 32,768 coded patterns, which is three-quarters less than the modulation patterns required by traditional single-pixel technology. Furthermore, with the increase in the number of detectors... As the number of modulation patterns increases, the required number of modulation patterns will increase accordingly. The proportion is further reduced. If the resolution is increased to... Traditional single-pixel imaging requires 524,288 modulation patterns, while using the hybrid coding technology of this invention with 4 detectors, only 131,072 patterns are needed. This not only significantly reduces imaging time but also reduces the pressure of data storage and processing.

[0050] Example 4 This embodiment provides a supplementary description of the hardware system for executing the above method.

[0051] The parallel single-pixel imaging system described in this embodiment includes a spatial light modulator as its core component. Specifically, this spatial light modulator is a digital micromirror device (DMD), which features high refresh rate and high contrast, making it ideal for projecting high-speed binary patterns. In another embodiment, a projector can be used as an integrated light source and modulator to directly project structured light onto the object. In applications where light intensity requirements are not high or phase modulation is needed, a liquid crystal display (LCD) can also be used as the spatial light modulator.

[0052] The detection module consists of multiple single-point photodetectors, which can be photodiodes, avalanche photodiodes, or photomultiplier tubes. They are connected to a computer via a data acquisition card (DAQ). The computer acts as the control center, sending synchronization signals and pattern data to the spatial light modulator while simultaneously reading multi-channel voltage signals from the DAQ card.

[0053] After the demixing algorithm is completed, the computer uses image processing software to stitch together the reconstructed images of each sub-region. The stitching process can be done directly based on the geometric boundaries of the sub-regions, or it can be fine-tuned using cross-correlation calculations of overlapping image areas to eliminate stitching gaps, ultimately forming a complete high-resolution image of the target scene. This non-array detection architecture avoids the need for custom-designed and expensive array detectors, while also solving the problems of inter-pixel crosstalk and heat dissipation difficulties associated with array detectors, providing an effective hardware solution for low-cost, high-performance computational imaging.

[0054] Furthermore, the illumination encoding required for the reconstruction process described in this invention can also utilize different combinations of Fourier transform, Hadamard transform, and random illumination patterns such as speckle to complete the projection pattern modulation process for target object imaging. The active single-pixel imaging method in this invention can also be replaced by a passive single-pixel imaging method.

[0055] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention. Contents not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A hybrid encoding and decoding parallel single-pixel imaging method, characterized in that, include: Multiple photodetectors are configured to receive light signals from the target scene, and the target scene is considered to be composed of multiple sub-regions; A spatial light modulator is controlled to project a series of hybrid coded patterns onto the target scene, wherein the coded patterns present corresponding modulation codes on the multiple sub-regions respectively; The light intensity signal is acquired in parallel using the multiple photodetectors, wherein the light intensity signal acquired by each photodetector contains a mixed light intensity component from multiple sub-regions; Obtain a demixing matrix, which characterizes the spatial response weight distribution of each photodetector to the optical signal in each sub-region; Based on the demixing matrix, the light intensity signals collected by the multiple photodetectors are demixed to separate the independent response signals corresponding to each sub-region. Based on the independent response signal and the modulation coding of the coding pattern in the corresponding sub-region, the image of each sub-region is reconstructed.

2. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 1, characterized in that, The unmixing matrix is ​​constructed using the following system of linear equations: in, Indicates the first The first photodetector collected the data of the first... Secondary light intensity signal Indicates the first Independent response signals for each sub-region Indicates the first The photodetector pairs with the first The spatial response weighting coefficients of each sub-region, the unmixing matrix being derived from... The coefficient matrix formed.

3. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 2, characterized in that, The steps for obtaining a demixing matrix include: The spatial light modulator is controlled to project a specific coded pattern, which is used to illuminate individual sub-regions individually or in combination; The corresponding light intensity response values ​​are collected using the multiple photodetectors; Calculate the weight coefficients of each element in the coefficient matrix based on the collected light intensity response values. The value of .

4. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 2, characterized in that, The step of performing demixing operations on the light intensity signals collected by the multiple photodetectors based on the demixing matrix includes: Calculate the inverse of the unmixing matrix; The vector formed by the light intensity signals collected by the multiple photodetectors is multiplied by the inverse matrix on the left to calculate the independent response signal. .

5. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 1, characterized in that, The encoded pattern is composed of a combination of different types of modulation codes; The first sub-region presents a Hadamard pattern, and the second sub-region presents a Fourier pattern. The Hadamard pattern and the Fourier pattern are combined to form the coded pattern.

6. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 1, characterized in that, The encoded pattern is composed of a combination of the same type of modulation codes; The first sub-region presents a first Hadhamaki pattern, the second sub-region presents a second Hadhamaki pattern, and the first Hadhamaki pattern and the second Hadhamaki pattern are combined to form the coded pattern.

7. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 1, characterized in that, The modulation coding is Fourier coding, and the generation expression of the Fourier coding is: in, The average light intensity. For contrast, and For spatial frequency, For phase; Furthermore, the Fourier encoding employs a four-step phase-shift method, with phase... Take respectively , , and .

8. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 1, characterized in that, The plurality of photodetectors are two in number, namely a first photodetector and a second photodetector; the plurality of sub-regions are two in number, namely a first sub-region and a second sub-region. Wherein, the spatial response weighting coefficient of the first photodetector to the first sub-region is greater than its spatial response weighting coefficient to the second sub-region, and the spatial response weighting coefficient of the second photodetector to the second sub-region is greater than its spatial response weighting coefficient to the first sub-region. The unmixing matrix is: A matrix is ​​used to separate the independent response signals of the first sub-region and the second sub-region.

9. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 1, characterized in that, The spatial light modulator is a digital micromirror device (DMD), a projector, or a liquid crystal display (LCD).

10. The hybrid encoding and decoding parallel single-pixel imaging method according to claim 1, characterized in that, Also includes: The reconstructed images of each sub-region are stitched together to form a complete high-resolution image of the target scene.

Images