A synchronous sampling type single-pixel imaging system and image reconstruction method based on DLP projection

Abstract

The application provides a synchronous sampling type single-pixel imaging system and an image reconstruction method based on DLP projection, and the method comprises the following steps: constructing an orthogonal coding pattern sequence based on a pattern library, dividing the coding pattern sequence into different pattern sections, each pattern section containing at least one frame of pattern, and assigning a pattern number to each pattern; loading the coding pattern sequence to a DLP control module and projecting to a target object; starting the integration sampling of the light intensity signal collected by a photoelectric detector by a trigger signal to obtain an effective light intensity value corresponding to the pattern number; estimating the sampling energy distribution of each region of the pattern according to the collected pattern number and the corresponding light intensity value; identifying a low coverage region with energy lower than a threshold value to generate a pattern rescheduling set; fusing and averaging the light intensity values of the repeatedly sampled patterns, giving a higher weight to the sampling values of the corresponding patterns, and reconstructing the target pattern by a weighted back projection algorithm.

Description

Technical Field

[0001] This invention belongs to the field of image processing, and particularly relates to a synchronous sampling single-pixel imaging system and image reconstruction method based on DLP projection. Background Technology

[0002] Single-pixel imaging technology is a novel imaging method that uses a spatial light modulator to encode the light field, then a single-point detector collects the overall light intensity and combines it with pattern information to complete image reconstruction. This method avoids the use of expensive or difficult-to-manufacture area array detectors, making it particularly suitable for imaging applications in special bands such as infrared and terahertz, as well as in scattering media environments. Therefore, it has significant value in scientific research and teaching. In recent years, single-pixel imaging systems based on digital light processing (DLP) projection have been widely used due to their compact structure, flexible control, and low cost. The system typically consists of a DLP projection module, an avalanche photodiode detector (APD), a data acquisition card, and a computing unit. The DLP module projects a series of coded patterns onto the target, and the reflected light is sampled by the detector and sent to the computing unit for decoding and reconstruction. However, in practical applications, existing systems still have key problems. First, the timing asynchrony between pattern projection and sampling is severe, easily leading to frame misalignment and data mismatch, resulting in distorted reconstructed images. Secondly, most current systems rely solely on preset, fixed pattern sequences, failing to flexibly adjust based on the response of different regions during sampling. This often results in undersampling in some areas and blurred image edges. Thirdly, the uneven distribution of light intensity across regions during the imaging process leads to a lack of effective compensation mechanisms when local areas contribute insufficient data, resulting in poor uniformity in the reconstructed image. Finally, even with repeated sampling, existing methods lack reasonable fusion strategies, potentially leading to overweighting of repeated patterns or insufficient utilization of information. These issues make it difficult to guarantee data stability and result reliability in single-pixel imaging systems during teaching experiments and scientific research verification. There is an urgent need for a novel system approach that can simultaneously address temporal synchronization, pattern scheduling, and spatial uniformity issues without increasing hardware complexity. Summary of the Invention

[0003] The purpose of this invention is to design a synchronous sampling single-pixel imaging system and image reconstruction method based on DLP projection, which can achieve the stability of synchronous sampling, the flexibility of pattern scheduling and the uniformity of image reconstruction, and improve the usability and reliability of single-pixel imaging technology in complex environments and high-precision teaching and scientific research.

[0004] To achieve the above objectives, a first aspect of the present invention provides a synchronous sampling single-pixel imaging method based on DLP projection, the method comprising:

[0005] An orthogonal coded pattern sequence is constructed based on a pattern library, and the coded pattern sequence is divided into different pattern segments. Each pattern segment contains at least one pattern frame, and a pattern number is assigned to each pattern frame. The first and last frames of each pattern segment are set as redundant frames. The number of pattern segments, the number of patterns in each segment, and the numbering rules are preset by the system.

[0006] The encoded pattern sequence is loaded into the DLP control module and projected onto the target object; the data acquisition card is activated by a trigger signal to perform integral sampling on the light intensity signal collected by the photodetector, and a rate of change suppression term is introduced to correct the sampled value to obtain the effective light intensity value corresponding to the pattern number;

[0007] Based on the collected pattern numbers and their corresponding light intensity values, and combined with the spatial structural features of the patterns, the sampling energy distribution of each region of the pattern is calculated; low-coverage regions with energy below the threshold are identified, and at least one pattern number that contributes the most to the region is selected to form a pattern rescheduling set;

[0008] The pattern is repeatedly sampled, the light intensity values ​​of the repeatedly sampled pattern are fused and averaged, and the sampled values ​​of the corresponding pattern are assigned higher weights according to the pattern rescheduling set. The target pattern is reconstructed through a weighted back projection algorithm to achieve compensation and enhancement for weak coverage areas.

[0009] Furthermore, the number of pattern segments M and the number of patterns per segment N are dynamically set according to the pattern resolution and system synchronization requirements; wherein, the total number of patterns is the product of M and N, the value of N ranges from 128 to 256, and M is determined by rounding down the ratio of the total number of patterns to N; the numbering rule is a continuous integer number, which is automatically generated by the calculation unit and stored in the pattern control index table.

[0010] Furthermore, the integral sampling mechanism is as follows: after receiving the trigger signal from the DLP control module, the data acquisition card opens a fixed time window for a preset duration to integrate and average the voltage signal output by the photodetector.

[0011] Furthermore, the rate of change suppression term is achieved by calculating the integral of the signal rate of change within the sampling time window; the sampled value is corrected by subtracting a term proportional to the absolute value of the signal derivative from the original integrated sampled value, the proportionality coefficient of which is an empirical constant determined in advance through system calibration.

[0012] Furthermore, the pattern is divided into Q non-overlapping rectangular sub-regions; the estimated sampling energy of each sub-region is calculated by the normalized average brightness value of all projected patterns in that region, its corresponding corrected light intensity value, and a regularization term used to balance the differences in contributions between patterns.

[0013] Furthermore, the steps to determine whether an area is a low-coverage area are as follows: calculate the median value of the estimated sampling energy of all areas and set a proportional threshold; if the energy value of a certain area is lower than the product of the median value and the proportional threshold, it is determined to be a low-coverage area.

[0014] Furthermore, the steps for generating the pattern rescheduling set are as follows: from the current pattern segment, select the top K patterns with the highest normalized average brightness value in the region, and include their numbers in the pattern rescheduling set.

[0015] Furthermore, the specific steps for fusing and averaging the light intensity values ​​of the repeatedly sampled pattern are as follows: add up the corrected light intensity values ​​of all the samples and divide by the number of samples to obtain the fused light intensity value corresponding to the pattern number.

[0016] A second aspect of the present invention provides a synchronous sampling single-pixel imaging system based on DLP projection, the system comprising:

[0017] The calculation unit is used to generate an orthogonal coded pattern sequence and divide the coded pattern sequence into different pattern segments. Each pattern segment contains at least one frame of pattern and assigns a pattern number to each frame of pattern. The first and last frames of each pattern segment are set as redundant frames. The number of pattern segments, the number of patterns in each segment, and the numbering rules are preset by the system.

[0018] The projection unit is used to load the coded pattern sequence into the DLP control module and project it onto the target object; the data acquisition card is activated by a trigger signal to perform integral sampling on the light intensity signal collected by the photodetector, and a rate of change suppression term is introduced to correct the sampled value to obtain the effective light intensity value corresponding to the pattern number;

[0019] The scheduling unit is used to calculate the sampling energy distribution of each region of the pattern based on the acquired pattern number and its corresponding light intensity value, combined with the spatial structure characteristics of the pattern; identify low-coverage regions with energy below a threshold, and select at least one pattern number that contributes the most to the region to form a pattern rescheduling set.

[0020] The output unit is used to repeatedly sample the pattern, fuse and average the light intensity values ​​of the repeatedly sampled pattern, and assign higher weights to the sampled values ​​of the corresponding pattern according to the pattern rescheduling set. The target image is reconstructed through a weighted back projection algorithm to achieve compensation and enhancement of weak coverage areas.

[0021] Furthermore, the projection unit includes a light source, an illumination system, a DMD controller, a DMD chip, and an adjustable projection lens; the DMD controller is used to receive and buffer the encoded pattern sequence from the computing unit, and control the micromirror array on the DMD chip to flip to spatially modulate the light beam; the adjustable projection lens is used to project the modulated light beam onto the target object and adjust the size of the projected image.

[0022] The beneficial technical effects of the present invention are at least as follows:

[0023] To address the aforementioned issues, this invention provides a synchronous sampling single-pixel imaging system and image reconstruction method based on DLP projection. By employing a segmented pattern organization and setting redundant frames for each segment, frame-level triggering synchronization and abnormal frame removal are achieved, ensuring a one-to-one correspondence between sampled data and pattern numbers. An integral mechanism and a rate-of-change suppression term are introduced during the sampling process to effectively counteract noise interference from high-frequency micromirror switching and low-light conditions, ensuring more stable and reliable light intensity values. In the image coverage analysis stage, the energy distribution of each region is inferred by combining the overall sampled values ​​with pattern structural features, and differential regularization is used to suppress local deviations, thereby automatically selecting the pattern numbers with the most compensation value for weakly covered areas, forming a pattern rescheduling set. In the final image reconstruction stage, multiple sampled values ​​are fused, and a weighting factor based on the rescheduling results is introduced, so that the patterns corresponding to weakly covered areas are reasonably enhanced during the decoding process, resulting in a more balanced and clearer final output image. Through this adaptive closed-loop design, the present invention enables the system to achieve stability in synchronous sampling, flexibility in pattern scheduling, and uniformity in image reconstruction without changing the hardware architecture, significantly improving the usability and reliability of single-pixel imaging technology in complex environments and high-precision teaching and research. Attached Figure Description

[0024] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0025] Figure 1 This is a flowchart of a synchronous sampling single-pixel imaging method based on DLP projection according to the present invention.

[0026] Figure 2 This is a framework diagram of a synchronous sampling single-pixel imaging system based on DLP projection according to the present invention.

[0027] Figure 3 This is a schematic diagram of the projection unit in an embodiment of the present invention. Detailed Implementation

[0028] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0029] In one or more embodiments, such as Figure 1 As shown, a synchronous sampling single-pixel imaging method based on DLP projection is disclosed, the method comprising the following:

[0030] S1: Construct an orthogonal coded pattern sequence based on the pattern library, and divide the coded pattern sequence into different pattern segments. Each pattern segment contains at least one frame of pattern, and a pattern number is assigned to each frame of pattern. The first and last frames of each pattern segment are set as redundant frames. The number of pattern segments, the number of patterns in each segment, and the numbering rules are preset by the system.

[0031] Specifically, the first step is to construct an orthogonal coded pattern sequence based on a pattern library. This pattern library is typically generated in advance using the Hadamard transform method and is programmed by a computer. The pattern size is consistent with the resolution of the final reconstructed image, such as 64×64 or 128×128 pixels. The patterns are stored in binary form (0 or 1) and are numbered according to their pattern number. Arrangement, among which Indicates the number of the pattern segment. This indicates the frame number within the current segment. The pattern size, format, and order are all preset in the calculation unit and uploaded to the DLP control module memory.

[0032] This step introduces a segmented pattern structure and encodes it into a numbering system that facilitates subsequent trigger control and data alignment. Traditional single-pixel imaging systems often load all patterns frame by frame in complete sequence, making it difficult to establish a synchronized logical structure between sampling and pattern numbering, and making it difficult to locate synchronization errors. This step addresses this by setting the pattern segment length... The entire pattern sequence is divided into Section, that is:

[0033] ;

[0034] Among them, the pattern number It is the first The first in the paragraph The pattern is automatically labeled by the numbering generator during the pattern loading phase. The numbering generator is a set of integer sequence generation functions running within a computational unit; its essence is a two-dimensional integer matrix mapping function, defined as:

[0035] ;

[0036] in, The pattern number within the segment is an integer with a value range of 1. The numbering function generates the numbering information, which is then stored in the computer as an array and output to the DLP controller. It represents the pattern segment number, which is an index of the segment structure and is used for segment identification and subsequent synchronization control. This indicates the pattern number within the segment, serving as the sequential index for the patterns in the current segment. The number of patterns per segment, usually 128 or 256, is a system setting parameter that controls the segment length. The total number of segments is determined by the total number of patterns and the segment length, and is equal to the total number of patterns divided by 1 / 2. .

[0037] The advantage of this numbering method lies in its ability to create a pattern organization structure with boundaries, allowing the DLP controller to insert synchronization control, sampling status judgment, and pattern adjustment operations after each projection segment is completed. The numbering sequence is generated by a computer control program and uploaded to the DLP control module via USB or HDMI interface during system initialization. This module sequentially reads the numbered patterns through a caching mechanism and drives the micromirror array to flip for projection. Each pattern segment has a default preset redundant first and last frame (numbered respectively) during loading. and These two frames are used to stabilize the output state of the DLP system and will be marked as waste frames in the sampled data.

[0038] Specifically, if 2048 patterns are generated, the number of frames per pattern segment is set. The total number of pattern segments Pattern number from Start, numbered sequentially to The numbering process requires no manual intervention; it is completed automatically by the numbering function and stored in the pattern control index table. During system operation, the system uses instructions from this table to perform pattern scheduling and frame synchronization control.

[0039] S2: Load the coded pattern sequence into the DLP control module and project it onto the target object; start the data acquisition card through a trigger signal to perform integral sampling on the light intensity signal collected by the photodetector, and introduce a rate of change suppression term to correct the sampled value to obtain the effective light intensity value corresponding to the pattern number;

[0040] Specifically, the input for this step is a set of pattern numbers. and pattern segment structure index . This indicates that the pattern number sequence after segmented structure encoding has been loaded into the DLP control module in order; This indicates the segment number, which serves as the grouping basis in pattern loading and sampling result labeling. Each number... A 64×64 pixel binary pattern is spatially modulated by the DMD chip in the DLP6500 module, illuminated by an LED white light source, and projected onto the target surface in numerical order after reflection by the DMD micromirror array. The DLP module has received and loaded the complete number sequence, and each segment initiates the sequential projection operation within the segment via a command frame before execution.

[0041] The core task of this step is to sequence the numbered patterns. Frame by frame projection, and through a frame-level synchronization mechanism with the sampling system, a one-to-one match is achieved between the pattern number and the light intensity sample value, thereby forming a sampling data pair. This is used for subsequent reconstruction and region analysis. To achieve this goal, the system uses each frame of the pattern... When projected by a DLP, a data acquisition card (such as the NI USB-6356) is simultaneously triggered via a TTL interface to sample the APD signal at the corresponding moment. The APD (Avalanche Photodiode) is fixed on the light reflection path of the target area and is responsible for converting the light reflected from the target pattern into an analog current signal. The signal is first converted into voltage form by a transimpedance amplifier, and then digitally sampled by the ADC module of the data acquisition card.

[0042] Because the pattern change frequency is high (usually set above 240Hz), and the DLP module and sampling module in the system are controlled by independent clocks, timing jitter is prone to occur in light intensity sampling. Therefore, this step introduces a sampling integration mechanism and sets a fixed sampling time window. Integral sampling is initiated immediately after each TTL trigger:

[0043] ;

[0044] in, Indicates pattern number The sampled light intensity value corresponding to the projection. This is the curve showing the change of the APD output voltage over time; this curve is output by the amplifier circuit. The moment when the pattern begins projection is defined by the TTL-triggered edge. The fixed sampling window set for the system is usually set in the range of 3~5ms to adapt to different optical stability conditions.

[0045] To address the sampling instability caused by low light and high pattern variation, a pattern variation rate suppression term is further introduced into the integral formula. This term constrains excessively large instantaneous fluctuations and suppresses the strong instantaneous reflection interference caused by high-frequency micromirror state switching. The specific modified sampling result is expressed as follows:

[0046] ;

[0047] variable This represents the corrected sampled value, used in subsequent processing. This correction term introduces a differential energy measure to detect the amplitude of electrical signal fluctuations during the integration period. When the reflected signal changes too drastically (usually caused by disturbances in the DMD switching mirror group), the sampled value is appropriately reduced. Coefficient These are empirical parameters, typically set to... The time interval is determined once during the system startup calibration phase and then fixed. In this way, the system effectively suppresses short-term peaks caused by micromirror oscillation and instability of the reflection focal plane, improving the stability of the final image reconstruction.

[0048] After the sampling data collection is completed, the system will associate each number with its corresponding... Structured data pairs are formed, and the sampled data is cached in real time in the memory of the acquisition card, and then transmitted to the main control computer via USB interface. Segment number Similarly, it is marked as an index in the cache structure for use in subsequent sampling grouping and pattern filtering operations. Furthermore, in each sampling process, it is numbered... and The corresponding frame remains a stable redundant frame; its sampled values ​​are recorded but not used for any subsequent analysis. The set of valid sampled numbers for the entire segment is: ,in The corresponding sample value is .

[0049] S3: Based on the collected pattern numbers and their corresponding light intensity values, combined with the spatial structure characteristics of the patterns, estimate the sampling energy distribution of each region of the pattern; identify low-coverage regions with energy below the threshold, and select at least one pattern number that contributes the most to the region to form a pattern rescheduling set;

[0050] Specifically, the input to this step is the set of sampled data structures output from the previous step. ,in Indicates the first Duan Di The numbering of the frame pattern This is the effective light intensity value of the frame corresponding to the numbered pattern after projection, corrected by integral sampling and micromirror perturbation suppression mechanism. Each pattern is numbered... Corresponding unique spatial pattern layout The pattern is pre-loaded in the DLP module, and the pattern content can be directly extracted from the main control computer using a numbered index. Since this system uses a single-pixel architecture and lacks spatial resolution, it is necessary to perform region mapping analysis on these overall light intensity responses to determine whether different regions in the image have received sufficient light intensity sampling.

[0051] The core objective of this step is to estimate the sampling energy distribution of the entire image in different regions using existing sampling pattern numbers and corresponding light intensity values, identify low-coverage areas, and construct a set of pattern numbers that require sampling compensation based on this information. This is used for the scheduling of the next sampling pattern.

[0052] To achieve region-aware sampling distribution evaluation, the image space is equally divided into Non-overlapping sub-regions Each region For a subset of pixels in an image matrix, for example, dividing a 64×64 image into an 8×8 grid, then each Each frame is an 8x8 pixel block. The structural features are spatially deconstructed using the regional modulation weight matrix, that is, the normalized average brightness of the pattern in each region is calculated, representing the modulation capability of the pattern in that region:

[0053] ;

[0054] Here Number the pattern In the region The average brightness within the area is an indicator of the pattern's contribution to that region. Since the pattern is binary, therefore... Combining the spatial modulation capability of the pattern with the light intensity response of its corresponding frame. Construct the sampling intensity contribution of the pattern in this region:

[0055] ;

[0056] In this expression, Indicates the region Estimated sampling energy; For the region Average modulation capability across all patterns within this segment. It is the modulation balance adjustment coefficient, usually set to 1. The range is used to suppress abnormal weight boosting when the contribution of pattern structure to the region varies too much. The introduction of this regularization term is designed to address the typical problem of a small number of high-structure patterns making misleading contributions. It effectively avoids misjudgments caused by individual patterns strongly covering a local area but having a low response.

[0057] In actual calculations, all patterns This was generated in the first step and stored in the data table corresponding to the number index. The main control program then uses the current segment number... and frame number Extract pattern, calculate Then compared with the sampled value Multiplication yields the local light intensity contribution of the region. This operation is performed at the segment level, meaning it is executed once for the entire region within each segment. The calculation results form the sampling energy distribution map of the current segment of the image.

[0058] The method to determine whether a certain region is undersampled is: calculate the energy of the entire region. median And set a proportional threshold. (e.g., 0.65), if a certain region satisfies If so, it is determined to be a low-coverage area. The system will then backtrack all instances within this segment. China's largest contributor to the region A pattern (usually taken) ), and add these pattern numbers to the pattern collection. Prepare to load repeated sampling in the next segment.

[0059] Specifically, in the section In the middle, the area was discovered energy ,and far below The system filters out items from segment 4 that are... superior The three largest pattern numbers, such as , , These three will be added And then, in the next pattern scheduling, a supplementary sampling insertion is performed.

[0060] S4: Repeatedly sample the pattern, fuse and average the light intensity values ​​of the repeatedly sampled pattern, and assign higher weights to the sampled values ​​of the corresponding pattern according to the pattern rescheduling set. The target pattern is reconstructed through a weighted back projection algorithm to achieve compensation and enhancement for weak coverage areas.

[0061] Specifically, the input for this step is the set of valid sample pairs accumulated in the second step. and the set of rescheduling pattern numbers output in the third step. .in, It is the pattern number. Is the pattern in the 1st The corrected light intensity value is obtained from the second sampling. This is because the third step identifies and outputs low-coverage areas. These pattern numbers are actually the most sensitive codes for weakly covered areas. Therefore, in the image decoding process of this step, it is necessary to give extra weight to the sampled values ​​corresponding to these pattern numbers in order to compensate for the spatial imbalance problem in the imaging process.

[0062] First, duplicate sample values ​​for all pattern numbers are merged to ensure that each number corresponds to only one stable sample value. For patterns... It is used throughout the sampling process. The sample value is [number]. The fusion value is calculated as follows:

[0063] ;

[0064] in Indicates number The average sample value, This represents the number of sampling times. This step ensures that the patterns projected from different segments do not become numerically unbalanced due to varying numbers of occurrences.

[0065] To incorporate the region identification results from the third step, the system introduces a pattern weighting factor. This factor is composed of Control: If ,but To enhance the contribution of the pattern corresponding to weakly covered areas; if ,but In implementation, The typical range of values ​​is The formula was determined through experimental calibration. The final image reconstruction uses a weighted linear back-projection formula:

[0066] ;

[0067] in To reconstruct the image at the pixel level The value, It is a pattern The value of this pixel (0 or 1), and through Convert to Symmetric encoding; These are the fused sampled values; These are weighting coefficients used for compensation in areas with weak coverage.

[0068] Specifically, assuming a pattern It was identified as a key pattern in low-coverage areas in the third step and was therefore added. During reconstruction, Setting it to 1.1 increases the sampling contribution of the pattern by 10% during backprojection, thereby enhancing the restoration of that region in the target image. For those not... The pattern number in the image is kept with a weight of 1 to avoid over-correction.

[0069] The final output is the final reconstructed image. This is a two-dimensional matrix, with pixel values ​​being floating-point numbers. The image not only fuses all sampled data but also incorporates data from a third step. Compensatory enhancements were made to areas with weak coverage.

[0070] In one or more embodiments, such as Figure 2 As shown, a synchronous sampling single-pixel imaging system based on DLP projection is disclosed, the system comprising:

[0071] The calculation unit is used to generate an orthogonal coded pattern sequence and divide the coded pattern sequence into different pattern segments. Each pattern segment contains at least one pattern frame, and a pattern number is assigned to each pattern. The first and last frames of each pattern segment are set as redundant frames. The number of pattern segments, the number of patterns in each segment, and the numbering rules are preset by the system.

[0072] The projection unit is used to load the coded pattern sequence into the DLP control module and project it onto the target object; the data acquisition card is activated by a trigger signal to perform integral sampling on the light intensity signal collected by the photodetector, and a rate of change suppression term is introduced to correct the sampled value to obtain the effective light intensity value corresponding to the pattern number;

[0073] The scheduling unit is used to estimate the sampling energy distribution of each region of the pattern based on the acquired pattern number and its corresponding light intensity value, combined with the spatial structure characteristics of the pattern; identify low-coverage regions with energy below a threshold, and select at least one pattern number that contributes the most to the region to form a pattern rescheduling set;

[0074] The output unit is used to fuse and average the light intensity values ​​of the repeatedly sampled patterns, and assign higher weights to the sampled values ​​of the corresponding patterns according to the pattern rescheduling set, and reconstruct the target image through a weighted back projection algorithm to achieve compensation and enhancement of weak coverage areas.

[0075] Among them, such as Figure 3 As shown, the projection unit includes a light source, an illumination system, a DMD controller, a DMD chip, and an adjustable projection lens; the DMD controller is used to receive and buffer the encoded pattern sequence from the computing unit, and control the micromirror array on the DMD chip to flip to spatially modulate the light beam; the adjustable projection lens is used to project the modulated light beam onto the target object and adjust the size of the projected image.

[0076] Specifically, the base image is first loaded into the DMD controller, enabling the DMD chip to flip and display the corresponding pattern. Then, one or a combination of RGB light sources is selected, and light is shone onto the DMD chip through an illumination system. The micromirrors on the DMD chip flip according to the loaded pattern, modulating the light beam. Finally, after adjustment by the projection lens, these modulated light beams can change the image size and project it onto the target object, forming the desired projection effect.

[0077] It is worth noting that the specific workflow of the synchronous sampling single-pixel imaging system based on DLP projection provided in this embodiment of the invention is the same as that of the synchronous sampling single-pixel imaging method based on DLP projection described in the above embodiment, and will not be repeated here.

[0078] This invention also provides a synchronous sampling single-pixel imaging device based on DLP projection, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps described in the above embodiment of the synchronous sampling single-pixel imaging method based on DLP projection, for example... Figure 1 The steps S1 to S4 described above; or, when the processor executes the computer program, it implements the functions of each module in the above system embodiments.

[0079] For example, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the DLP projection-based synchronous sampling single-pixel imaging device.

[0080] The aforementioned DLP projection-based synchronous sampling single-pixel imaging device can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. This device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the device may also include input / output devices, network access devices, and buses.

[0081] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASACs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the DLP projection-based synchronous sampling single-pixel imaging device, connecting all parts of the device via various interfaces and lines.

[0082] The memory can be used to store the computer program and / or modules. The processor implements various functions of the synchronous sampling single-pixel imaging device based on DLP projection by running or executing the computer program and / or modules stored in the memory, and by calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created according to the operation of the air conditioner controller, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0083] The module integrated into the synchronous sampling single-pixel imaging device based on DLP projection, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0084] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0085] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method of image reconstruction based on DLP projection, characterized in that, The method comprises: Based on the pattern library, an orthogonal coding pattern sequence is constructed, and the coding pattern sequence is divided into different pattern segments, each pattern segment containing at least one frame of pattern, and each frame of pattern is assigned a pattern number; the first and last frames of each pattern segment are set as redundant frames; the number of pattern segments, the number of patterns in each segment and the numbering rule are preset by the system; The coding pattern sequence is loaded into a DLP control module and projected onto a target object; a data acquisition card is triggered by a trigger signal to integrate and sample the light intensity signals collected by a photodetector, and a change rate suppression term is introduced to correct the sampling value, thereby obtaining effective light intensity values corresponding to the pattern numbers; According to the collected pattern numbers and their corresponding light intensity values, and in combination with the spatial structure characteristics of the patterns, the sampling energy distribution of each region of the pattern is calculated; low-coverage regions with energy lower than a threshold value are identified, and at least one pattern number with the largest contribution to the region is selected from the low-coverage regions to form a pattern rescheduling set; The patterns are repeatedly sampled, the light intensity values of the repeatedly sampled patterns are fused and averaged, and the sampling values of the corresponding patterns are given higher weights according to the pattern rescheduling set, and the target pattern is reconstructed by a weighted back projection algorithm to realize compensation and enhancement of the weak-coverage regions.

2. The method of claim 1, wherein, The number of pattern segments M and the number of patterns in each segment N are dynamically set according to the pattern resolution and system synchronization requirements; wherein the total number of patterns is the product of M and N, the value range of N is 128 to 256, and M is determined by rounding according to the ratio of the total number of patterns to N; the numbering rule is a continuous integer number, which is automatically generated by the calculation unit and stored in the pattern control index table.

3. The method of claim 1, wherein, The integral sampling specifically comprises: after receiving a trigger signal from the DLP control module, the data acquisition card opens a fixed time window with a preset duration, integrates and averages the voltage signals output by the photodetector.

4. The method of claim 1, wherein, The change rate suppression term is realized by calculating the integral of the signal change rate in the sampling time window; the correction of the sampling value is to subtract a term proportional to the absolute value of the signal differential from the original integral sampling value, and the proportional coefficient is an empirical constant determined by system calibration in advance.

5. The method of claim 1, wherein, The pattern is equally divided into Q non-overlapping rectangular sub-regions; the estimated sampling energy of each sub-region is calculated by the normalized average brightness value of all projected patterns in the region, the corresponding corrected light intensity value, and a regularization term for balancing the contribution difference between patterns.

6. The DLP projection based image reconstruction method of claim 1, wherein, The step of determining whether a region is a low-coverage region is: calculating the median value of the estimated sampling energy of all regions, and setting a proportional threshold δ; if the energy value of a region is lower than the product of the median value and the proportional threshold, it is determined to be a low-coverage region.

7. The DLP projection based image reconstruction method of claim 1, wherein, The generation step of the pattern rescheduling set is: from the current pattern segment, the top K patterns with the highest normalized average brightness value on the region are selected, and their numbers are included in the pattern rescheduling set.

8. The DLP projection based image reconstruction method of claim 1, wherein, The specific steps of fusing and averaging the light intensity values of the repeatedly sampled patterns are: adding all the secondary collected corrected light intensity values of the pattern and dividing by the number of samplings to obtain the fused light intensity value corresponding to the pattern number.

9. A DLP projection based synchronous sampling type single-pixel imaging system, characterized in that, The system comprises: A computing unit is configured to generate a sequence of orthogonal coded patterns, divide the sequence into different pattern segments, each of which contains at least one frame of pattern, and assign a pattern number to each frame of pattern; the first and last frames of each pattern segment are set as redundant frames; the number of pattern segments, the number of patterns in each segment, and the numbering rule are preset by the system; A projection unit is configured to load the sequence of coded patterns into a DLP control module and project onto a target object; a trigger signal is used to start the integration sampling of the light intensity signal collected by the photodetector by a data acquisition card, and a change rate suppression term is introduced to correct the sampling value, thereby obtaining the effective light intensity value corresponding to the pattern number; A scheduling unit is configured to calculate the sampling energy distribution of each region of the pattern according to the collected pattern number and the corresponding light intensity value, in combination with the spatial structure characteristics of the pattern, identify the low-coverage region with energy lower than a threshold, and select at least one pattern number that contributes most to the region from the low-coverage region to form a pattern rescheduling set; An output unit is configured to repeatedly sample the pattern, fuse and average the light intensity values of the repeatedly sampled patterns, give a higher weight to the sampling value of the corresponding pattern according to the pattern rescheduling set, and reconstruct the target image by a weighted back-projection algorithm to realize the compensation and enhancement of the weak-coverage region.

10. The DLP projection based synchronous sampling type single-pixel imaging system according to claim 9, wherein, The projection unit includes a light source, an illumination system, a DMD controller, a DMD chip, and an adjustable projection lens; the DMD controller is configured to receive and cache the sequence of coded patterns from the computing unit, control the micro-mirror array on the DMD chip to flip to spatially modulate the light beam; and the adjustable projection lens is configured to project the modulated light beam onto the target object and adjust the size of the projected image.

Images