A method for frame synchronization and outlier removal in light intensity sampling based on a DLP projection system

Abstract

This invention proposes a frame synchronization and light intensity sampling outlier removal method based on a DLP projection system. The method includes: acquiring the original image to be projected and extracting a structural feature vector; calculating the structural predicted light intensity of the original image based on the structural feature vector; acquiring at least one light intensity sample value within a fixed time window after the original image is projected to form a light intensity sequence; calculating the average light intensity of the light intensity sequence, and combining the structural predicted light intensity and the structural feature vector, introducing a structural response correction term and a structural adaptability tolerance term to determine whether the sampling of the current frame is successfully synchronized with the pattern projection; if the sampling of the current frame is successfully synchronized, constructing an outlier removal threshold, and identifying and removing outliers from the light intensity sequence; calculating the variance of the remaining sampling sequence after outlier removal; and weighted fusing the arithmetic mean with the median of the remaining sequence to finally generate a light intensity output value and submit it to the image reconstruction module.

Description

Technical Field

[0001] This invention belongs to the field of image processing, and in particular relates to a method for frame synchronization and light intensity sampling outlier removal based on a DLP projection system. Background Technology

[0002] Single-pixel imaging is an imaging method that reconstructs images by spatially modulating a light field and using a single-point photodetector. Compared to traditional area array detectors, single-pixel imaging has significant advantages in non-visible light bands, low-light environments, and scattering imaging scenarios, and is particularly suitable for conditions involving high cost, high noise, or where area array detectors are difficult to use. With the development of digital light processing technology, DLP-based projection systems, due to their high-speed modulation, digital control, and programmable pattern loading capabilities, have gradually become the core spatial light modulators in single-pixel imaging systems. However, in practical applications, due to differences in the hardware triggering mechanisms between the DLP module and the sampling module (such as the APD and the data acquisition card), misalignment between pattern projection and light intensity sampling timing often occurs, leading to inaccurate frame synchronization, trigger delay fluctuations, or misalignment between pattern switching and sampling acquisition. These problems are more pronounced in high-speed projection or low-exposure tolerance experiments, easily causing frame misalignment or missampling, directly resulting in image reconstruction distortion. Meanwhile, the light intensity sampling process itself is also affected by factors such as minor jitter of optical devices, external ambient light disturbances, and detector noise, resulting in random outliers or sudden fluctuations in the sampling sequence, especially when the pattern complexity is high or the bright spots are unevenly distributed. Traditional methods usually rely on simple time delay compensation or first and last frame removal to solve the synchronization problem, and on fixed threshold methods (such as the 3σ criterion) to remove outliers. However, these methods are too coarse and cannot adapt to the dynamic changes of different pattern structures and actual system states, often resulting in misjudgments and omissions, reducing the reliability of the sampled data. Since the data of single-pixel imaging systems is highly compressed, each sampled value has a significant impact on the final reconstruction. Therefore, relying solely on fixed delays or global thresholds is insufficient to guarantee data accuracy. How to achieve a mechanism that can both ensure frame synchronization and dynamically remove outliers has become a major bottleneck in the promotion of high-precision applications of this type of system. Summary of the Invention

[0003] The purpose of this invention is to design a frame synchronization and light intensity sampling outlier removal method based on a DLP projection system, which can achieve frame synchronization and outlier removal in single-pixel imaging without relying on external high-precision synchronization hardware, thereby significantly improving the accuracy of reconstructed images and the quality of sampling data.

[0004] To achieve the above objectives, this invention provides a method for frame synchronization and outlier removal in light intensity sampling based on a DLP projection system, the method comprising:

[0005] The original image to be projected is acquired and a structural feature vector is extracted. The structural feature vector includes the number of bright spots in the pattern and the average brightness of the pattern. Based on the preset mapping relationship between structural features and light intensity response, the structural predicted light intensity of the original image is calculated according to the structural feature vector.

[0006] At least one light intensity sample value is collected within a fixed time window after the original image is projected to form a light intensity sequence; the average light intensity of the light intensity sequence is calculated, and combined with the structure prediction light intensity and the structure feature vector, a structure response correction term and a structure adaptability tolerance term based on the spatial distribution location of the bright spot are introduced to determine whether the sampling of the current frame is successfully synchronized with the pattern projection;

[0007] If the sampling synchronization of the current frame is determined to be successful, then based on the structural feature vector and the structural predicted light intensity, an outlier removal threshold that depends on both the pattern structure complexity and the overall sampling deviation is constructed. Outlier identification and removal are performed on the light intensity sequence, and the arithmetic mean of the remaining sampled values ​​after removal is calculated.

[0008] Based on the remaining sampled sequence after removing outliers, its variance is calculated; the arithmetic mean is then weighted and fused with the median of the remaining sequence, where the weighting weight is determined by the structural variance sensing control factor, and finally a stable light intensity output value is generated and submitted to the image reconstruction module.

[0009] Furthermore, the steps for extracting the number of bright spots in the pattern include: performing binarization on the pattern image and counting the number of pixels with gray values ​​higher than a set threshold.

[0010] Furthermore, the mapping relationship is established by collecting the structural features of at least one sample pattern and its average light intensity value measured under standard reflection conditions during the system initialization phase, and then performing a weighted linear fit.

[0011] Furthermore, the structural response correction term is calculated based on the distance between the centroid of all bright spots in the current pattern and the image center.

[0012] Furthermore, the structural adaptability tolerance term is dynamically adjusted based on the ratio of the number of bright spots in the current pattern to the maximum number of bright spots observed by the system.

[0013] Furthermore, the construction of the outlier removal threshold introduces a structural bias regularization term, which weights the pattern bright spot density using a nonlinear power function.

[0014] Furthermore, the construction of the outlier removal threshold also introduces an overall residual index, which is the relative deviation between the average value of the light intensity sequence and the light intensity predicted by the structure.

[0015] Furthermore, the value of the structural variance sensing control factor takes the minimum of the following two terms: the first term is the ratio of the current frame sampling variance to the system maximum variance, and the second term is the product of the pattern structure sparsity and the structural sensitivity coefficient.

[0016] Furthermore, the condition for determining successful synchronization is that the absolute value of the algebraic difference between the average light intensity, the predicted light intensity of the structure, and the structure response correction term is less than or equal to the structure adaptability tolerance term.

[0017] Furthermore, the light intensity sampling value is generated by the reflected light from the response pattern of the avalanche photodetector and is collected through a data acquisition card.

[0018] The beneficial technical effects of the present invention are at least as follows:

[0019] To address the aforementioned issues, this invention provides a method for frame synchronization and outlier removal in light intensity sampling based on a DLP projection system. By quantitatively analyzing the structural features of the projected pattern, a light intensity prediction model corresponding to the number of bright spots and average brightness is established, providing a benchmark for subsequent judgment. Based on this benchmark, this invention introduces a structure-response-aware synchronization calibration mechanism in the light intensity sampling stage. This allows the average light intensity within the sampling window to be compared with the predicted light intensity and structural correction, ensuring strict alignment between the sampled values ​​and the projected pattern and avoiding frame misalignment due to triggering errors. Furthermore, this invention proposes a pattern-sensitive outlier removal strategy. By combining local statistical indicators with structural features to construct a dynamic threshold, it achieves accurate identification and removal of sudden outlier sampling points, avoiding the failure of traditional fixed-threshold methods under complex patterns. Finally, this invention proposes a stable light intensity generation method based on variance and structural feature regulation. A structure-aware suppression factor is introduced into the weighted fusion of the mean and median, effectively suppressing the interference of residual fluctuations on the final output, thereby ensuring stable and reliable output light intensity values ​​that can be directly used for image reconstruction. Through the above series of steps, this invention achieves a systematic solution for frame synchronization and anomaly removal in single-pixel imaging without relying on external high-precision synchronization hardware, significantly improving the quality of sampled data and the accuracy of reconstructed images. It is suitable for widespread application in teaching experiments, portable imaging devices, and high-speed imaging scenarios. Attached Figure Description

[0020] 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.

[0021] Figure 1 This is a flowchart of a method for frame synchronization and light intensity sampling outlier removal based on a DLP projection system according to the present invention. Detailed Implementation

[0022] 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.

[0023] In one or more embodiments, such as Figure 1 As shown, a method for frame synchronization and light intensity sampling outlier removal based on a DLP projection system is disclosed, the method comprising the following:

[0024] S1: Obtain the original image to be projected and extract the structural feature vector, which includes the number of bright spots in the pattern and the average brightness of the pattern; based on the preset mapping relationship between structural features and light intensity response, calculate the structural predicted light intensity of the original image according to the structural feature vector;

[0025] Specifically, the purpose of this step is to extract physically meaningful and calculable structural features for each frame of the pattern to be projected in the DLP projection system, and to establish a quantitative relationship between these structural features and their actual sampled light intensity. This relationship will play a crucial supporting role in subsequent judgments of successful sampling and outlier removal. This modeling process relies solely on the pattern image and the existing light intensity sampling equipment in the system, without introducing other types of image data, historical information, or additional hardware.

[0026] This step uses a pattern image. As input, the image is a standard grayscale image sent from the host computer to the DLP controller via a USB interface. The image data is in 8-bit grayscale format and has a size of [missing information]. The pixel grayscale value ranges from [0, 255]. The image format can be parsed using common image processing tools (such as Python and OpenCV) and used as a standard two-dimensional matrix for structural feature extraction.

[0027] The analysis of pattern structure involves extracting two core features: the number of bright spots in the pattern and the average brightness of the pattern. (Based on the image...) As input, a grayscale thresholding operation is first performed, marking all locations in the image with grayscale values ​​greater than a set threshold (e.g., 128) as bright spots. This threshold is set based on optical response experiments of the DLP system and is the minimum driving grayscale value to ensure "projection micromirror flip is ON". The count of bright spots can be completed by scanning pixel by pixel, resulting in the total number of bright spots. , representing the number of areas in the pattern that participate in effective reflection.

[0028] After extracting the number of bright spots, the average grayscale value of all pixels in the pattern is calculated to obtain the pattern brightness level. This value represents the average level of the overall light energy distribution of the pattern and is the main basis for subsequent judgment on whether the reflected light intensity deviates from the structural expectation. This operation can be achieved by summing the pixels of the entire image and then dividing by the number of pixels, regardless of the image content distribution.

[0029] Next, to establish a quantitative correspondence between structural features and light intensity sampling values, it is necessary to collect actual sampling data of the sample patterns during the system initialization phase. Several representative patterns (e.g., 50 frames) are selected, loaded using a DLP, and then irradiated onto a standard reflective surface (e.g., a matte white board). The reflected light intensity of the corresponding pattern is collected using an APD (Avalanche Photodetector). Light intensity sampling is performed by a data acquisition card connected to the APD, such as the NI USB-6001. Each frame is sampled 5 times, and the average value is taken to obtain the actual light intensity of the pattern under standard illumination conditions. .

[0030] A sample set is constructed using the above samples. The prediction function is constructed using a weighted linear fit:

[0031] ;

[0032] in, and These are the maximum values ​​of the number of bright spots and the average brightness in the selected samples, respectively, used to normalize the variables to [0,1] to prevent differences in numerical magnitude from causing bias in the results. Current pattern The expected sampled light intensity is used as a standard value for subsequent frame synchronization and anomaly detection; The total number of pixels with a grayscale value greater than 128 in the pattern is obtained by the main control terminal after binarization using OpenCV. The average grayscale value of all pixels in the pattern is obtained by summing the number of pixels in the image and then dividing by the number of pixels. , The maximum number of bright spots and the maximum brightness value observed in the initial sample; , : Structural response weights, satisfying It is obtained by fitting with the minimum mean square error, for example. , .

[0033] Specifically, if a certain pattern With 24,000 bright spots, an average image brightness of 112, and a maximum number of bright spots of 40,000 and a maximum average brightness of 180 in historical samples, the normalized feature is: .like Then the predicted light intensity of the pattern is:

[0034] ;

[0035] This value is the pattern. Light intensity reference under standard reflection conditions.

[0036] S2: Collect at least one light intensity sample value within a fixed time window after the original image is projected to form a light intensity sequence; calculate the average light intensity of the light intensity sequence, and combine the predicted light intensity of the structure with the structural feature vector, introduce a structural response correction term and a structural adaptability tolerance term based on the spatial distribution location of the bright spot, and determine whether the sampling of the current frame is successfully synchronized with the pattern projection;

[0037] Specifically, this step aims to solve the alignment problem between sampling timing and pattern projection in a DLP single-pixel imaging system, i.e., the "frame synchronization" problem. Because the DLP projection module in this system uses digital pattern modulation, its inter-frame switching frequency is high. However, sampling devices (such as APDs and DAQ cards) have response delays or triggering errors in signal processing, easily leading to the phenomenon of "pattern switching but sampling still lagging," causing image distortion in the reconstructed image. Traditional methods often use fixed delays or discarding the first and last frames as simple synchronization means, but these ignore the differences in reflectivity and response of different patterns, resulting in unstable performance under experimental conditions with many pattern types and high sampling frequencies. Therefore, this step constructs a synchronization mechanism based on pattern structural features... and predicted light intensity value A structure-response-aware synchronous calibration mechanism is established based on this, and it is used to automatically determine whether the sampling has successfully aligned with the pattern, providing a controllable source of sampling data for subsequent outlier removal. This strategy minimizes the reliance on additional hardware trigger signals.

[0038] To ensure that the light intensity sampling is aligned with the pattern frame, this step involves sampling the light intensity in each pattern frame. After loading, set a sampling time window of a fixed length. Collect within this window Subsequent consecutive light intensity values ​​form a light intensity sampling sequence. . The value is determined through the initial system calibration, covering the complete exposure stage after the DLP pattern is stably output.

[0039] After sampling, calculate the average light intensity within the window:

[0040] ;

[0041] This average light intensity and the predicted value Deviation comparison is performed. However, since different patterns have different bright spot distributions, the actual reflection path may introduce specific deviations, so a simple fixed tolerance cannot be used for judgment. Therefore, we introduce a structural response correction term. This is used to model the non-ideal response of a pattern in the imaging path, such as energy loss due to large reflection angles when bright spots are concentrated in edge regions. The correction term is designed based on the spatial offset characteristics of the bright spot distribution (e.g., concentration at image edges).

[0042] ;

[0043] in, Pattern response correction term, used to adjust frame synchronization error judgment; The distance between the centroid of the bright spots in the pattern and the center of the image is calculated by weighting the centroid of each pixel. The diagonal length of the image in the system is used as a normalization reference. The corrected weighting coefficients are set through system experiments to reflect the degree of influence of edge effects on reflection loss.

[0044] Finally, the predicted light intensity, edge correction term, and structural adaptability tolerance term are integrated into the overall structure prediction. To determine whether the frame was successfully sampled, the following conditions are used:

[0045] ;

[0046] in, The structural adaptability tolerance term, calculated from the proportion of bright spots in the pattern, is defined in the same way as in step one.

[0047] If the determination is valid, the sampling is effective and the sampled data is retained; otherwise, the sampling of that frame is considered to be out of sync with the pattern, and the frame is invalid.

[0048] Specifically, in the pattern In the image, if the centroid of the bright spot is close to the edge, near ,but near ,like This indicates that the pattern may have an energy loss of approximately 5% under ideal conditions. If with The difference is no more than Even if the pattern is not fully synchronized, it can still be considered a successful synchronization. This mechanism improves the system's flexibility in sampling and judging complex patterns, and avoids edge distribution patterns being misjudged as faulty frames due to energy mismatch.

[0049] Unlike traditional methods that rely solely on light intensity error for judgment, this step introduces structural positional features (bright spot centroid distribution) as part of the synchronization criterion, giving the judgment standard "pattern perception" capabilities. Furthermore, correction terms... The design uses a proportional term rather than a constant offset, ensuring good scalability and applicability in different sized patterns and different scene devices.

[0050] S3: If the sampling synchronization of the current frame is determined to be successful, then based on the structural feature vector and the structural predicted light intensity, an outlier removal threshold that depends on both the pattern structure complexity and the overall sampling deviation is constructed, and outlier identification and removal are performed on the light intensity sequence, and the arithmetic mean of the remaining sampled values ​​after removal is calculated.

[0051] Specifically, this step aims to further process the sampled light intensity sequence that has been determined to be successfully synchronized. To improve the stability of light intensity input, while ensuring consistency between the pattern and the sampling time sequence, abruptly abnormal sampling points are removed. Unlike conventional image processing, in DLP single-pixel imaging systems, light intensity sampling is a single-channel time series without spatial dimension or redundant structure. Therefore, the validity of each data point directly determines the accuracy of image reconstruction, and outliers have a significant impact on the final image. The unique aspect of this step is that all sampling points cannot be judged based on spatial structural features; it is necessary to combine the structural information of the pattern itself. To determine whether the corresponding sampled values ​​are reasonable, it is necessary to construct an outlier removal model that combines structure awareness, local statistics, and structure bias correction.

[0052] To eliminate local abrupt changes or random errors in sampling, this step first uses the structural prediction values. and sampling mean The residuals between them are defined as a "structural stability index". To measure whether the current sampled sequence deviates overall:

[0053] ;

[0054] in: The average value of the current frame samples, derived from... The average of the sampling points is obtained; :pattern The predicted light intensity is obtained from the formula in step one; : A positive infinitesimal number, to avoid a denominator of 0, is usually set as the smallest quantization unit of the system; The larger the value, the greater the overall deviation, requiring a more stringent local elimination strategy.

[0055] Next, from This method identifies local anomalies. Unlike traditional methods based on range or fixed proportions, this approach introduces a structural bias regularization term, mapping the pattern's structural complexity (i.e., bright spot density) to a local fluctuation tolerance control factor, defined as:

[0056] ;

[0057] in: : Eliminate sensitivity thresholds to determine the scope of outlier definition; The influence coefficient of the bright spot structure is usually set to 0.05 to 0.1; The maximum number of bright spots observed in the entire system, used for normalization; Nonlinear structural response amplification factor, recommended value is 1.5 to 2.0; The residual guiding coefficient controls the impact of overall deviation on the rejection intensity; a value of 0.3–0.5 is recommended. (Second item) This is an innovative regularization term used to improve the ability to eliminate deviations in the overall structure.

[0058] Based on this sensitivity threshold, a local rejection rule is set: if the light intensity at a certain point... satisfy:

[0059] ;

[0060] The point is then identified as an outlier and removed.

[0061] The rejection threshold of this rejection mechanism is structure-driven, rather than a uniform constant, and can automatically adjust the tolerance range according to the density of bright spots in the pattern; by introducing The method feeds back the residual between the sampling mean and the structural model to the anomaly detection, realizing the "global residual control over local decision-making"; the nonlinear power function weights the bright spot density, making the removal threshold of the bright spot dense pattern show an increasing trend.

[0062] Specifically, in the pattern In the middle, if , , , ,but:

[0063] ;

[0064] If we assume , , ,but:

[0065] ;

[0066] That is, sampling points that deviate from the sampling mean by more than 7% will be marked as abnormal, avoiding the shortcomings of using a fixed ±5% threshold.

[0067] All removal operations can be completed in linear time, with low processing complexity, making them easy to run in real time in teaching systems or lightweight imaging platforms.

[0068] This step outputs a single variable. The final effective light intensity value after removing outliers is calculated as the arithmetic mean of the remaining points:

[0069] ;

[0070] in: The set of sample indices retained after removal; : Number of samples to retain, if If the frame fails to reconstruct, it is marked as a failed frame, and reconstruction is terminated.

[0071] S4: Based on the remaining sampled sequence after removing outliers, calculate its variance; weight the arithmetic mean and the median of the remaining sequence and fuse them, where the weighting weight is determined by the structural variance perception control factor, and finally generate a stable light intensity output value and submit it to the image reconstruction module.

[0072] Specifically, the goal of this step is to calculate the final output light intensity of the frame pattern based on the remaining valid sample values ​​after outlier removal in step three. This information is submitted to the image reconstruction module within the imaging system. The image reconstruction module relies on light intensity and the projection pattern. Since there is a one-to-one correspondence, this step not only needs to output the final value, but also needs to confirm that the value is reliable, i.e., unaffected by residual outliers and does not change excessively with the number of samples. To this end, a stability enhancement mechanism based on local variance suppression is designed. This mechanism enhances the stability of the output value through structural constraints, while ensuring that the processing method is simple, fast, and can be executed in real time, meeting the deployment requirements of teaching systems and experimental platforms.

[0073] Considering It is based on the arithmetic mean after removing outliers, if the number of remaining samples... Small sample sizes, or residual sample biases, can lead to unstable output. To enhance output stability, a structural repression factor is introduced. This is used to weight and suppress the offset of high-variance sampled outputs. It is defined as follows:

[0074] ;

[0075] in: The final stable light intensity output value used for image reconstruction; The mean value after removing outliers is the main source of output. The median of the removed sequence is used as a robust estimation reference. : Structure-variance perceived control factor, representing the degree of influence of current frame sampling fluctuations on the final output.

[0076] The expression is designed to weight and fuse the mean and median, thereby increasing the influence of the median and improving output stability when there are few sampling points and large variance. When there are many samples and the distribution is stable, the output is still dominated by the mean.

[0077] Regulatory factors The definition is as follows:

[0078] ;

[0079] in: : The sample variance of the sampled points after removal; The maximum sampling variance of the entire system recorded during the initialization phase; : Number of bright spots in the pattern; The maximum number of bright spots observed in the entire system; : Structural sensitivity coefficient, usually set to 0.5 to 0.8, is used to adjust the impact of structural complexity on output stabilization.

[0080] This definition models the coupling of current frame light intensity fluctuations with pattern complexity: for patterns with dense bright spots and a large reflection range (i.e., ... Larger, more tolerant of fluctuations; while for patterns with sparse bright spots ( If the median value is relatively small, then the stability requirement is increased, thereby increasing the median influence.

[0081] Specifically, if a certain pattern Samples after removal The mean is The median is Sample variance The system's maximum variance Number of highlights Maximum number of highlights ,but

[0082] First item: ;

[0083] Second item: ;

[0084] Take the minimum value: ;

[0085] The final output is:

[0086] ;

[0087] It is evident that this strategy effectively constrains the variance fluctuations caused by outliers within a reasonable range through structural control, resulting in more stable output.

[0088] This invention also provides a frame synchronization and light intensity sampling outlier removal device based on a DLP projection system, 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 frame synchronization and light intensity sampling outlier removal method for a DLP projection system, 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.

[0089] 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 frame synchronization and light intensity sampling outlier removal device based on a DLP projection system.

[0090] The frame synchronization and light intensity sampling outlier removal device based on a DLP projection system 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, buses, etc.

[0091] 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 frame synchronization and light intensity sampling outlier removal device based on a DLP projection system, connecting various parts of the device via various interfaces and lines.

[0092] The memory can be used to store the computer program and / or modules. The processor, by running or executing the computer program and / or modules stored in the memory, and by calling the data stored in the memory, realizes various functions of the frame synchronization and light intensity sampling outlier removal device based on a DLP projection system. 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 based on 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 devices.

[0093] The module integrating frame synchronization and light intensity sampling outlier removal equipment based on a DLP projection system, 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, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0094] 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.

[0095] 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 for frame synchronization and outlier rejection of light intensity sampling based on DLP projection system, characterized in that, The method comprises: obtaining an original image to be projected and extracting a structural feature vector, the structural feature vector comprising a pattern bright spot number and a pattern average brightness; calculating a structural predicted light intensity of the original image according to the structural feature vector based on a preset mapping relationship between a structural feature and a light intensity response; collecting at least one light intensity sampling value in a fixed time window after projection of the original image to form a light intensity sequence; calculating an average light intensity of the light intensity sequence, and combining the structural predicted light intensity and the structural feature vector, introducing a structural response correction term based on a spatial distribution position of a bright spot and a structural adaptability tolerance term to judge whether sampling of a current frame is successfully synchronized with pattern projection; if it is judged that sampling of the current frame is successfully synchronized, then based on the structural feature vector and the structural predicted light intensity, constructing an outlier rejection threshold value which depends on both pattern structural complexity and overall sampling deviation, performing outlier identification and rejection on the light intensity sequence, and calculating an arithmetic mean value of the remaining sampling values after rejection; based on the sampling sequence remaining after rejection of outliers, calculating a variance thereof; performing weighted fusion of the arithmetic mean value and a median value of the remaining sampling sequence, wherein a weighted weight is determined by a structural variance perception regulation factor, and finally generating a stable light intensity output value and submitting it to an image reconstruction module.

2. The frame synchronization and light intensity sampling outlier rejection method based on the DLP projection system according to claim 1, characterized in that, The extraction step of the pattern bright spot number comprises: performing binaryzation processing on a pattern image, and counting a number of pixel points with a gray value higher than a set threshold value.

3. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, characterized in that, The mapping relationship is established by weighted linear fitting through collection of structural features of at least one sample pattern and its actually measured average light intensity value under standard reflection conditions in a system initialization stage.

4. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, characterized in that, The structural response correction term is calculated based on a distance between a centroid of all bright spots in a current pattern and a center of an image.

5. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, characterized in that, The structural adaptability tolerance term is dynamically adjusted based on a ratio of a bright spot number of the current pattern to a maximum bright spot number observed by the system.

6. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, characterized in that, The construction of the outlier rejection threshold value introduces a structural bias regularization term, which is weighted by a pattern bright spot density through a nonlinear power function.

7. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, characterized in that, The construction of the outlier rejection threshold value also introduces an overall residual index, which is a relative deviation of the average value of the light intensity sequence from the structural predicted light intensity.

8. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, characterized in that, The value of the structural variance perception regulation factor takes a minimum value of the following two items: a first item is a ratio of a sampling variance of the current frame to a maximum variance of the system, and a second item is a product of a pattern structural sparsity and a structural sensitivity coefficient.

9. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, characterized in that, The condition for judging that sampling of the current frame is successfully synchronized is that an absolute value of an algebraic difference of the average light intensity, the structural predicted light intensity and the structural response correction term is less than or equal to the structural adaptability tolerance term.

10. The frame synchronization and light intensity sampling outlier rejection method based on DLP projection system according to claim 1, wherein, The light intensity sampling value is generated by an avalanche photodetector in response to pattern reflected light and is collected by a data acquisition card.

Images