A method for locating a moving target based on single-pixel imaging in a complex background

Abstract

This invention discloses a moving target localization method based on single-pixel imaging in complex backgrounds, comprising: a rapidly moving target entering the scene against a complex grayscale background; playing several binarized Fourier substrate patterns using a digital micromirror device; the light field of the target being rapidly modulated and received by a single-point silicon detector, then converted from analog to digital by a high-speed acquisition card and stored in a computer; then using the Fourier slicing theorem to convert some Fourier coefficients into a one-dimensional line integral signal projected along the x and y axes; comparing the difference between the one-dimensional line integral signals with and without the target in the time dimension; and using a threshold division method to determine the coordinates of the target center. This invention addresses the problem that the imaging frame rate in the single-pixel domain cannot meet the high resolution and real-time requirements of object tracking, effectively broadening the application scope of image-free tracking methods and promoting the application of single-pixel imaging in fields such as moving target detection.

Description

Technical Field

[0001] This invention pertains to the field of moving target tracking and localization technology, specifically a method for locating moving targets based on single-pixel imaging in complex backgrounds. Background Technology

[0002] Tracking and imaging of fast-moving targets holds significant promise for applications in navigation, biomedicine, and computer vision. Traditional target tracking methods based on planar array imaging primarily rely on photography, using image sensors to capture images of the target and further employing image post-processing and image analysis algorithms to determine the trajectory of the moving object within the image. Therefore, the accuracy of target tracking depends heavily on the quality of the captured sequence of images and the performance of the algorithm used. Some high-performance high-speed cameras can capture continuous images with high signal-to-noise ratios in short exposure times. However, in real-time tracking of moving objects, high-speed cameras typically have enormous data throughput, directly proportional to the camera's frame rate, leading to excessively high hardware requirements in practical applications. Furthermore, while high-speed cameras exhibit superior performance in the visible light region, their performance fails in some invisible light regions, such as the infrared and terahertz bands, thus limiting the application scope of planar array imaging technology in moving target tracking.

[0003] In recent years, with the rapid development of computational imaging technology, single-pixel imaging, as a novel computational imaging technique, has been widely studied. It modulates the light field of a target object sequentially using different illumination patterns, records one-dimensional light intensity signals using a single-pixel sensor without spatial resolution, and then reconstructs the image of the object based on the correlation between the illumination patterns and the single-pixel signals. Due to the advantages of single-pixel detectors, such as wide-spectrum imaging and low detector cost, single-pixel imaging exhibits superior performance in some invisible wavelengths. Therefore, when single-pixel imaging technology is used for target tracking, it can operate normally in wavelengths where area array cameras cannot function effectively. Currently, single-pixel imaging technology has been widely studied in various fields of moving target tracking and imaging.

[0004] Currently, there are two methods for obtaining the trajectory of moving objects using single-pixel imaging. The first method is similar to planar array imaging, analyzing the trajectory of the moving target from continuously reconstructed images. However, the frame rate of single-pixel imaging is limited by the modulation frequency of the spatial light modulator, and continuous measurement is required, resulting in a very large number of patterns used in the imaging process. To reduce the number of patterns required for imaging, some researchers have proposed using compressed sensing algorithms. However, due to the high computational cost of compressed sensing algorithms, the number of measurements is still large. To further improve the frame rate, some researchers have proposed a second method: directly acquiring the trajectory of the moving target without reconstructing the target image. The key to this method is to increase the frame rate by reducing the number of measurement patterns used. Zha et al. proposed a method for real-time tracking of moving objects at a frequency of 177Hz using 128 Hadamard patterns. Zhang et al. used 6 Fourier basis patterns to achieve real-time tracking of a specific type of moving object at 1666Hz in two-dimensional space. Subsequently, Zha et al. proposed a fast moving target tracking method based on geometric moment patterns, achieving a frame rate of 7.4kHz. Based on this, the team proposed a complementary measurement scheme, increasing the frame rate of the method to 11.1kHz. In the same year, Xiao et al. proposed an imaging method for randomly moving targets based on geometric moment analysis. This method was the first to reconstruct the shape and motion state of a target when the translational velocity, translational direction, rotation center, rotational velocity, and rotational direction were unknown. This method can realistically reconstruct a randomly moving object at a rotational speed of 1800 rpm.

[0005] However, the above methods are all for high-speed moving target localization under no background or pure black background, and their effectiveness in localizing moving targets in complex scenes remains unclear. In real-world scenarios, target localization often requires complex background environments and the absence of any prior knowledge such as movement speed or direction. Therefore, in the field of single-image imaging, methods for target localization under complex backgrounds urgently need further research. Summary of the Invention

[0006] The purpose of this invention is to provide a moving target localization method based on single-pixel imaging in complex backgrounds, so as to solve the problems of low frame rate and inaccurate localization of moving targets in complex background environments, and promote the application and development of single-pixel imaging technology in the field of moving target detection.

[0007] The technical solution to achieve the purpose of this invention is: a method for locating moving targets based on single-pixel imaging in complex backgrounds, the specific steps of which are as follows:

[0008] Step 1: Obtain partial Fourier coefficients of complex backgrounds using a passive single-pixel imaging system;

[0009] Step 2: Use a single-pixel imaging system to obtain partial Fourier coefficients of multiple frames of moving target images against a complex background;

[0010] Step 3: Calculate the line integral curve of the complex background image and the multi-frame motion image;

[0011] Step 4: Determine the target motion trajectory based on the line integral curve of the complex background image and multiple frames of motion images.

[0012] Preferably, the specific method for obtaining partial Fourier coefficients of complex backgrounds using a single-pixel imaging system is as follows:

[0013] Step 1.1: Generate several grayscale Fourier basis patterns with different spatial frequencies and different initial phases;

[0014] Step 1.2: Upsample the gray-level Fourier base pattern and use the Floyd-Steinberg error jitter algorithm to convert the upsampled gray-level Fourier base pattern into a binary pattern.

[0015] Step 1.3: Import the binary pattern into the DMD memory and set the modulation rate to modulate the scene;

[0016] Step 1.4: Use a detector to synchronously receive the modulation signal of the complex background reflected light and use a digital acquisition card to collect and save at least 3 light response values;

[0017] Step 1.5: Calculate the partial Fourier coefficients of the background target image using the three-step phase-shift formula based on at least three light response values;

[0018] Preferably, the grayscale Fourier substrate pattern is as follows:

[0019]

[0020] Where x and y are the spatial coordinates of the image, and u and v are the spatial frequencies f. x ,f y The discretized form of the image, with a reconstructed image size of M×N, and spatial frequency u having M discretized values: The spatial frequency v has N discretized values: a is the average intensity of the substrate pattern, and b is the modulation amplitude. This is the initial phase.

[0021] Preferably, the modulation signal of the complex background reflected light obtained by the single-pixel detector Specifically:

[0022]

[0023] Where S represents the region where the Fourier basis pattern modulation target is located. Let I(x,y) be a grayscale Fourier basis pattern, and let I(x,y) be the target scene image.

[0024] Light response value of a single pixel detector for:

[0025]

[0026] Among them, D n β is the light response value caused by background illumination at the position of a single pixel detector, and β is the amplification factor.

[0027] Preferably, the specific method for calculating the partial Fourier coefficients of the background target image using the three-step phase-shift formula based on at least three light response values ​​is as follows:

[0028] Initial phase of activity The photoresponse values ​​corresponding to 0, 2π / 3, and 4π / 3 can be D0(u,v), D... 2π / 3 (u,v),D 4π / 3 (u,v);

[0029] The Fourier coefficients F(u,v) corresponding to the spatial frequency points are calculated using a three-step phase-shift formula:

[0030]

[0031] By utilizing spectral symmetry, the corresponding spectrum is filled in to obtain the partial Fourier coefficients F of the background target image. ref (u,v).

[0032] Preferably, the specific method for obtaining partial Fourier coefficients of multi-frame moving target images under complex backgrounds using a single-pixel imaging system is as follows:

[0033] Step 2.1: Set the DMD image playback mode to loop K times;

[0034] Step 2.2: The target moves at a constant speed across the complex background on the electric guide rail. The detector synchronously receives the modulation signals of the reflected light from the moving target in multiple frames against the complex background, and the digital acquisition card collects and saves the light response value of the kth measurement.

[0035] Step 2.3: Calculate the partial Fourier coefficients F of the multi-frame moving target image. k (u,v).

[0036] Preferably, the specific method for calculating the line integral curves of complex background images and multi-frame motion images is as follows: Based on the Fourier slice theorem, calculate the line integral curves p of the targetless complex scene and the partial keyframe images of the target moving within the complex scene in the x and y directions, respectively. ref (x),pref (y),…,p k (x),p k (y), the specific formula is:

[0037]

[0038] Where θ is the angle between the x-axis containing the projection direction and the original x-axis, T(x,θ) is the projection integral curve of the image along the θ direction, and F -1 Inverse Fourier transform, p ref (x),p ref (y) represents the line integral curves of the background image in the x and y directions, respectively, p k (x) represents the integral curve of the moving target in the x-direction at the k-th measurement, p k (y) represents the integral curve of the moving target in the y-direction during the k-th measurement.

[0039] Preferably, the specific method for determining the target motion trajectory based on the line integral curve of a complex background image and multiple frames of motion images is as follows:

[0040] Calculate the x and y coordinates C of the target center. x (k),C y (k) curve:

[0041]

[0042]

[0043] In the formula, p ref (x),p ref (y) represents the line integral curves of the background image in the x and y directions, respectively, p k (x) represents the integral curve of the moving target in the x-direction at the k-th measurement, p k (y) represents the integral curve of the moving target in the y-direction during the k-th measurement.

[0044] Compared with the prior art, the present invention has the following significant advantages: (1) The present invention expands the application scenarios of the target localization method based on single-pixel imaging, and can accurately locate the target in a complex gray-scale background instead of a simple black and white background. The present invention is adaptable to complex scenes; (2) The present invention uses the Fourier slice theorem to convert the two-dimensional image frequency domain information into a one-dimensional line integral signal, which greatly reduces the signal quantity while reducing the information dimension, which is conducive to achieving high frame rate measurement and avoiding information waste; (3) In the target localization process, the present invention does not rely on any information about the target to assist in the target localization, such as prior knowledge of the moving speed, moving direction, etc., so it is adaptable to randomly moving targets.

[0045] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the overall process of a moving target localization method based on single-pixel imaging in complex backgrounds.

[0047] Figure 2 This is the optical path diagram of a single-pixel imaging system.

[0048] Figure 3 These are keyframe images from a test video that verifies the present invention.

[0049] Figure 4 This is a test video of keyframe images, line integrals, and a three-dimensional schematic diagram to verify the present invention.

[0050] Figure 5 This is a comparison diagram of the target positioning results and the actual motion results of this invention.

[0051] Figure 6 This invention is based on p under different sampling numbers k (x),p k (y) Comparison chart.

[0052] Figure 7 This is a comparison chart of the target localization results and actual results under different sampling point numbers according to the present invention.

[0053] Figure 8 These are keyframe images of the experimental scene used to verify the invention, captured by a high-speed camera.

[0054] Figure 9 This is a comparison chart of the three-dimensional line integral curve results of a horizontally moving target under different measurement numbers k when the total number of samples SP is 7 and the DMD modulation rate is 10kHz and 20kHz.

[0055] Figure 10 This is a comparison chart of the positioning results of a horizontally moving target and the camera capture results under different measurement counts k when the total sampling number SP is 7 and the DMD modulation rate is 10kHz and 20kHz. Detailed Implementation

[0056] The present invention will now be described in further detail with reference to the accompanying drawings.

[0057] This invention is a moving target localization method based on single-pixel imaging in complex backgrounds, utilizing a single-pixel system to measure scene information of the moving target. A single-pixel system is an optical system in which the reflected light from a target illuminated by a light source is modulated by a spatial light modulator and received by a single-point detector. The overall flowchart of this invention is shown below. Figure 1As shown, the main experimental components include an LED light source, a digital micromirror device (DMD), a silicon detector, a target, a motorized guide rail, a high-speed camera, a data acquisition card, and a computer. The actual system optical path diagram is shown below. Figure 2 As shown, the target moves in a straight line across the scene on an electric guide rail with a speed of 70 mm / s and an effective travel of 100 mm. The scene is illuminated by an LED light source, and the reflected light is imaged onto the DMD target surface through an imaging lens. After being modulated by a Fourier substrate pattern, the light is received by a silicon detector through a collecting lens and finally acquired and stored in computer memory by a digital acquisition card (NI9222). The Fourier slice theorem is used to convert some of the acquired Fourier coefficients into one-dimensional line integral signals in the x and y directions. The one-dimensional line integral signal with the moving target is compared with the background line integral signal by differential comparison. An appropriate threshold is set to identify areas with large signal fluctuations, thereby locating the center of the target. Under the same field of view, a high-speed industrial camera acquires the image information of the moving target passing through the scene completely at 1000 fps, saving it in both video and image formats to computer memory. Since the number of patterns played is as low as 7*3, combined with the high modulation rate of the DMD (up to 22.727 kHz), this invention can achieve moving target localization at approximately 1000 fps.

[0058] A method for localizing moving targets based on single-pixel imaging in complex backgrounds, the specific steps of which are as follows:

[0059] Step 1: Obtain partial Fourier coefficients F of complex backgrounds using a single-pixel imaging system. ref (u,v);

[0060] Step 1.1: The computer generates several grayscale Fourier basis patterns with different spatial frequencies and different initial phases.

[0061] The gray-level Fourier basis pattern generation method is shown in Equation 1. Where x and y are the spatial coordinates of the image, and u and v are the spatial frequencies f. x ,f y The discretized form, assuming the size of the reconstructed image is M×N, and the spatial frequency u has M discretized values: The spatial frequency v has N discretized values: However, when u and v take values, one of them must be 0. Let a be the initial phase, a be the average intensity of the base pattern, and b be the modulation amplitude. Their values ​​are: a = b = 0.5.

[0062]

[0063] Step 1.2: Upsample the gray-level Fourier base pattern and use the Floyd-Steinberg error jitter algorithm to convert the upsampled gray-level Fourier base pattern into a binary pattern.

[0064] Step 1.3: Import the binary pattern into the DMD memory and set the DMD modulation rate to modulate the scene;

[0065] Step 1.4: The detector synchronously receives the modulation signal of the complex background reflected light and the digital acquisition card collects and saves three photoresponse values.

[0066] Complex backgrounds refer to spatial scenes with continuously varying gray levels; most natural scenes exhibit this characteristic. Assuming the target in a complex background is a reflective object, its light intensity distribution, i.e., the target scene image, is denoted as I(x,y). The reflected light from the target object forms a Fourier basis pattern. Under the modulation, Target reflected light intensity obtained by a single-pixel detector Formula 2 can be used for calculation:

[0067]

[0068] Where S represents the region where the Fourier basis pattern modulation target is located. for:

[0069]

[0070] Among them, D n β is the light response value caused by background illumination at the position of a single pixel detector, and β is the amplification factor, which is related to the detector itself and the spatial relationship of the object.

[0071] For different spatial frequencies (u, v), the single-pixel detector will obtain the corresponding photoresponse value for every 3 frames of modulation pattern played by the DMD. Among them, the initial phase The values ​​are 0, 2π / 3, and 4π / 3 respectively, therefore the photoresponse values ​​can be denoted as D0(u,v), D... 2π / 3 (u,v),D 4π / 3 (u,v).

[0072] Step 1.5: Calculate the partial Fourier coefficients F of the background target image. ref (u,v);

[0073] Finally, the corresponding Fourier coefficients F(u,v) can be calculated using the three-step phase-shift formula:

[0074]

[0075] The Fourier coefficients corresponding to the above spatial frequency points are calculated using Formula 4, and the corresponding spectra are then filled in using spectral symmetry, denoted as F. ref (u,v).

[0076] Step 2: Use a single-pixel imaging system to obtain partial Fourier coefficients F of multiple frames of moving target images against a complex background. k (u,v);

[0077] Step 2.1: Set the DMD image playback mode to loop K times;

[0078] Step 2.2: The detector synchronously receives the modulation signals of multiple frames of light reflected from moving targets against a complex background, and the digital acquisition card acquires and saves the light response value of the kth measurement.

[0079] An electrically driven guide rail is placed directly in front of a complex background. The target moves at a constant speed across the complex background on the guide rail. Since the DMD loops through several measurement patterns K times, the single-pixel detector performs multiple repeated measurements of the scene containing the target. The light response value obtained from the k-th measurement is denoted as...

[0080] Step 2.3: Calculate the partial Fourier coefficients F of the multi-frame moving target image. k (u,v);

[0081] Since the target is measured K times during its movement in the scene, the Fourier coefficient F of the moving target portion in the kth measurement can be calculated according to Formula 4. k (u,v). Therefore, we can finally obtain the partial Fourier spectra of K+1 images. The partial Fourier coefficients of each frame are derived from SP. X +SP Y -1 Fourier basis patterns with different spatial frequencies were obtained through three-step phase shift measurements. Among them, SP... X SP Y These represent the number of sampling points in the x and y directions, respectively.

[0082] To facilitate the demonstration of the positioning process results, the measured data are processed at fixed intervals of measurement number T (unit: times), thereby calculating the spectrum of K / T+1 images. Key frames from the test video used to verify this invention are shown below. Figure 3 As shown, the frame interval is 25 frames, from Figure 3 As can be seen, the white puppy target moves from left to right through a complex grayscale scene.

[0083] Step 3: Calculate the line integral curves of complex background images and multi-frame motion images;

[0084] According to the Fourier slice theorem, the line integral curves p of some keyframe images in the test video in the x and y directions can be calculated. ref (x,p ref (y,…,p k (x,p k (y, as shown in Formula 5:

[0085]

[0086] Where θ is the angle between the x-axis containing the projection direction and the original x-axis, T(x,θ) is the projection integral curve of the image along the θ direction, and F -1 Inverse Fourier transform, p ref (x),p ref (y) represents the line integral curves of the background image in the test video in the x and y directions, respectively. k (x) represents the integral curve of the moving target in the x-direction at the k-th measurement, p k (y) represents the integral curve of the moving target in the y-direction during the k-th measurement. The result is as follows: Figure 4 As shown, Figure 4 (a) and (b) represent the number of sampling points SP in the x and y directions, respectively. X SP Y The line integral curves calculated at 129 are shown in Figures (c) and (d). It can be seen that the line integral in the x-direction fluctuates drastically, while the line integral in the y-direction only shows significant local fluctuations. k (x),p k (y) A schematic diagram of the three-dimensional curve as the number of measurements k increases. It can be seen that as k increases, p k The peak point of the curve (x) will shift towards the direction with larger x-coordinate values, while p k The fluctuation of (y) is not obvious, and it only shows some fluctuation around the y coordinate value of 200.

[0087] Step 4: Calculate the target's trajectory;

[0088] Based on formulas 6 and 7, set appropriate threshold coefficients λ and μ, and calculate the horizontal and vertical coordinates C of the target center. x (k),C y (k) curve:

[0089]

[0090]

[0091] The trajectory of the target center's movement was plotted based on the curves showing the change in the target center's horizontal and vertical coordinates with the number of measurements. The λ and μ values ​​in the test video were both set to 0.9. The positioning results of this invention when the total sampling points SP were 257 are as follows: Figure 5 As shown, the target localization trajectory closely matches the actual trajectory, indicating that the localization effect of this method is significant. To further improve the localization frame rate, we first consider reducing the number of sampling points. When the number of sampling points in the x and y directions is SP... X SP Y As the curve gradually decreases, the integral curve of the target scene is as follows: Figure 6 As shown in the figure, it can be seen that as the number of sampling points gradually decreases, the line integral curve gradually becomes smoother, and the originally clear curve fluctuations (as shown in the red and green boxes in the figure) gradually disappear, making the difference imperceptible to the human eye. Therefore, we tested the positioning effect of the method of this invention under different numbers of sampling points, and the results are as follows. Figure 7 As shown, the total number of sampling points SP = SP X +SP Y -1, from Figure 7 As shown in (c), when SP is 3, the localization result shows a significant deviation, while the results for other sampling point numbers are closer to the actual results. Therefore, a minimum of 7 sampling points should be used to achieve high frame rate target localization.

[0092] Keyframes of the target scene recorded by a high-speed camera in the experiment, such as Figure 8 As shown, the frame interval is 15 frames, indicating that the target moves from left to right across the complex background. In the experiment, the sampling point SP was set to 7, and the DMD looped through K+1 times a 7*3 Fourier basis pattern. The result obtained from the scene modulation signal obtained using the single-pixel system after the above calculations is shown below. Figure 9 As shown, this is a comparison of the three-dimensional line integral curve results of a horizontally moving target under different measurement counts k when the total number of samples SP is 7 and the DMD modulation rate is 10kHz and 20kHz. In order to ensure that the target motion process is completely recorded, when the DMD modulation rate is 10kHz and 20kHz, we set the total number of measurements K to 600 and 800 times respectively, and the threshold coefficients λ and μ are both set to 0.9. Figure 10 This is a comparison chart of the localization results of a horizontally moving target and the camera capture results under different measurement counts k when the total sampling number SP is 7 and the DMD modulation rate is 10kHz and 20kHz. Figure 10 As can be seen from the trajectory diagram in (e), the target localization result is still relatively accurate even when the DMD modulation rate is as high as 20kHz, thus proving that the present invention can achieve moving target localization at about 1000fps.

Claims

1. A method for locating moving targets based on single-pixel imaging in complex backgrounds, characterized in that, The specific steps are as follows: Step 1: Obtain partial Fourier coefficients of complex backgrounds using a passive single-pixel imaging system; Step 2: Use a single-pixel imaging system to obtain partial Fourier coefficients of multiple frames of moving target images against a complex background; Step 3: Calculate the line integral curves of the complex background image and the multi-frame motion images. Specifically, based on the Fourier slice theorem, calculate the line integral curves of the complex scene without a target and the keyframe images of the target moving within the complex scene. The integral curve along the direction is given by the following formula: in, The projection direction is located Axis and original The included angle of the axis, For the image along Projected integral curve in the direction, Inverse Fourier Transform These represent the background image in Line integral curve in the direction, Indicates the moving target in the 1st... The measurement Direction line integral curve, Indicates the moving target in the 1st... The measurement Directional integral curve; Spatial frequency Discretized form; Step 4: Determine the target's motion trajectory based on the line integral curve of the complex background image and multiple frames of motion images. The specific method is as follows: Calculate the x and y coordinates of the target center curve: 。 2. The method for locating moving targets in complex backgrounds based on single-pixel imaging according to claim 1, characterized in that, The specific method for obtaining partial Fourier coefficients of complex backgrounds using a single-pixel imaging system is as follows: Step 1.1: Generate several grayscale Fourier basis patterns with different spatial frequencies and different initial phases; Step 1.2: Upsample the gray-level Fourier base pattern and use the Floyd-Steinberg error jitter algorithm to convert the upsampled gray-level Fourier base pattern into a binary pattern. Step 1.3: Import the binary pattern into the DMD memory and set the modulation rate to modulate the scene; Step 1.4: Use a detector to synchronously receive the modulation signal of the complex background reflected light and use a digital acquisition card to collect and save at least 3 light response values; Step 1.5: Calculate the partial Fourier coefficients of the background target image using the three-step phase-shift formula based on at least three light response values.

3. The method for locating moving targets in complex backgrounds based on single-pixel imaging according to claim 2, characterized in that, The grayscale Fourier base pattern is as follows: in, These are the spatial coordinates of the image. Spatial frequency The discretized form, the size of the reconstructed image is spatial frequency for Discretized values: spatial frequency for Discretized values: , It is the average intensity of the base pattern. It is the modulation amplitude. This is the initial phase.

4. The method for locating moving targets based on single-pixel imaging in complex backgrounds according to claim 2, characterized in that, Modulation signal of complex background reflected light obtained by single-pixel detector Specifically: in, The region where the target is located is modulated by the Fourier basis pattern. It is a grayscale Fourier base pattern. For the target scene image; Light response value of a single pixel detector for: in, This represents the light response value caused by background illumination at the location of a single pixel detector. This is the magnification factor.

5. The method for locating moving targets based on single-pixel imaging in complex backgrounds according to claim 2, characterized in that, The specific method for calculating the partial Fourier coefficients of the background target image using the three-step phase-shift formula based on at least three light response values ​​is as follows: Initial phase of activity They are respectively The corresponding photoresponse values ​​can be respectively ; The Fourier coefficients corresponding to spatial frequency points are calculated using a three-step phase-shift formula. : By utilizing spectral symmetry, the corresponding spectrum is filled in to obtain partial Fourier coefficients of the background target image. .

6. The method for locating moving targets based on single-pixel imaging in complex backgrounds according to claim 1, characterized in that, The specific method for obtaining partial Fourier coefficients of multi-frame moving target images against a complex background using a single-pixel imaging system is as follows: Step 2.1: Set the DMD image playback mode to loop playback. Second-rate; Step 2.2: The target moves at a constant speed across the complex background on the electric guide rail. The detector synchronously receives the modulated signals of the reflected light from the moving target in multiple frames against the complex background, and the digital acquisition card collects and saves them. The light response value measured in this study , ; Step 2.3: Calculate the partial Fourier coefficients of the multi-frame moving target images. .

Images