29 results about "Ghost imaging" patented technology

The invention discloses a ghost imaging image reconstruction method and device and a storage medium, and the method comprises the steps: firstly obtaining a low-quality reconstructed image generated by a ghost imaging system, and then inputting the low-quality reconstructed image into a trained LD-GIRNet network for processing, and obtaining a high-quality target reconstructed image. The network is of an encoder-decoder architecture, and the core of the network is that a plurality of lightweight decoding reconstruction units are embedded in an encoder and a decoder. The LDRU integrates a channel-space attention joint enhancement module or a detail enhancement multi-scale residual module to enhance features and suppress noise. Meanwhile, the network is connected with a decoder and an encoder through a residual compensation fusion module, and shallow detail features and deep semantic features are effectively fused. According to the method, the lightweight design of the model is realized, the parameter quantity is remarkably reduced, the image quality, noise robustness and detail recovery capability of ghost imaging reconstruction at a low sampling rate are effectively improved, and the method has excellent generalization performance.

This invention discloses a method, device, and storage medium for ghost imaging image reconstruction. The method first acquires a low-quality reconstructed image generated by a ghost imaging system, and then inputs it into a trained LD-GIRNet network for processing to obtain a high-quality target reconstruction image. The network employs an encoder-decoder architecture, with multiple lightweight decoding and reconstruction units embedded in both the encoder and decoder. LDRU integrates a channel-spatial attention joint enhancement module or a detail enhancement multi-scale residual module to enhance features and suppress noise. Simultaneously, the network connects the decoder and encoder through a residual compensation fusion module, effectively fusing shallow detail features with deep semantic features. This invention achieves a lightweight model design, significantly reducing the number of parameters while effectively improving the image quality, noise robustness, and detail recovery capability of ghost imaging reconstruction at low sampling rates, exhibiting excellent generalization performance.

A scan circuit applicable to driving a plurality of light-emitting-diodes of a plurality of scan lines of a display panel is provided. A plurality of switching components are configured, having each switching component connected with one of the scan lines. A plurality of power transistors are disposed corresponding to the scan lines, having each power transistor connected between one scan line and a ground terminal. A low-dropout regulator circuit or a global operational amplifier is alternatively connected to the plurality of switching components in common for selectively generating an output voltage to at least one of the scan lines and avoiding ghost imaging on the display panel. By employing the proposed scan circuit, the invention is effective in achieving de-ghost functioning and avoiding conventional signal coupling issues on the display panel. In addition, redundant power waste consumption can be significantly reduced at the same time.

The application discloses a kind of ghost imaging reconstruction and classification method and equipment based on hybrid network, belong to the field of computational imaging technology, will be based on Hadamard matrix generation light speckle sequence formed by fixed light speckle sequence;Then, based on this light speckle sequence sampling expanded MNIST data set, generate light intensity sequence, light speckle sequence data;Then, construct and train GRU-CNN network model;Finally, using the model trained to calculate correlation imaging reconstruction and target identification classification.The conventional ghost imaging technique is often faced with the problem of poor image quality under low sampling conditions.In order to solve this problem, the application proposes a GRU-CNN network (GCGI) model combined with deep learning, which can efficiently reconstruct high-quality images and accurately identify and classify targets under very low sampling rates by fusing light intensity sequence and light speckle information, and has strong practical value.

The invention provides a video image defogging and low illumination enhancement method and system, and relates to the technical field of video processing, and the method comprises the steps: synchronously obtaining a video frame sequence, a reference frame and a light intensity correlation value of a scene, carrying out the ghost imaging reconstruction through employing a Fourier lamination imaging method, and carrying out the ghost imaging reconstruction; fusing the reconstructed image with the fog concentration gradient, the illumination attenuation coefficient and the turbulence disturbance intensity extracted from the video frame sequence to generate a fused feature map; then, a pulse coupling neural network is adopted to divide fog-effect, low-illumination and turbulence regions, and a capsule network is utilized to perform dynamic routing and feature reconstruction on each region to obtain a feature vector set; and finally, performing adaptive restoration on each region in combination with the features of the reference frame to generate a clear image. According to the invention, the image quality and stability of the digital twin system in complex outdoor scenes such as foggy days, low illumination and turbulence can be effectively improved.

The application discloses a two-stage modulation-based underwater polarization ghost imaging method and device, and relates to the fields of optical imaging and ocean detection. The method comprises the following steps: acquiring basic data of a target water area and constructing a backscattering point spread function; performing statistical decorrelation modulation on an original reference speckle sequence based on a first correlation coefficient, a second correlation coefficient and the backscattering point spread function, to obtain a primary modulation reference speckle distribution sequence; performing polarization space modulation on the primary modulation speckle sequence, to obtain a secondary modulation reference speckle distribution sequence; irradiating a target to be detected according to the secondary modulation reference speckle distribution sequence, acquiring a double-polarization bucket detection signal, and constructing a polarization difference signal sequence; and performing correlation reconstruction on the polarization difference signal sequence and the secondary modulation reference speckle distribution sequence, to obtain a reconstructed image. The application solves the problems of serious backscattering interference and insufficient imaging signal-to-noise ratio in a strong turbid underwater environment, and realizes high-quality underwater polarization ghost imaging.

This invention discloses a noise-free ghost imaging method and system, relating to the field of photoelectric imaging technology. The method includes generating a two-dimensional matrix of random numbers, binaryizing the two-dimensional matrix according to grayscale information to obtain a binary two-dimensional matrix, dividing the two-dimensional matrix into P groups of two-dimensional binary matrices, using the P groups of two-dimensional binary matrices as projection matrices, projecting them onto a target object X, and using a bucket detector composed of an array detector and an FPGA board to acquire the reflected light intensity values ​​of the P groups of target objects as collected data, simultaneously performing binarization processing to obtain the P groups of bucket measurement signals; calculating the target object matrix based on the P groups of two-dimensional binary matrices and the P groups of bucket measurement signals, and reconstructing the target object image. This invention utilizes the binary nature of binary data to achieve noise-free imaging, overcoming the problem of existing methods where unavoidable noise prevents the reconstruction of high-quality images.

This invention discloses a computational ghost imaging method based on conditional generative adversarial networks (GANs) for low sampling rates. This method can extract internal features from images in harsh environments or at low sampling rates. By further improving and optimizing traditional autoencoder algorithms, it effectively solves the problems of image quality and speed in reconstructed images under low sampling conditions. Compared with traditional CGI, CSCGI, CNN-CGI, DCAN-CGI, and DGI methods, this invention can reconstruct target images faster at low sampling rates, and significantly improves both PSNR and SSIM, effectively filtering out background speckle. Practical physical experiments further verify the feasibility of this method. While ensuring the quality of the reconstructed image, the high efficiency of GPU multi-core parallel computing can significantly shorten both the training time of the neural network and the target object reconstruction time, which is of great significance for engineering applications.

The invention relates to a ghost imaging neural differential analysis method under a selected plaintext attack condition in the technical field of computational imaging and optical security. The method comprises the following steps: firstly, designing a differentiated plaintext, and deducing a projection speckle by adopting differential calculation; then, constructing a dimensionality reduction simulation training set and an experimental test set based on the structural features of the projection speckles; then, constructing a one-dimensional signal noise reduction neural network model, and training the one-dimensional signal noise reduction neural network model by adopting the dimension reduction simulation training set; and finally, inputting the experimental test set into the trained one-dimensional signal noise reduction neural network model to obtain deciphered speckles, and based on the deciphered speckles, adopting ghost imaging correlation operation to obtain a cracked plaintext image. That is to say, based on the basic principle of ghost imaging encryption, the inherent linear property of the ghost imaging light path is utilized, and the fusion architecture of the differential attack and the neural network is used, so that the cracking of various ghost imaging encryption methods can be realized, and the universality and reliability of ghost imaging encryption analysis can be improved.

The present application provides a kind of ghost imaging monitoring method and system based on environmental coupling simulation and image enhancement, the method comprises: setting initial light source, the initial light source is handled with splitter matrix model to obtain probe light path light source and reference light path light source, based on the probe light path light source and reference light path light source, respectively construct probe light path and reference light path;In probe light path, lens, object to be measured and bucket detector are set in sequence, and surface detector is set in reference light path, high gauss blur is introduced in probe light path to simulate heavy fog environment, and random phase screen based on Perlin noise is introduced to simulate turbulent environment.The present application builds ghost imaging system through Python, uses Gaussian simulation to simulate heavy fog environment and turbulent environment, and combines the concept of histogram equalization in digital image processing, the imaging quality of ghost imaging is optimized by non-linear mapping gray scale chart, and the imaging quality of ghost imaging system under long-distance condition can be improved.

The invention discloses a calculation ghost imaging reconstruction method based on a Vision Transform and a U-Net network, and the method comprises the following steps: 1, enabling a Hadamard matrix of a cut cake sequence to serve as a phase mask, sequentially embedding the Hadamard matrix into a spatial light modulator, sequentially projecting the Hadamard matrix to a target image, and collecting a total light intensity sequence through a barrel detector; 2, partitioning the Hadamard matrix of the cake cutting sequence and the light intensity sequence, performing linear projection on each block to obtain a high-dimensional vector token, adding a position code to each high-dimensional vector token, inputting the high-dimensional vector token into a Transform encoder for processing, and converting the output into a two-dimensional matrix; and step 3, inputting the two-dimensional matrix obtained in the step 2 into the U-Net network to enhance transmission and fusion of feature information until a loss function of the ViT-U-Net network is kept stable, and obtaining a result of calculating ghost imaging. The image reconstructed by the method is high in quality and high in reconstruction efficiency.

A high signal-to-noise ratio ghost imaging method based on orthogonal modulation of light spot of light source, a special light spot pattern is used to irradiate a sample to obtain a high signal-to-noise ratio image. The special light spot pattern makes the functions of the intensity of the light source irradiated on different positions of the sample changing with time orthogonal to each other, for example, the light intensity fluctuations at different positions have different frequencies. By detecting the light signal intensity fluctuations through the object and analyzing the frequency components of the signal, the shape of the object can be restored. The signal-to-noise ratio of the image is greatly improved. Correspondingly, the number of light spot patterns required for clear imaging is greatly reduced, thereby improving the imaging speed.

The application discloses a kind of turbulent flame geometry three-dimensional tomography method and system based on multi-probe CGI, comprising: N single-pixel detector is arranged around flame to form multi-probe detection array, dynamic space illumination modulator is used to generate structured light field to illuminate flame, each detector synchronously acquires barrel detection signal;Through compressed sensing reconstruction algorithm combined with convex optimization solving method, flame two-dimensional projection image under each view angle is recovered;Using adaptive histogram equalization, combined with filtering method, the calculation ghost imaging noise is inhibited, and two-dimensional projection image preprocessing is realized;Based on multi-view geometric projection relationship, construct tomographic projection matrix, use algebraic reconstruction technique combined with full variation regularization iterative algorithm to solve three-dimensional voxel distribution, and the flame front geometry three-dimensional structure is reconstructed.The application uses single-pixel detector to replace traditional area array camera, with the advantages of low cost, strong ability to resist harsh environment, high time resolution, suitable for high space-time resolution three-dimensional measurement.

The invention discloses a rapid micro-nano grating defect detection method and system based on Fourier space imaging, and relates to the technical field of ghost imaging systems.The method is characterized by comprising the steps that an ultraviolet spectrum image and a visible light spectrum image of a to-be-detected micro-nano grating are obtained; obtaining the defect type of the micro-nano grating according to the ultraviolet spectrum image, the visible light spectrum image and the dual-channel convolutional neural network; according to the ultraviolet spectrum image and two-dimensional short-time Fourier transform, determining the position of a non-periodic defect; bayesian reasoning is carried out according to the ultraviolet spectrum image and the visible light spectrum image, the position of the mixed defect is determined, the global spectrum is decomposed into the local spectrum through two-dimensional short-time Fourier transform, the defect area is obtained through the local outlier factor, accurate mapping of frequency anomaly and the spatial position is achieved, and the accuracy of the mixed defect is improved. The problem of position information loss caused by traditional Fourier transform is solved, and non-periodic defects and mixed defects can be effectively detected.

The application discloses a signal reconstruction method in a ghost imaging system based on compressed sensing, and comprises the following steps: S1, performing block processing on an original signal to obtain a pretreatment signal; S2, obtaining a sparse representation of the pretreatment signal and a measurement signal; S3, performing inner product of each atom in a sensing matrix and a current residual, and using L*t atoms with the maximum inner product to form a matrix A t ; S4, using a least square method to obtain a current sparse signal; S5, selecting t items with the maximum absolute value from the current sparse signal, and selecting corresponding atoms in the sensing matrix to form a new matrix A tL ; S6, calculating a residual of the measurement signal based on A tL ; if the residual is less than or equal to a threshold value, outputting the current sparse signal; if the residual is greater than the threshold value and greater than a residual of a last round of iteration, setting t=t+1 and jumping to S3; otherwise, directly jumping to S3; and S7, obtaining a reconstruction signal of the original signal based on the current sparse signal. In this way, the requirement of hardware in the sampling process can be reduced, and the signal recovery effect can be improved.

The invention belongs to the technical field of correlated imaging, and provides a non-equidistant ghost imaging method and system based on a Talbot self-imaging effect, and the method comprises the steps: building a light path, enabling emergent light of a laser to load a hologram containing a specific randomly distributed Bessel mode through a spatial light modulator, and generating a partially coherent light beam with a periodic self-imaging characteristic; the partially coherent light beam is split into a signal light path and a reference light path through a beam splitter, the signal light path irradiates an object to be measured after passing through a first transmission distance, and the light intensity is received by a barrel detector D1; after the reference light path passes through a second transmission distance, a detector D2 records light intensity distribution; the distance difference between the first transmission distance and the second transmission distance is set to be an integral multiple of the Talbert distance; synchronously acquiring light intensity received by the multiple groups of bucket detectors D1 and light intensity distribution recorded by the detectors D2; and performing second-order statistical correlation calculation on the multiple groups of light intensity data, and reconstructing an image of the object to be measured. The problem that the ghost imaging technology seriously depends on light path equidistance is solved.

This application discloses an underwater polarization ghost imaging method and apparatus based on two-level modulation, relating to the fields of optical imaging and marine exploration. The method includes acquiring basic data of the target water area and constructing a backscattering point spread function; statistically decorrelating the original reference speckle sequence based on first and second correlation coefficients and the backscattering point spread function to obtain a first-level modulated reference speckle distribution sequence; performing polarization spatial modulation on the first-level modulated speckle sequence to obtain a second-level modulated reference speckle distribution sequence; illuminating the target under test according to the second-level modulated reference speckle distribution sequence to acquire a dual-polarization barrel detection signal and construct a polarization difference signal sequence; and reconstructing the image by correlating the polarization difference signal sequence with the second-level modulated reference speckle distribution sequence. This application solves the problems of severe backscattering interference and insufficient imaging signal-to-noise ratio in highly turbid underwater environments, achieving high-quality underwater polarization ghost imaging.

The present application relates to a kind of underwater long-distance ghost imaging methods for modulating compensation speckle light field phase distortion in real time.The light emitted by light source is intensity-modulated by DMD to produce speckle light field, and then phase-modulated by SLM, and divided into two ways after passing through beam splitter, one way directly enters camera, and records real-time speckle light field information;The other way is imaged to target after passing through underwater isotropic medium.The scattered light field returns to beam splitter after being reflected by target, and is received by the camera on the other side.By analyzing and comparing the intensity information of speckle light field received by two cameras, and inverting phase information, phase distortion is fed back to spatial light modulator, and each micromirror pixel unit of digital micromirror device for modulating speckle light field is compensated in real time.After that, repeat the above imaging process, calculate the sum of each pixel of return speckle light field received by camera (8), and correlation calculation is carried out with modulated speckle light field, and finally the image of target is obtained.The method is a new way to correct speckle light field phase by real-time acquisition and calculation of light field distortion phase, and realizes long-distance ghost imaging in underwater isotropic environment.

The application discloses a kind of based on time-domain ghost imaging mid-infrared data encryption optical communication system and method, the system includes: continuous light laser, for producing tunable near-infrared continuous light;Signal encryption module, for generating target data signal and generating key pair target data signal encryption, obtain encrypted signal;Near-infrared high-speed light modulation module, for responding to encrypted signal, tunable near-infrared continuous light is time-domain high-speed modulation, generates near-infrared modulation light signal;Dispersion optical fiber module, for introducing dispersion compensation optical fiber to near-infrared modulation light signal time-domain stretch, after physical encryption, send to nonlinear frequency down-conversion module;Near-infrared single-frequency laser, for outputting near-infrared single-frequency continuous light;Nonlinear frequency down-conversion module, for near-infrared modulation light signal and near-infrared single-frequency continuous light beam combination, through frequency down-conversion generation carries encrypted signal's mid-infrared light signal;Mid-infrared low-speed light detection module, for detecting mid-infrared light signal and recording signal integral intensity;Time-domain ghost imaging data decryption module, for signal integral intensity and key cross-correlation, decryption obtains target data signal, realizes mid-infrared waveband data encryption transmission.The application solves the problem of mid-infrared detector bandwidth limitation by combining frequency down-conversion technology and computing time-domain ghost imaging, and improves encryption rate.

The application discloses a phase time-domain ghost imaging method. The device uses light pulses generated by a non-coherent light source in a time domain as an imaging light source, divides the light pulses into two paths, and one path is used for acting on a to-be-measured time signal and is received by a photodiode. The time-domain signal is converted into a frequency-domain signal on the photodiode, a filter function is used to filter out a low-frequency part and reserve a high-frequency part, and the high-frequency part is converted into a time-domain signal to record light intensity information. The other path is used for recording a signal intensity waveform by another photodiode, and second-order correlation operation is performed on light intensities of the two photodiodes to obtain an image of phase modulation. The device is a phase time-domain ghost imaging device. The method and the device are simple, easy to modulate, and have application value in the field of optical imaging.

The invention discloses an underwater polarization ghost imaging method and device based on linear and circular polarization composite modulation, and relates to the technical field of optical imaging and ocean exploration, and the method comprises the steps: collecting the multi-polarization-state intensity distribution of a reflected laser signal of a target region and a barrel detection light intensity value, and calculating a Stokes vector based on the intensity distribution; calculating a polarization ghost image based on the barrel detection light intensity value, and calculating a stabilization correction parameter based on the polarization ghost image; calculating a linear polarization background scattering parameter and a water body transmission function based on the Stokes vector and the stabilization correction parameter, and calculating a linear polarization de-scattering restored image based on the polarization ghost image, the linear polarization background scattering parameter and the water body transmission function; and calculating a circular polarization background scattering parameter and a circular polarization water body transmission function based on the linear polarization de-scattering restored image and the stabilization correction parameter, and calculating a circular polarization de-scattering restored image based on the linear polarization de-scattering restored image and the circular polarization water body transmission function. And the imaging quality is obviously improved.

The invention provides a ghost imaging monitoring method and system based on environmental coupling simulation and image enhancement, and the method comprises the steps: setting an initial light source, carrying out the beam splitting of the initial light source through a beam splitter matrix model, so as to obtain a detection light path light source and a reference light path light source, and carrying out the detection of a ghost imaging image based on the detection light path light source and the reference light path light source. Respectively constructing a detection light path and a reference light path; a lens, an object to be detected and a barrel detector are sequentially arranged on a detection light path, a surface detector is arranged on a reference light path, Gaussian blur is introduced into the detection light path to simulate a heavy fog environment, and a Perlin noise-based random phase screen is introduced to simulate a turbulent environment. According to the method, the ghost imaging system is built through Python, a heavy fog environment and a turbulence environment are simulated by utilizing Gaussian simulation, and gray-scale map optimization is performed on the imaging quality of ghost imaging through nonlinear mapping in combination with a histogram equalization concept in digital image processing, so that the imaging quality of the ghost imaging system under a long-distance condition can be improved.

The invention discloses a ghost imaging system based on adjoint particles, belongs to the technical field of quantum imaging and particle detection crossing, and aims at solving the problems that a traditional particle imaging system is high in cost, limited in precision and complex in architecture. The system comprises a neutron source, a barrel-type detector, a high-precision alpha particle detector array and a data processing and reconstruction unit, a D-T neutron source is preferably selected as the neutron source, and alpha particles in the direction close to 180 degrees are generated when neutrons are emitted; the barrel-type detector does not need spatial resolution capability and only measures total energy deposition and flux fluctuation of neutron beams; the alpha particle detector array collects three-dimensional information of alpha particles (x, y, t) to construct a space illumination mode; and the data processing unit reconstructs modulation information of an object on the neutron beam and generates an image by calculating a second-order correlation function of an alpha particle position mode and neutron flux fluctuation. Hardware configuration is simplified, cost is reduced, imaging precision is improved through particle quantum correlation, and the method is suitable for the fields of medical imaging, material analysis and nuclear technology application.

The application discloses a high-resolution image reconstruction method and related device based on single-pixel detection, and belongs to the technical field of computational imaging and deep learning; the method generates a single-pixel detection signal through a target image and a trainable binary coding pattern; preliminary image information is reconstructed from single-pixel data by using a differential ghost imaging (DGI) technology; a combined loss function is designed to ensure the reconstruction accuracy of the image at the pixel level and the structure level; and the preliminary image information is input into a trained MSAW-UNet model to obtain a reconstructed image. The application can effectively reconstruct a high-resolution and high-quality image by using sparse single-pixel data under a low sampling rate, and overcomes the limitations of traditional reconstruction methods in resolution and image quality.

The invention discloses a secret image sharing method and system fusing calculation ghost imaging and federal learning, and belongs to the field of information security. The system comprises a user side and a quantum cloud server; the system further comprises a homomorphic encryption module, a secret sharing module, a model training module, a parameter aggregation module, an image encryption module and an image decryption module. According to the method, a measurement matrix is distributed through a Shamir secret sharing technology, and high randomness and unpredictability of the measurement matrix are ensured by utilizing quantum random number generation. The system is protected by adopting a homomorphic encryption technology, and cooperative training of local data of each user side is realized through federal learning, so that each user side can obtain a global model finally without directly sharing original data. In the image transmission stage, the user side can perform secure communication by using the recovered measurement matrix and the calculation ghost imaging generation bucket detection value, so that the efficient decryption and sharing of the image are realized on the premise of ensuring the data privacy.

This application belongs to the field of optical computational imaging and machine vision technology, and more specifically, relates to a ghost imaging method and system based on a physically driven untrained dual-network architecture. The method first constructs a two-stage untrained neural network architecture of "initial reconstruction - fine reconstruction". The first-stage network quickly locks the low-frequency spatial structure of the object and introduces it as a priori constraint into the optimization process of the second-stage network. Combining a physical imaging model and regularization constraints, high-frequency detail features are recovered from bucket probe signals with extremely low sampling rates. Finally, through a physically consistent light intensity scale recovery mechanism, the normalized features output by the network are mapped back to the real physical light intensity space. This invention does not require pre-training data and effectively overcomes the technical difficulties of existing technologies that easily get trapped in local optima, suffer from significant noise, and lose physical light intensity scale at extremely low sampling rates (e.g., 5%), achieving high signal-to-noise ratio and high-fidelity target imaging.

The application provides a kind of based on physical drive deep learning's non-diffraction computing ghost imaging optimization method, using the deep learning method of cyclic consistent generative adversarial network model based on physical drive inhibits the image quality degradation caused by quasi-non-diffraction speckle point spread function multi-stage side lobe, improves the imaging quality and depth of field of computing ghost imaging, comprising: using annular phase template generates quasi-non-diffraction speckle;The binary image data set is reconstructed to obtain the initial reconstructed image;The real physical imaging degradation process is embedded into the network as an explicit physical priori, a physical driven double cycle generative adversarial network model is constructed and trained;The transmission light intensity of the actual target is collected by using a single-pixel detection system and is correlated and reconstructed, and the reconstructed image with high-order oscillation is input into the trained network model, and finally a high-quality target image is reconstructed.The application significantly removes the high-order oscillation caused by the side lobe of quasi-non-diffraction speckle point spread function, while maintaining the advantage of large depth of field, realizes high-resolution and high-quality image reconstruction.This also provides a new idea for the subsequent research of long-distance imaging of computing ghost imaging.

The invention is suitable for the technical field of video processing, and provides a multi-dimensional vector matrix-based ghost imaging video inverse filtering motion blur removal method, which comprises the following steps of: 1, acquiring a ghost imaging video based on four-dimensional vector Walsh transformation; step 2, ghost imaging video filtering processing; 3, optical flow field calculation and motion blur detection are carried out; and step 4, carrying out inverse filtering deblurring. The ghost imaging video based on the multi-dimensional vector matrix is processed by using an inverse filtering method for fuzzy function estimation, and the fuzzy kernel is estimated through the optical flow information of the fuzzy frame and the previous frame, so that the motion blur is effectively eliminated, and the structural contour integrity of a relative motion object is kept; and the processing result can ensure that the video sequence has a consistent clear effect in space. The algorithm is suitable for fuzzy recovery of uniform linear motion in a ghost imaging system, and a new thought is provided for solving the systematic motion fuzzy problem.

The embodiment of the invention discloses a light collecting system which comprises a lens group and a lens barrel, and the lens group is installed on the lens barrel. The mask is used for realizing a spatial filtering function and is arranged in the lens cone; the inner wall of the lens cone is provided with a low scattering layer. Meanwhile, the embodiment of the invention discloses a defect detection system for the to-be-detected object with the pattern, and the light collecting system is adopted. As the space filtering function is embedded in the lens cone, the diffraction light is ensured not to form interference on the external space of the light collecting system; therefore, the interference of stray light generated by the spatial filter on the whole detection system is effectively avoided; aberration correction of the detection system is optimized, the imaging quality is improved, the detection precision of the system is further improved, and the detection quality of the system is guaranteed; meanwhile, the inner wall of the lens cone of the light collecting system is provided with the low scattering layer, and the low scattering layer can absorb part of irradiated light, reduce scattering intensity and absorb part or all of reflected light beams, so that ghost imaging, flare or dazzle light phenomena caused by diffraction stray light on imaging are avoided.