36 results about "Blind deconvolution" patented technology

The invention relates to the technical field of medical image processing, in particular to a digestive endoscopy image deblurring enhancement method and system.The method comprises the steps that firstly, an input digestive endoscopy original image is processed through a blurred region classification network, and a pixel-level blurred classification map capable of distinguishing an adhesion blurred region and a motion blurred region is generated; then, parallel processing is carried out according to the classification graph: for an adhesion fuzzy region, physical model restoration and color correction are carried out by estimating a transmissivity graph and an ambient light value; for a motion blur region, a self-adaptive non-blind deconvolution kernel is constructed to perform deconvolution sharpness. And finally, inputting the two processing results and the original clear area into a multi-scale feature fusion network together, carrying out adaptive feature weighted fusion and image reconstruction, and outputting a globally clear and detail-enhanced final image. According to the method, accurate identification and targeted enhancement of composite blurring are realized, and the visual quality and diagnosis availability of the digestive endoscopy image are effectively improved.

The invention discloses a fan blade aerial image deblurring method fused with auxiliary image prior, which comprises the following steps of: establishing an image degradation model, estimating a blurring kernel and a potential image in the image degradation model by optimizing a target function, and after the blurring kernel is obtained, performing image deblurring on the image degradation model; non-blind deconvolution is applied to the input blurred image to generate a final clear image. According to the method, the auxiliary image is introduced in the process of estimating the blurring kernel, the structural consistency and defect identification degree of the recovered image are remarkably improved, a breakthrough from natural image priori to task specific priori is achieved, after the blurring kernel estimation is completed, standard deconvolution is not directly used, and the algorithm is simple and convenient to operate. Compared with the prior art, the non-blind deconvolution energy model fusing multiple image structure priori is innovatively proposed, the common problems of artifacts, edge blurring and the like in a traditional deconvolution image are greatly solved, high-fidelity image reconstruction is achieved, and the accuracy and robustness of a subsequent defect detection system are enhanced.

The present application relates to the adaptive signal denoising decomposition method for mechanical transmission system compound fault identification, belongs to the mechanical vibration signal processing and fault diagnosis technical field. Adopting the autoregressive model to remove the harmonic component of vibration signal adaptively, through the construction filter group decomposition AR denoising vibration signal, obtains a series of modes, estimates the fault period of mode, and with the help of blind deconvolution theory, adaptive iteration updates filter and mode, carries out mode selection according to multidomain correlation coefficient and fault period consistency coefficient, determines the optimal filter length according to the weighted square envelope harmonic noise ratio, carries out square envelope spectrum analysis to the optimal mode obtained after adaptive denoising decomposition, and finally realizes compound fault identification. The present application can carry out adaptive denoising decomposition to the vibration signal of mechanical transmission system, does not need to construct priori base function and fault period priori knowledge, and significantly improves the accuracy of compound fault identification.

The invention discloses an image deblurring method and device based on blind deconvolution, and relates to the technical field of engine turbine blade strain measurement image deblurring. In order to solve the defect that the dual requirements for image quality and processing efficiency in engine blade strain detection are difficult to meet in the prior art, the technical scheme provided by the invention comprises the following steps: acquiring images of a blade in a static state and at different rotating speeds, taking a static image as a reference, and taking a dynamic image as a to-be-processed image; initializing a blind deconvolution parameter based on a Gaussian blur kernel; performing blind deconvolution iterative optimization under the constraint of a still image, and jointly estimating a clear image and a blurred kernel to obtain a preliminary deblurred image; de-noising is carried out; and adaptively selecting a gamma value according to the histogram mean value difference between the de-noised image and the still image, and executing gamma correction to enhance the contrast and details. According to the method, a deblurring-denoising-enhancing link is formed, the image definition can be effectively recovered under the condition of high-speed rotation, and the strain detection precision and stability are remarkably improved.

The invention provides a target signal and underwater acoustic channel decoupling method and system based on a conditional generative adversarial network, and the method comprises the steps: inputting a preprocessed receiving signal and condition information into a trained neural network model, and outputting an estimated channel response; deconvolution is carried out on the estimated channel response and the received signal to obtain a source signal estimation value, and decoupling of the target signal and the underwater acoustic channel is achieved; the neural network model is a conditional generative adversarial network. Compared with a traditional blind deconvolution method, the method has the advantages that the channel decoupling capacity is higher, and decoupling of signals and channels can be more effectively achieved; compared with the prior art, the method is higher in robustness, can generate target signals with consistency and authenticity in different complex environments, supports the continuous updating and optimization of the model according to the actual deployment environment, and remarkably improves the adaptability to the environment change. The method has better real-time performance and is suitable for real-time application of an underwater unmanned platform.

The present invention discloses a mechanical fault diagnosis method and system based on blind deconvolution. The method comprises: picking up a mechanical vibration observation signal through an acceleration sensor; constructing a weighted energy threshold coefficient, and performing single source point detection on the mechanical vibration observation signal collected by the sensor based on the weighted energy threshold coefficient to obtain feature data and an estimated number of sources H; normalizing the filtered feature data to obtain feature data represented by directional angles; clustering the feature data represented by directional angles to obtain cluster centers and membership degrees; reconstructing and separating H source signals containing single faults from a signal containing composite faults based on the cluster centers and membership degrees, performing an inverse short-time Fourier transform on the separated source signals to obtain time-domain source signals; performing feature enhancement on the time-domain source signals; and performing envelope analysis on the enhanced signals to determine the fault type. The present invention can achieve blind separation of composite faults when the number of sources is unknown, and use mechanical vibration signals for fault feature extraction and diagnosis.

The invention discloses an ophthalmology ultrasound RF signal enhancement method and device and a model training method and device, and relates to the technical field of ophthalmology ultrasound and the technical field of deep learning, and the method comprises the steps: obtaining to-be-enhanced ophthalmology ultrasound RF signal data of a target object, the to-be-enhanced RF signal data is obtained by performing ophthalmologic ultrasonic scanning on eyeballs of a target object through eyelids by adopting high-frequency pulse sound waves; rF signal data to be enhanced are input into a pre-trained signal enhancement model, blind deconvolution calculation is carried out on the RF signal data to be enhanced, obtained output serves as enhanced RF signal data, and the signal enhancement model is a deep learning model. By the adoption of the scheme, the comfort level of a patient during anterior segment ophthalmic ultrasonic examination is improved, and the examination efficiency is improved.

The present application relates to the technical field of image processing, and more particularly, to a kind of aluminized film quality detection method and system based on image recognition, comprising: obtaining the fuzzy image of aluminized film, and the aluminized film is printed with Chinese and English characters;The fuzzy image is divided into multiple image blocks, and for each image block, based on the gradient feature of the image block, its fuzzy intensity index is calculated.The present application divides the image into blocks, and calculates the fuzzy intensity index and the fuzzy sensitivity degree for each image block respectively, and then generates a spatial variability guide weight by combining the two.The weight can intelligently identify the key areas in the image that are both severely blurred and contain complex characters, and guide the blind deconvolution algorithm to focus on accurately restoring these areas, effectively solving the defect that traditional methods are difficult to handle spatially varying blur, and significantly improving the restoration quality of the blurred image.

The application relates to an OCT image processing method and system based on adaptive weighted bilateral filtering deconvolution, which comprises the following steps: preprocessing an original OCT signal, including removing background noise, wave number calibration, spectrum shaping and dispersion compensation, to obtain an OCT reconstructed image; adjusting the spatial weight and intensity weight of the OCT stretched image through histogram stretching to enhance contrast, and adaptively denoising the image; further optimizing the image by using a blind deconvolution algorithm, iteratively updating the estimated image and a point spread function (PSF), and obtaining a high-definition OCT reconstructed image. By using the adaptive weighted bilateral filtering and blind deconvolution technology, the application realizes efficient denoising and detail reservation of the OCT image, thereby obtaining a high-definition reconstructed image.

The invention discloses an unsupervised reconstruction method for adaptive optical image restoration, which is applied to the technical field of astronomical observation and aims to solve the problem that in the prior art, the imaging resolution is reduced due to atmospheric turbulence in the observation of a ground-based solar telescope. According to the method, multi-frame blind deconvolution is creatively fused into an unsupervised deep learning framework, an encoder-decoder network is constructed to generate a potential clear image, meanwhile, a depth priori condition network is constructed to optimize a fuzzy kernel, and efficient recovery is achieved by solving network parameters and the multi-frame blind deconvolution problem through alternate iteration. A GCM model is introduced into a corrector to ensure stable convergence, and a BM3D algorithm is adopted in the denoising process. Compared with the prior art, the method does not need a large amount of training data, is high in generalization capability, is high in recovery quality, can process non-isoplanatic and multiband images, achieves the complementary advantages of a conventional algorithm and deep learning, is high in calculation efficiency, is high in practicality, and provides a more advanced and reliable image processing technology for the fields of astronomical observation and the like.

A data receiving device with a blind deconvolution mechanism. A descrambling circuit descrambles the received data based on a single-antenna assumption and N data position assumptions within a transmission period, generating N sets of soft-bit data. A soft information processing circuit extracts bit position information to determine invariant and variable bit positions for a blind deconvolution process, including: cyclically storing and superimposing N sets of soft-bit data to generate N sets of superposition results, corresponding to N data position assumptions in N buffers of a buffer circuit; and retaining data corresponding to invariant bit positions at the start of a new transmission period. A post-processing circuit deinterleaves and decodes the N sets of superposition results corresponding to each data sub-block to generate N sets of decoded results for verification. When any N sets of decoded results pass verification, the soft information processing circuit terminates the blind deconvolution process.

The invention provides a computational adaptive optical microscopic imaging method based on sparse blind deconvolution. According to the method, sparse prior of a fluorescence sample and an aberration correction algorithm based on a Zernike polynomial are combined, compared with traditional Richardson-Lucy deconvolution, a point spread function and sample information can be reconstructed from a single blurred image at the same time, the requirement for accurate point spread function calibration is avoided, and the method is suitable for large-scale popularization and application. And the method is better in the aspects of improving the robustness to noise and stabilizing the deconvolution process. According to the invention, the image resolution and contrast of the wide-field and confocal fluorescence microscope system are obviously improved, and the wide-field and confocal fluorescence microscope system is suitable for urgent demands in the fields of dynamic life science research, medical imaging and the like, and has a wide application prospect.

The application discloses an on-board multi-sensor space-time online calibration method in a fully mechanized coal mining face complex environment and belongs to the technical field of image processing and environment perception. With the main IMU time axis as a unified reference, the time offset of the camera and the laser radar inertial measurement is estimated and compensated online, and a continuous time pose prediction model is established. The point cloud is subjected to motion compensation and time unification, combined with the time sequence motion and echo intensity consistency discrimination to remove the jitter ghost points, and the de-ghosting point cloud is obtained. Combined with the camera continuous time pose and the rolling shutter exposure model of each row, a spatially changing blur kernel is constructed, the non-blind deconvolution of the image is carried out to obtain a clear image. Finally, based on the de-ghosting point cloud and the clear image, a cross-modal structure consistency constraint is constructed, and the camera-laser radar space external participation time offset is solved through nonlinear optimization. The application can realize high-precision, lightweight and sustainable online space-time calibration, and improve the stability and reliability of the multi-modal data fusion of the on-board equipment in the fully mechanized coal mining face.

The invention specifically discloses a fault impact deconvolution tracking method for adaptive period detection of a rail transit intelligent bearing, and the method comprises the following steps: obtaining a vibration acceleration signal of the intelligent bearing, setting an expected filter length range, dividing the filter length range into two subintervals, determining three special points, searching an optimal filter coefficient; and calculating and comparing target harmonic energy ratios THER in the blind deconvolution signal envelope spectrums of the three special points, determining the position of the maximum value, determining a filter length search interval, if the interval length reaches a threshold condition, outputting the filter length and the optimal filter coefficient corresponding to the maximum index, and outputting the final filtered blind deconvolution signal. And obtaining an envelope spectrum of the blind deconvolution signal, and evaluating the degradation state of the intelligent bearing. By adopting the technical scheme, vibration signal acquisition, cyclic frequency estimation, recursive search and fault diagnosis are performed, the fault period of the intelligent bearing is positioned, and the calculation efficiency is remarkably improved while the diagnosis precision is ensured.

This invention discloses an automatic method for measuring the long axis of the aorta in left ventricular ultrasound images, relating to the field of medical image processing. The method involves collecting left ventricular ultrasound images, converting the original images into grayscale images, and processing each grayscale image using an improved semi-blind deconvolution algorithm. Key points are sequentially labeled with their serial numbers and coordinate positions on each processed image. A training set and a validation set are then divided from the labeled image set. An improved HRNet model is trained using the training set to obtain a key point detection model. Images from the validation set are input into the key point detection model, detecting six key points in each input image that correspond to the endpoints of the aortic long axis. Straight lines are drawn using each pair of key points as endpoints. The length of each straight line is calculated based on the coordinates of its endpoints and multiplied by the image scaling factor to obtain the aortic long axis measurement result.

The present disclosure discloses a data receiving apparatus and a data receiving method having blind deconvolution mechanism. A descrambling circuit descrambles received data according to an antenna assumption and N data position assumptions within a transmission period to generate N groups of soft-bit data. A soft-bit processing circuit retrieves bit position data to determine non-variable bit positions and variable bit positions. N circular buffers of a storage circuit store and superimpose the N groups of soft-bit data corresponding to N data position assumptions in a circular manner to generate N groups of superimposed results and keep the data corresponding to the non-variable bit positions. A post-processing circuit performs de-interleaving and decoding on the N groups of superimposed results to generate N groups of decoded results to perform redundancy check thereon. When one decoded results passes the redundancy check, the soft-bit processing circuit stops performing the blind deconvolution process.

The present application relates to a hyperspectral image sub-pixel positioning method with edge preservation, comprising: S1, using a Gaussian blur kernel and an original hyperspectral image to perform non-blind deconvolution to reduce the point spread function effect in the original hyperspectral image; S2, using a domain transform recursive filter to filter the result of step S1 to play a role in edge preservation and noise reduction; S3, obtaining an abundance image through a linear spectral unmixing technique; S4, using an interpolation method to up-sample the abundance image to obtain soft class values of each type of sub-pixel; S5, using a class assignment strategy to assign a class label to each sub-pixel to obtain the final sub-pixel positioning result. The present application can effectively process the point spread function effect and the noise such as texture details in the image, thereby improving the sub-pixel positioning accuracy.

The application relates to a multi-focus photoacoustic imaging registration fusion method based on blind deconvolution, which comprises the following steps: acquiring photoacoustic images of different laser wavelengths and different depths of field by using a fast photoacoustic imaging device; establishing a blind deconvolution model and initializing a blind deconvolution blur kernel; based on the blind deconvolution model, deconvolving the photoacoustic image and updating the blur kernel to obtain an optimized photoacoustic image through blind deconvolution, and continuously optimizing by using an image quality judgment function; using an intensity-based photoacoustic image registration algorithm to determine optimal transformation parameters based on intensity information, and registering the photoacoustic images of different depths of field after blind deconvolution optimization; using an image fusion algorithm to fuse the registered photoacoustic images to generate a focused fusion high-quality super-resolution photoacoustic image. Compared with the prior art, the application has the advantages of non-invasiveness, non-destructiveness, convenience, fast realization of photoacoustic image feature registration and fusion of different depths of field, and recovery of high-resolution imaging.

The present application relates to the technical field of unmanned aerial vehicle, especially to a kind of unmanned aerial vehicle aerial target detection repair method for optical distortion, comprising the following steps: obtaining unmanned aerial vehicle aerial video sequence, extract continuous frame image;Based on the statistical characteristics of thermal turbulence in a short time, construct space-time varying distortion model, distortion estimation is carried out to video sequence.The present application can effectively estimate and reverse optical distortion caused by thermal turbulence by constructing space-time varying distortion model and combining blind deconvolution technology, thereby significantly improving the definition and geometric consistency of unmanned aerial vehicle aerial image, this method makes full use of the time sequence information between continuous frames and turbulence statistical characteristics, realizes the accurate modeling of distortion field, and then restores the image with more detailed and clearer edges by inverse processing, provides high-quality input for subsequent target detection.

This invention discloses a mechanical fault diagnosis method based on maximizing the generalized cyclostationary index, belonging to the fields of signal processing and mechanical equipment condition monitoring and fault diagnosis. It mainly involves blind deconvolution of the signal based on maximizing the generalized cyclostationary index to extract the cyclostationary features of interest and perform envelope analysis. The method includes shape parameter estimation based on a generalized Gaussian cyclostationary model; blind deconvolution based on maximizing generalized cyclostationarity; and β-order envelope spectrum analysis. The beneficial effects of this invention are that, based on maximizing the generalized cyclostationary index, it improves the robustness of cyclostationary feature extraction for non-Gaussian distributed signals; and by setting a tolerance bandwidth in the weighting matrix generation, it solves the problem of inconsistency between the theoretical and actual cyclostationary frequencies.

The invention discloses a monitoring data processing method and system for a bee colony node of an unmanned aerial vehicle, and relates to the technical field of unmanned aerial vehicle data processing, and the method comprises the steps: collecting the junction temperature data and external flow field data of the unmanned aerial vehicle; calculating the Reynolds number based on the flow field data and the fuselage characteristic length, and judging the air flow state of the fuselage heat dissipation surface; in response to the fact that the junction temperature data exceeds a safety threshold value and the air flow state is laminar flow, generating a micro-maneuvering disturbance instruction, driving the unmanned aerial vehicle to generate micro-amplitude vehicle body vibration, and converting a vehicle body heat dissipation surface flow field into turbulent flow; and constructing a point spread function based on the waveform parameter of the micro maneuver control signal, carrying out non-blind deconvolution restoration on original monitoring data collected in a micro-amplitude machine body vibration state, and inputting the original monitoring data into a target detection network to carry out identification and analysis of a monitoring target. According to the method, the laminar flow boundary layer is damaged by micro-motor vibration, the heat dissipation efficiency is improved, non-blind deconvolution is performed by using the vibration parameters to eliminate imaging blur, and the continuous operation capability of the bee colony node under high load is ensured.

The present invention discloses a defocused image deblurring method based on boundary neighborhood gradient difference, which belongs to the technical field of computer vision. In view of the fact that the existing defocused image deblurring method cannot accurately obtain the blur amount of the boundary position of the defocused image for static scenes with multiple depth layers, the present invention makes full use of the relationship between the boundary neighborhood gradient difference and the blur amount to accurately obtain the blur amount of the boundary position of the defocused image, thereby solving the problem of boundary ringing artifacts appearing in the deblurring result; in view of the problem that the ability of the non-blind deconvolution algorithm to retain image detail information is not strong enough, resulting in the loss of detail information in the deblurring result, the present invention combines the discrete blur amount selection strategy and the sparse prior to design a non-blind deconvolution algorithm to enhance the ability of the non-blind deconvolution algorithm to retain image detail information, solves the problem of the loss of detail information in the deblurring result, and can be used for defocused image deblurring processing in static scenes with multiple depth layers.

An image processing method for measuring object displacement is provided. The method includes acquiring a first sequence of images and a second sequence of images, two adjacent images of the first sequence of images including a first overlap portion, two adjacent images of the second sequence of images including a second overlap portion, the first sequence of images corresponding to a first three-dimensional (3D) surface including a curved surface on an object in a first state, the second sequence of images corresponding to a second 3D surface on the object in a second state. The method also includes deblurring the first sequence of images and the second sequence of images based on a blind deconvolution method to obtain sharp focal plane images, stitching the sharpened first sequence of images and the sharpened second sequence of images into a first 3D image and a second 3D image based on camera pose estimation performed by solving a perspective-n-point (PnP) problem using a refined robustly weighted Levenberg-Marquardt (RRWLM) algorithm. The method also includes forming a first two-dimensional (2D) image and a second 2D image by unwrapping the first 3D image and the second 3D image, respectively, and generating a displacement map image from the first 2D and second 2D images by performing a two-dimensional digital image correlation (DIC) method.

The invention relates to the technical field of image processing, in particular to an aluminum laminated film quality detection method and system based on image recognition, and the method comprises the steps: obtaining a fuzzy image of an aluminum laminated film printed with Chinese and English characters; and dividing the blurred image into a plurality of image blocks, and for each image block, based on the gradient feature of the image block, calculating to obtain a blurring intensity index of the image block. According to the method, the image is divided into blocks, the fuzzy intensity index and the fuzzy sensitivity degree are calculated for each image block, and then a spatial variability guide weight is generated by combining the fuzzy intensity index and the fuzzy sensitivity degree. The weight can intelligently identify key areas which are seriously blurred and contain complex characters in the image, and guide the blind deconvolution algorithm to focus on accurate restoration of the areas, so that the defect that a traditional method is difficult to process space change blurring is effectively overcome, and the restoration quality of the blurred image is remarkably improved.

The present invention discloses a railway switch fault analysis method and system. The method includes: converting a three-phase operating current into a three-phase operating current digital signal according to a preset digital conversion strategy; extracting cyclic frequency information from the three-phase operating current digital signal and generating an enhanced envelope spectrum; quantizing and accumulating the relative characteristics of multiple cyclic frequencies in the enhanced envelope spectrum to obtain a diagnostic characteristic spectrum, and identifying the equal-frequency-interval harmonic structure in the enhanced envelope spectrum based on the diagnostic characteristic spectrum to obtain the cyclic frequency; feeding the cyclic frequency into a preset maximum cyclostationary blind deconvolution, the maximum cyclostationary blind deconvolution outputs a repetitive transient pulse caused by a local defect, and determining the railway switch fault signal based on the repetitive transient pulse. The method improves the accuracy of railway switch fault identification, enables good identification of fault-related signals even in the presence of strong external noise interference, and is beneficial to the troubleshooting of railway switch faults.

A fault detection system is provided for extracting fault signatures from time-domain stator current signals of a motor under varying load operation by solving a joint blind deconvolution demodulation problem. The stator currents of the motor under varying operating conditions are modeled as stator currents under steady-state operating conditions affected by a system response vector and a load modulation vector. A proximal alternating linearization minimization method is used to solve the joint blind deconvolution demodulation problem, assuming that the spectrum of the desired signal is sparse. Motor fault detection is then performed using the recovered stator currents under steady-state operating conditions.

The invention discloses a computational adaptive optical microscopic imaging method based on sparse blind deconvolution. The method comprises the following steps: firstly, carrying out multi-frequency fusion filtering preprocessing on an acquired fluorescence image, inhibiting low-frequency noise and enhancing medium-high frequency details; then, based on image gradient sparse prior, the image and a point spread function are updated through blind deconvolution alternate iteration; then utilizing a Zernike polynomial to fit aberration to obtain an aberration-containing point spread function; and finally, performing deconvolution to realize image deblurring and aberration correction. According to the method, extra hardware and precise point spread function calibration are not needed, and high noise robustness and system universality are achieved; compared with traditional Richardson-Lucy deconvolution, the method has the advantages that the point spread function and the sample information can be reconstructed from a single blurred image at the same time, the requirement for accurate point spread function calibration is avoided, the image resolution and contrast of a wide-field and confocal fluorescence microscope system are improved, and the method is suitable for being applied to the fields of dynamic life science research, medical imaging and the like.

The invention discloses a bearing residual life prediction method based on elastic blind deconvolution and a time convolution neural network, and relates to the technical field of vibration signal intelligent fault diagnosis. The method comprises the following steps: collecting bearing acceleration signals; z-score normalization processing is carried out on the collected acceleration signals of the bearing; convolution layer parameter initialization is realized through uniform initialization; minimizing a nonlinear L1 / 2 norm through an Adam optimizer so as to enhance fault features in the acceleration signal; a step-by-step ablation strategy based on a harmonic energy distribution index is adopted, and irrelevant interference components and repeated fault components are eliminated in a fault feature enhancement process to improve the calculation efficiency and realize end-to-end fault feature output; monitoring an output signal by adopting a harmonic energy distribution index threshold value so as to determine the occurrence time of a bearing fault; and mapping the output signal into residual life by using a residual life prediction module.

A profile generation method and device based on an intelligent agent for high-resolution processing of seismic data, comprising processing original seismic data through a blind deconvolution model to obtain an initial wavelet and an initial reflection coefficient; convolving the initial wavelet and the initial reflection coefficient to obtain an initial seismic profile; extracting initial profile features of the initial seismic profile; inputting the initial profile features, parameters of the initial wavelet, and parameters of the initial reflection coefficient into a pre-trained reinforcement learning intelligent agent for calculation; updating the parameters of the initial wavelet and the parameters of the initial reflection coefficient by the reinforcement learning intelligent agent to obtain parameters of a final wavelet and parameters of a final reflection coefficient; and convolving the final wavelet and the final reflection coefficient to obtain a target high-resolution seismic profile. The method of the present application uses a convolution model as the physical basis of the reinforcement learning environment, realizes reinforcement learning driven by a physical model, and obtains a seismic profile with higher resolution and more consistent with actual geological structure.

The present invention relates to the field of blur detection technology, and discloses a motion blur detection method and system based on deep learning and blind deconvolution, including: collecting an original image and performing sub-region division of the image; performing singular value decomposition on each sub-region respectively; detecting the blur region of the sub-region according to the decomposition result; performing image restoration on the blurred image; using the method of deep learning to segment the target contour information in the blurred restored image, and outputting a clear image with the target contour. By introducing SPD-Conv, the improved YOLOv8 model can more accurately segment low-resolution blurred targets, effectively improving the segmentation accuracy. It can adapt to the low-resolution blurred target segmentation tasks in different scenarios. While improving the segmentation accuracy, the method proposed in the present invention does not significantly increase the computational complexity and inference time of the model, and still maintains the real-time advantage of YOLOv8, meeting the requirements for real-time performance in practical applications.