535 results about "Motion blur" patented technology

The invention discloses an unmanned aerial vehicle laser and vision fusion inspection method and system for bridge bottom disease detection, and the method comprises the steps: carrying out the synchronous data collection through employing a calibrated laser radar, a camera and an IMU, and obtaining a three-dimensional laser point cloud and a two-dimensional visual image of the appearance of a bridge; sharpening the image containing the motion blur and completing brightness self-adaption of the image; stable feature points are extracted, multi-frame matching is carried out, the corresponding poses of the images are estimated, and bridge dense point cloud reconstruction is completed; performing geometric component segmentation on the point cloud to generate a geometric prior region; component segmentation is carried out on a support area in the image, and a continuous and accurate component segmentation result is obtained in combination with a geometric prior area; screening the image, calling a targeted disease detection model in a corresponding component area, and generating a segmentation mask for the disease; obtaining a real disease three-dimensional point cloud, and carrying out quantitative calculation on the physical size of the disease; and displaying the real disease three-dimensional point cloud data and the physical size of the disease. The method is high in efficiency and precision.

The invention relates to the technical field of fabric detection, and discloses a fabric defect intelligent detection method and system based on AI visual identification. According to the method, motion blur is quantized through motion state data, optical blur caused by fabric motion is eliminated through deconvolution solution, so that motion interference in the fabric transmission process is processed in a targeted mode, self-adaptive balance of the deblurring capacity and the feature retention capacity is achieved, and then based on the optical interference principle, the deblurring capacity and the feature retention capacity are improved. Through a dynamic calibration system combining hardware-level real-time compensation and multi-dimensional optical parameter calibration, dynamic optical parameter calibration of primary correction data is realized, then fabric defect characterization data is extracted to accurately obtain defect features, and finally, a detection-production line control closed loop is constructed through a quality quantitative index and a comprehensive risk value, so that fabric defect detection is realized. The fabric defect detection precision can be improved, so that the problem of high defect missing detection and false detection rate caused by optical data distortion due to movement and environment interference in a traditional method is effectively solved.

The invention discloses a light guide plate defect detection method and system based on a neural network, and particularly relates to the technical field of machine vision detection, and the method comprises the following steps: aiming at the problem of image instability of a light guide plate in a dynamic transmission or rotation process, continuously collecting an image sequence and extracting time domain features; and performing interference judgment in combination with the inter-frame consistency prediction coefficient and a first threshold to realize accurate identification of the abnormal image frame. For an abnormal image frame, further correcting the recognition credibility of the abnormal image frame by adopting a confidence adjustment and fusion mode, and meanwhile, introducing a frequency domain transformation and image enhancement strategy to compensate detail loss caused by motion blur; according to the method, inter-frame consistency analysis, confidence fusion regulation and control and frequency domain fuzzy recognition and compensation mechanisms are introduced, abnormal judgment and image quality restoration of the light guide plate image in the dynamic scene are realized, the recognition accuracy and stability of the neural network model on the defect type, position and confidence are improved, and the false detection and omission ratio is effectively reduced.

The invention discloses an image restoration method, an image restoration device and a storage medium, which are used for improving the structure restoration precision and the detail restoration capability of image restoration. The method comprises the following steps: acquiring multi-dimensional inertial data in real time; performing frequency domain analysis on the multi-dimensional inertial data by adopting sliding window short-time Fourier transform to obtain vibration intensity; if the vibration intensity does not exceed the preset threshold value, acquiring an image; calculating a definition index of the image; determining whether the image is a blurred image according to a preset definition standard and the definition index of the image; if the image is judged to be a blurred image, dividing the blurred image into a motion blurred image and a focusing blurred image; respectively constructing point spread function models of the motion blurred image and the focusing blurred image; performing deconvolution processing or depth reconstruction on the blurred image through a point spread function model to obtain a clear image; and recalculating the definition index of the clear image, and if the definition index does not exceed the definition threshold, triggering reacquisition or switching the repair model to execute secondary repair.

The invention provides an image noise mark feature selection method and system, a storage medium and a computer. The method comprises the steps of obtaining a to-be-processed image noise mark data set; embedding a sample set in the image noise mark data set into a multi-granularity fuzzy cluster to construct a dynamic fuzzy membership evaluation matrix; dynamically evolving a multi-level high-precision granular ball cluster; obtaining mark distribution with high identification degree; constructing a rough perception feature evaluation framework based on granular ball topology driving, extracting decision equivalence classes by combining rough set upper and lower approximation and extended positive domain models, and determining and measuring the contribution degree of each feature to a decision system by fusing multi-granular-ball decision boundary information based on a dependency degree quantitative model; a particle and ball structure consistency verification mechanism is introduced, and multi-level evaluation is carried out on the importance of the features through dependency and consistency. According to the method, the optimal feature subset with strong anti-noise performance and high discrimination capability is obtained, and stable and efficient input support is provided for a subsequent image noise mark learning model.

The invention relates to the technical field of computers, and discloses a real-time rendering and interaction method and system for an immersive virtual reality scene. The method comprises the following steps: fusing tuner inertial data and eyeball tracking data, and constructing a prospective state prediction model; generating a predictive focus field in combination with scene visual saliency; synthesizing an anisotropic temporal-spatial resolution graph according to the predicted head angular velocity; gPU variable-rate coloring is driven to realize non-uniform rendering; and re-projection or dynamic fuzzy correction is executed in a self-adaptive manner according to the attitude prediction error before display. According to the technical scheme, the perception delay and the rendering load are remarkably reduced, and the frame rate stability and the visual immersion in a high-dynamic scene are improved.

The invention provides a motion blurred bar code identification method and system based on multi-frame image fusion, and the method comprises the steps: obtaining a continuous image sequence containing a motion blurred bar code, carrying out the motion track feature extraction of the continuous image sequence, and obtaining a motion vector field and a pixel displacement track set of a bar code region in each frame image unit; performing multi-frame image fusion on the continuous image sequence based on the motion vector field and the pixel displacement track set to generate a candidate bar code image set; performing deblurring enhancement processing on the candidate bar code image set to obtain a clear bar code image unit after deblurring processing; and performing bar code area positioning and distortion correction processing on the clear bar code image unit to generate a standardized bar code image, and performing identification to obtain an identification result. According to the invention, the accuracy and reliability of bar code identification in a motion blurred scene can be obviously improved.

A time-of-flight pixel circuit includes a photodiode configured to generate charge in response to modulated light reflected from an object. First and second transfer transistors are coupled to the photodiode. The first transfer transistor transfers a first portion of charge from the photodiode in response to a first modulation signal and the second transfer transistor transfers a second portion of charge from the photodiode in response to a second modulation signal. The second modulation signal is an inverted first modulation signal. A first floating diffusion is coupled to the first transfer transistor to receive the first portion of charge in response to a first modulation signal. Each one of a first plurality of sample and hold transistors is coupled between a respective one of a first plurality of memory nodes and the first transfer transistor.

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 relates to the technical field of electric power detection, in particular to an electric power inspection robot detection method and system based on machine vision. According to the method, the three-dimensional model of the target area is constructed by pre-collecting the basic data of the area, the historical defect data is obtained to mark the detection key area, the initial path and the key detection point are optimized by using the genetic algorithm, redundant collection points are removed, and the accuracy of the visual detection path is improved; meanwhile, visual data are analyzed in real time through a path acquisition collaborative analysis module, an optimization instruction is generated to dynamically adjust a path, and the problems of low efficiency and missing detection caused by independent operation of vision and navigation are avoided; the visual perception module collects detection dynamic data of definition, vibration acceleration, environment wind speed and environment light, the environment adaptability module analyzes and outputs a comprehensive fluctuation value, a corresponding fluctuation value interval is obtained, a corresponding environment adaptability regulation and control instruction is generated, and influences caused by motion blur and environment sudden change are avoided.

According to the video stream image optimization method and device provided by the embodiment of the invention, a dynamic fuzzy kernel generation mechanism is innovatively designed, and the data processing security is ensured through space-time decoupling and physical constraint. A content-motion double-branch network is constructed, and a reliable feature reconstruction system is established in combination with residual connection and space-time transformation. A physical constraint corrector is introduced, and high-quality video output is provided while user privacy is protected through adaptive fusion and rule correction. According to the method, the defects of the traditional technology in the aspects of fuzzy processing, feature decoupling, image optimization and the like are effectively overcome.

The invention relates to the technical field of computer vision, in particular to an image target detection system and method based on deep learning, and the system comprises a dynamic feature alignment unit, a motion blur compensation unit and a feature fusion control unit. A dynamic feature alignment unit generates spatial deformation parameters through a deformable convolutional layer and an offset prediction sub-network, resamples a shallow high-resolution feature map, and realizes deep and shallow feature space alignment, and a motion blur compensation unit generates a motion vector based on brightness gradient field difference, constructs a mask and weights a suppression blur region, so as to realize deep and shallow feature space alignment. The feature fusion control unit analyzes local entropy and target size distribution, dynamically distributes feature weights and feeds back and optimizes offset parameters, a closed-loop learning loop is formed by the method, and the problems of inaccurate feature alignment, fuzzy interference and poor scene adaptation are solved.

The invention relates to the technical field of industrial automatic detection, in particular to a collaborative visual detection method and system for an SMT production line, and the method comprises the steps: obtaining a beat reference signal which is synchronous with the periodic motion beat of a mounting head in the SMT production line; determining a stable acquisition window of which the vibration intensity is lower than a preset vibration threshold value; in the stable acquisition window, sending a synchronous trigger instruction to a plurality of visual inspection cameras deployed on the SMT production line in a unified manner; performing space-time alignment processing on the multi-path image data; performing motion blur correction processing on the image data subjected to the space-time alignment processing; and carrying out validity verification on the result of the space-time alignment processing, and executing a preset exception processing strategy when the verification result indicates that the verification fails. According to the invention, the beat reference signal synchronized with the motion beat of the mounting head is directly obtained, and the opportunity decision of image acquisition and the real-time vibration monitoring result are deeply bound, so that the double strategies of beat synchronization and vibration avoidance are realized.

The invention relates to a traffic accident intelligent detection system and method based on a YOLOv12 improved architecture. The system comprises a YOLOv12 enhanced feature extraction network, a multi-scale detection head, a time sequence information fusion module, a real-time reasoning optimization engine and an intelligent decision fusion system, and realizes collaborative optimization of local feature enhancement and global context modeling by constructing six core technology modules and adopting collaborative learning of a C2f-Attention mechanism and deformable convolution. According to the method, a composite loss function special for traffic accidents is innovatively designed, and adaptive fusion of multi-scale features and difficult sample mining are realized through a multi-objective optimization mechanism of Enhanced Focus Loss, IoU-aware Loss and Severage-aware Loss. According to the method, the problems of low detection precision and false alarm and missing alarm caused by illumination variation, shielding and motion blur in a traffic monitoring scene are effectively solved, in the test of an AccidentsDesection YOLOv8 data set, the mAP at 0.5 reaches 91.27% and is improved by 8.6% compared with that of YOLOv8, the reasoning speed reaches 67 FPS, experimental results show that the system has excellent performance in the aspects of detection precision, real-time performance and model compression, and the method is suitable for popularization and application. The method achieves a remarkable effect in traffic accident intelligent identification, and has a remarkable technical effect and industrial application value.

The invention relates to the technical field of embankment inspection, and provides an embankment inspection robot environment detection method and system, which can determine a motion blur direction, extract linear features of an image and determine a linear feature direction by acquiring an embankment image and performing motion blur judgment when the image has motion blur. And on the basis, according to the geometrical relationship between the motion blur direction and the linear feature direction, whether the linear feature is the embankment crack or not is judged. By using the motion blur direction as a judgment basis, a pseudo crack and a real crack caused by motion blur can be effectively distinguished, and early-stage potential safety hazard detection omission caused by image quality reduction is avoided, so that the accuracy and the reliability of embankment crack detection are remarkably improved, and the continuity and the effectiveness of an embankment inspection task are guaranteed.

The invention discloses an X-ray machine inspection parameter automatic configuration system for pet inspection, and relates to the technical field of medical imaging equipment, a displacement prediction model is used for pre-judging a pet convulsion state before exposure, a dynamic exposure trigger is only activated in a stable interval, and motion blur is avoided from the source; compared with the scheme of relying on post-exposure image feedback adjustment in the prior art, invalid radiation and repeated exposure operation can be avoided; the partition parameter mapping module is combined with a species feature database to map the gray level of the preview image into equivalent thickness and independently generate region parameters; aiming at extreme body type difference, such as an abdominal fat layer of an obese dog and a rib region of an emaciated cat, the system automatically distributes differentiated kV / mA parameters, and the problem of overexposure or underexposure caused by a traditional fixed penetration rate standard is eliminated; and the radiation fusing unit monitors the accumulated dose in real time, dynamically adjusts a safety threshold according to the weight of the pet, and stops exposure before the dose exceeds the limit.

The invention discloses a video deblurring method based on a deformable space-time sparse converter. The method comprises the following steps: carrying out feature extraction on continuous n frames of fuzzy videos, and calculating forward and backward light streams and corresponding fuzzy images; iteratively updating features of each frame through a bidirectional feature propagation module guided by a multi-scale fuzzy image, and aggregating feature information of different propagation branches; a deformable space-time sparse Transform module is adopted to carry out refinement processing on the aggregation features; and inputting the refined features into a decoder module, and generating a final deblurring result in combination with original input residual connection. The method is suitable for scenes of security monitoring, mobile equipment shooting and the like, the effect of removing the dynamic fuzzy area in the video can be improved, and the overall visual quality of the video is effectively improved.

The invention relates to an automatic milling cutter setting method and system based on machine vision, and belongs to the technical field of milling cutter setting. The method comprises the following steps: firstly, positioning initial position coordinates of a milling cutter, and planning an initial tool setting path of the milling cutter by combining target tool setting position coordinates; then obtaining an image of the milling cutter in the initial cutter setting path, carrying out image denoising and image deblurring processing, carrying out edge detection after obtaining a second image, extracting an edge contour of the milling cutter, calculating sub-pixel coordinates of edge points of the contour, and carrying out parametric fitting to obtain a current milling cutter position and a current milling cutter posture; inputting the initial tool setting path, the wear degree of the milling cutter, the current position of the milling cutter and the posture of the milling cutter into an error prediction model to predict the current motion error of the milling cutter; calculating the path compensation amount according to the current motion error of the milling cutter, and adjusting the tool setting path of the milling cutter according to the compensation amount. The method can reduce the interference of the motion blur of the milling cutter and environmental factors, and realizes the quantitative adjustment and correction of the tool setting of the milling cutter.

The invention provides a physical driving measurement method for monocular three-dimensional dynamic displacement of a rotary machine, and belongs to the technical field of crossing of computer vision and industrial state monitoring. According to the method, a high-speed dynamic visual acquisition system is constructed, and a time sequence video stream is obtained by using a cooperative marker; constructing a'Gaussian + motion blur 'composite gradient model taking the motion blur width as an endogenous variable, and jointly resolving sub-pixel edge coordinates and ambiguity by adopting a nonlinear optimization algorithm; decoupling the pixel displacement of the radial X axis, the radial Y axis and the axial Z axis by using the width change and the edge displacement of the marker in combination with the geometric principles of sequential robust filtering and monocular imaging; and finally, outputting a three-way physical vibration waveform by using the calibration conversion factor. According to the method, the fuzzy influence is adaptively eliminated from a physical imaging mechanism, micron-level precision three-direction vibration synchronous measurement is realized only by a single camera, and the hardware cost and the deployment difficulty are reduced.

The invention discloses an object level motion blur three-dimensional scene synthesis method and an object level motion blur three-dimensional scene synthesis device based on three-dimensional Gaussian splash (3D Gaussian splash, 3DGS). According to the method, a multi-view image and corresponding camera parameters are used as input, and three-dimensional Gaussian point scene representation including spatial position, scale, rotation, opacity and appearance characteristics is constructed; generating a binary mask of a moving foreground object and a static background by using an image segmentation model; on the basis, camera exposure time is dispersed into a plurality of time steps, space pose parameters changing along with time are introduced only for three-dimensional Gaussian points corresponding to the foreground object, instantaneous three-dimensional scene representation under each time step is obtained, integration or weighted averaging is carried out on rendering results of the plurality of time steps, and the real-time three-dimensional scene representation of the foreground object is obtained. Generating a motion blur rendering result with real physical significance; meanwhile, a mask-guided foreground-background separation rendering strategy and a partition loss function are adopted, clear rendering supervision is applied to a static background area, fuzzy rendering supervision is applied to a moving foreground area, and joint optimization of three-dimensional Gaussian scene parameters is achieved; in the reasoning stage, the clear background and the fuzzy foreground are fused and output according to the mask, and a three-dimensional scene reconstruction result with the clear background and the reasonable motion fuzzy effect of the foreground is obtained. According to the method, the exposure integral process and the object movement track are explicitly modeled in the three-dimensional geometric space, and object-level controllable fuzzy three-dimensional scene generation is realized.

The invention relates to the field of image recognition, in particular to a safety monitoring method and system based on image recognition. The method comprises the following steps: acquiring monitoring videos of a bank from multiple angles, and performing directional motion convolution elimination so as to construct a dynamic fuzzy elimination monitoring video; performing multi-target depth detection and image frame segmentation on the dynamic fuzzy elimination monitoring video, and extracting a plurality of personnel real-time image frames; performing continuous frame target tracking according to the plurality of person real-time image frames, and performing personalized trajectory feature mining to generate a plurality of person personalized trajectory features; and carrying out visual identification and potential threat object judgment on the plurality of person real-time image frames by holding articles one by one, and extracting suspicious person image frames. According to the invention, full-automatic and real-time response safety behavior analysis is realized, and the safety risk of a bank is reduced to the greatest extent.

The invention relates to a method for reducing motion blur of an erf model by using events and frames, and belongs to the technical field of computer vision. The method comprises the following steps: converting an event stream into triple representation; generating a bidirectional optical flow field based on the event flow, and aligning the fuzzy frame and the event flow through event-guided deformation convolution; constructing a double-branch optimization framework; the double-branch features are dynamically fused through frequency domain attention gating, and the weight of the frequency domain attention gating is generated by event frequency spectrums through MLP; generating a pseudo-true value based on an event double integral model, aligning an event spectrum and rendering a high-frequency component, and carrying out chain constraint on the continuity of a cross-frame radiation field through event brightness change; event-frame non-linear response differences are modeled by a learnable camera response function. Clear reconstruction and rendering of the moving target in a complex dynamic scene are realized.

The invention relates to the technical field of machine vision detection, particularly discloses a mechanical part surface defect vision detection system, and aims to solve the problems of blind areas, environment light sensitivity, insufficient tiny defect detection capability and the like in the detection of complex curved surface high-reflection parts. Comprising an optical imaging module, a geometrical morphology sensing module, a dynamic polarization regulation and control module, an image fusion and enhancement module and a defect identification and decision module. The three-dimensional morphology of the surface of the part is sensed in real time, the imaging polarization state is dynamically regulated and controlled according to the three-dimensional morphology, multi-frame polarization image fusion and digital diffraction enhancement processing are combined, full-coverage and high-contrast detection of the complex curved surface part is achieved, motion blur is effectively restrained, and the detection capacity of sub-pixel-level tiny defects and the system environment adaptability are improved.

The invention discloses a wheat seedling missing detection method for performing semi-supervised learning by fusing spatio-temporal information, and belongs to the technical field of seedling missing detection methods. The invention discloses a wheat seedling missing detection method for performing semi-supervised learning by fusing spatio-temporal information. According to the method, a set of semi-supervised learning framework is constructed; an initial model is trained by using a small number of precisely-labeled frames, and pseudo labels are generated for unlabeled frames; and'time sequence consistency 'and confidence coefficient are introduced to form a joint screening mechanism, and high-quality spatio-temporal data are obtained at low cost. In order to break through the limitation of single-frame detection, a space-time attention module is integrated in a detection network, and multi-frame information is aggregated to improve the robustness to motion blur and the like. The invention also provides a new normal form of space anomaly detection, which does not directly identify the missing seedlings, but learns a normal seedling space distribution rule, and identifies a low-density area as a seedling missing area by using a K-nearest neighbor (KNN) algorithm. According to the method, an unstable classification task is converted into a robust statistical problem, and the method has the advantages of low cost, high robustness and strong generalization.

The invention discloses an intelligent security and protection monitoring method, system and device and a storage medium, and the method comprises the steps: carrying out the cross-modal fusion analysis of a roughly extracted first visual feature and first radar feature of a target, and generating a dynamic fuzzy intensity coefficient based on an abnormal behavior result after the fusion analysis, so as to dynamically adjust the fuzzy intensity of a target face region; according to the change of the target dynamic fuzzy intensity coefficient, judging whether a high-risk early warning signal is output or not; if the high-risk early warning signal is output, cross-modal deep fusion analysis is carried out on the extracted fine second visual feature and second radar feature of the target, and a target abnormal behavior judgment result is output based on the features after deep fusion; and tracking the target based on the judgment result, associating a visual detection target with a radar tracking target when the target is visually shielded, predicting the future motion trajectory of the shielded target, and recovering visual tracking when the shielding is removed. Accurate detection of abnormal behaviors is realized.

The invention discloses a blurred image optimization enhancement system based on artificial intelligence, which relates to the field of blurred image optimization enhancement and comprises a data acquisition module, a primary processing module and a deep processing module. The method comprises the following steps: analyzing dynamic blurring, out-of-focus blurring and atmosphere blurring of a crop remote sensing image to obtain a dynamic blurring characteristic value of the crop remote sensing image, an out-of-focus blurring characteristic value of the crop remote sensing image and an atmosphere blurring characteristic value of the crop remote sensing image; the method comprises the following steps: respectively judging dynamic blurring, defocus blurring and atmosphere blurring states of a crop remote sensing image, carrying out corresponding optimization adjustment, judging a comprehensive pathological state of a crop to obtain a crop pathological heterogeneity value of each region, and carrying out grading processing on each region of the crop remote sensing image to obtain a crop pathological heterogeneity value of each region; and corresponding optimization enhancement measures are matched, so that interference of various factors in the acquisition process of the crop remote sensing image can be comprehensively considered, and the accuracy of crop growth monitoring is ensured.

The invention provides a scene reconstruction and automatic correction method using an unmanned aerial vehicle, and belongs to the technical field of scene reconstruction and correction, and the method comprises the steps: carrying out the global data collection of different sub-regions, synchronously recording the flight path information of each round, and carrying out the timestamp alignment processing of sensor data; constructing a dynamic error compensation model, and correcting motion blur, environmental interference and trajectory deviation errors in the collected data by combining a dynamic parameter set of random flight of each round of the unmanned aerial vehicle; integrating multiple rounds of data after correction, establishing spatial mapping of subarea textures and point clouds, and generating an initial three-dimensional model; and constructing a closed-loop correction mechanism based on the GNSS absolute coordinate of the unmanned aerial vehicle and the ground high-precision control point, calculating the deviation of the initial three-dimensional model, and if the deviation exceeds a threshold value, reversely adjusting the random flight coverage density, the flight round and the error compensation coefficient until the deviation reaches the standard. And the precision of scene reconstruction and correction is improved.

The invention discloses a writing process analysis method based on computer vision, which relates to the technical field of handwritten character recognition, and comprises the following steps: carrying out perspective transformation on real-time writing data, obtaining a writing image frame sequence, carrying out motion blur correction and pen point coordinate positioning on the writing image frame sequence by adopting an RAFT (Reversible Addition Fragmentation Transform) optical flow algorithm, and forming a writing track data flow; and performing curvature segmentation on the writing trace data stream to obtain discrete stroke segments, and performing spatial topological correlation and geometric structure mapping on the discrete stroke segments to generate a writing stroke topological graph. Through the RAFT optical flow algorithm and the stroke semantic analysis model, the precision of stroke recognition is improved, and efficient and accurate analysis from dynamic visual input to structured character output is achieved.

The invention discloses a dynamic fuzzy small target detection method and system, and aims to solve the problems of weak feature response, information loss and precision-efficiency imbalance of a convolutional neural network in small target detection in a dynamic fuzzy scene. The method comprises the following steps: extracting a multi-scale feature map through a convolutional neural network, constructing a feature pyramid, retaining edge details of a small target by using shallow high-resolution features, and combining semantic information of deep features; a self-adaptive space fusion mechanism is adopted, space weights of different levels of features are dynamically learned, feature conflicts of a fuzzy region are suppressed through weighted summation, and feature response of a small target is enhanced; a multi-scale channel attention module is further introduced, multi-scale context information is extracted in combination with convolution operations of different receptive fields, and target related channels are enhanced through channel-level weighting, so that background interference is reduced; and finally, outputting a target category and a position based on the fused and optimized feature map.

The invention relates to the technical field of panoramic images, and discloses a self-correcting panoramic image splicing method for a vehicle around view system, which comprises the following steps: acquiring dynamic data of a vehicle to obtain an environment characteristic dynamic state, establishing a space-time incidence matrix according to the environment characteristic dynamic state, adjusting a fuzzy kernel estimation parameter, and performing accurate estimation and multi-frame super-resolution reconstruction of motion blur; fusing dynamic data, defining a vehicle attitude feature matrix, dynamically correcting vehicle attitude and camera external parameters, unifying cross-camera colors, performing ground constraint, constructing a multi-objective optimization framework, and balancing splicing quality and real-time performance; through monocular depth estimation and geometric constraint, the perspective difference of a close-range object is optimized, and through an edge retention loss function, the depth precision at the boundary of the object is enhanced; monocular depth estimation is optimized through binocular parallax constraint; constructing a multi-scale feature fusion network, and recovering image details; a factor graph model is constructed to integrate historical data, and accumulative errors of long-time-sequence splicing are eliminated.