84results about How to "Strong complementarity" patented technology

The invention provides a target identification method and system based on multi-source information fusion, and relates to the technical field of low-altitude target detection. The method comprises the following steps: acquiring target radar track data, interception equipment track data and position area information data; based on target radar track data, extracting a first feature in an RCS form dimension, and extracting a second feature in a motion dimension; based on the target radar track data and the track data of the monitoring equipment, determining a frequency spectrum monitoring correlation factor and regional position information features; performing feature fusion on the first feature, the second feature, the spectrum interception correlation factor and the regional position information feature to obtain a target feature, and identifying a target type; and identifying a target threat level based on the target type and the target radar track data. The method and the device are used in a target identification process based on multi-source information fusion, and the technical problem that the target threat degree cannot be accurately identified in a complex environment in the prior art is solved.

The invention discloses a wind field detection method, device and equipment based on wind measurement data fusion and a medium, and relates to the technical field of wind measurement, and the method comprises the steps: carrying out the quality control of initial coherent wind measurement data and initial direct wind measurement data, and obtaining target coherent wind measurement data and target direct wind measurement data which meet preset quality requirements; performing space-time consistency matching on the target coherent wind measurement data and the target direct wind measurement data to obtain a target matching result, and splicing the target direct wind measurement data to the target coherent wind measurement data according to the target matching result to obtain spliced wind measurement data; performing data fusion on data in a to-be-fused height range in the spliced wind measurement data by using a real-time fitting prediction method to obtain fused wind measurement data, and determining a target wind field detection result by using the fused wind measurement data; the to-be-fused height range is a height range determined based on the maximum value and the minimum value of the height corresponding to the target coherent wind measurement data in the fusion time sliding window; therefore, the accuracy of wind measurement data can be improved.

The invention relates to the technical field of computer vision, and discloses a camouflage target detection method based on cascade frequency domain perception and refinement feature guidance, comprising the following steps: a computer device obtains a camouflage image and extracts a multi-scale feature map; performing frequency domain separation by using a frequency domain sensing module, generating low-frequency and high-frequency characteristic sub-bands, and performing cross-hierarchy aggregation from bottom to top through a cascaded frequency domain sensor to obtain frequency fusion characteristics; receiving the frequency fusion features from top to bottom by using an aggregation guide decoding unit, performing guide refinement in combination with the deep decoding features, adapting to irregular edges by using a variable kernel convolution strategy in the refinement process, and generating a prediction feature map in combination with a partial convolution strategy; and outputting the predicted feature map as a detection result. According to the invention, through frequency domain information complementation and dynamic convolution refinement, the problem of difficult detection caused by highly similar texture of the camouflage target and the background is solved, and the positioning accuracy and edge segmentation integrity of the camouflage target are improved.

The invention discloses a coke residue feature detection method and system based on image and pressure alignment, and solves the problems that the subjectivity is high, the accuracy is difficult to guarantee and the physical and psychological health of detection personnel is influenced when coke residue feature detection is manually carried out in the prior art. The method comprises the steps of collecting a coke residue image and pressure data, performing preprocessing to form image input and pressure input, and constructing sample data; a neural network model is constructed, image features of image input are extracted by the image coding layer, pressure features of pressure input are extracted by the pressure coding layer, and the image features and the pressure features are fused and then are subjected to coke residue classification through a classifier; constructing a model comprehensive loss function; inputting sample data for training to obtain a coke residue classification model; and deploying the model, and inputting the to-be-detected coke residue image and the pressure data into the model to obtain a coke residue classification result. According to the method, the neural network is fully utilized, the image and the pressure are effectively fused, information complementation and integration are realized, and the classification detection precision of the coke residues is effectively improved.

This application belongs to the field of computer vision technology and discloses a cross-source point cloud registration method based on adversarial neural networks. The method includes: inputting a multi-frame LiDAR point cloud sequence and a UAV video sequence; segmenting each frame of the LiDAR point cloud into ground points and non-ground points and registering them to obtain a global LiDAR point cloud; performing motion recovery structure reconstruction on the UAV video sequence to generate a dense color point cloud; inputting the global LiDAR point cloud and the SfM dense color point cloud into an MMAlignNet network to construct a shared latent feature space; solving for the rigid transformation matrix from the LiDAR point cloud to the SfM dense color point cloud; and aligning the cross-modal point clouds according to the rigid transformation matrix to obtain a multimodal 3D reconstruction result. This application improves the registration accuracy, completeness, and robustness of large-scale outdoor environment 3D reconstruction by fusing UAV imagery and mobile LiDAR data, combined with cross-modal feature alignment and geometric constraint optimization methods.

The invention discloses a depression detection method based on EEG channel selection and multi-dimensional feature fusion, which avoids the problem of sensitivity of k-means to the clustering number and initial clustering centers, improves the k-means, determines the number of clustering centers by adopting a Karlinski-Halabass criterion, calculates the PageRank value of an EEG channel based on maximum and minimum similarity, and finally determines the number of the clustering centers by adopting the Karlinski-Halabass criterion and the PageRank value of the EEG channel based on the maximum and minimum similarity. Selecting the first k channels as initial clustering centers, designing a self-adaptive threshold model, and selecting channels with distances smaller than a threshold as key EEG channels; constructing a brain function network and a super-brain function network by using the key EEG channel, and extracting low-dimensional time domain features of the key EEG channel, low-dimensional spatial domain features of the brain function network and high-dimensional spatial domain features of the super-brain function network; and designing a multi-dimensional feature fusion strategy based on standard deviation, and fusing low-dimensional time domain features, low-dimensional spatial domain features and high-dimensional spatial domain features to realize depression detection with high accuracy.

The invention discloses a document-level relation extraction method and system based on knowledge enhancement and evidence guidance, and belongs to the technical field of natural language processing and information extraction. According to the invention, three core modules are mainly used for cooperative work: a document graph enhancement module is used for constructing a hierarchical heterogeneous graph and fusing co-reference analysis to enhance semantic representation; the knowledge enhancement module introduces an external knowledge base and adopts a confidence coefficient filtering mechanism to reduce knowledge noise; the evidence guidance reasoning module realizes multi-hop reasoning through axial attention and evidence supervision, and solves the technical problems of decentralized modeling of reasoning capability, large knowledge integration noise, insufficient evidence guidance and limited long-range dependence capture capability in the existing method. Experiments show that the method can effectively capture inter-sentence dependence, suppress knowledge noise and improve multi-hop reasoning stability, and can be widely applied to scenes such as knowledge graph construction, intelligent question and answer and information retrieval.

The application discloses an emotion recognition method based on online cross-modal knowledge distillation, comprising the following steps: acquiring electroencephalogram and electrocardiogram original signals and windowing and cutting; constructing electroencephalogram and electrocardiogram student models, extracting intermediate features from each modal data segment through an encoder, and obtaining non-normalized prediction output through a classifier; constructing a teacher probability distribution through a joint encoder fusion; introducing adaptive contrast loss to align the cross-modal intermediate features, introducing distillation loss to constrain the prediction probability distribution of each modal to align with the teacher probability distribution; synchronously optimizing new student model parameters through online collaborative training; and performing actual inference prediction based on the student model after training. The application combines double modal signals to make up for the defects of single modal information, excavates the complementarity of modes, realizes dynamic generation of teacher supervision signals and real-time collaborative learning of modes through online distillation, does not increase test calculation overhead, effectively improves the recognition accuracy, model robustness and generalization ability, and has good application prospect.

The invention discloses a blueberry tree branch segmentation method and system, and the method comprises the steps: carrying out the multi-scale feature extraction of a to-be-segmented blueberry tree image, and obtaining the feature representation containing low-layer details, middle-layer semantics and high-layer context; information complementation and semantic alignment of deep and shallow layer features are realized through a bidirectional cross-level information interaction module, and fusion features are optimized by using a space and channel attention mechanism to obtain enhanced features; and performing up-sampling and dynamic reconstruction on the enhanced features by adopting a frequency domain dynamic convolution module, refining branch edges and texture details, and outputting a high-resolution pixel-level segmentation result. The method effectively solves the problems of fuzzy boundary, fracture adhesion and background interference in the segmentation of the slender branches and trunks of the blueberries, and has good precision and generalization ability.

The invention discloses a hyperspectral image super-resolution optimization method based on a fused three-branch network, and belongs to the field of hyperspectral image processing, and the method comprises the steps: firstly obtaining and preprocessing training data, collecting LR-HSI, HR-MSI and corresponding high-resolution hyperspectral truth values, and completing normalization, data enhancement and cross-modal feature dimension matching and interaction; constructing a three-branch network model, inputting the LR-HSI into an HSI branch to extract spectral features, and inputting the HR-MSI into an MSI branch to extract spatial features; performing residual calculation and noise suppression on the spectral and spatial features, inputting the spectral and spatial features into a fusion branch, generating compensation features through convolution and jump connection, fusing three-branch output through learnable weight, generating a fusion image, and training a model through an optimizer. According to the method, a residual difference decoupling fusion strategy is introduced, and directional repair and enhancement are respectively carried out on missing space information in a hyperspectral image and insufficient spectral information in a multispectral image.

The application discloses a kind of gas diffusion layer and its preparation method and application, it is related to fuel cell technical field.The gas diffusion layer includes conductive substrate layer and hydrophobic filling layer, multiple channels are opened in conductive substrate layer and pass through its thickness, multiple channels include first channel and second channel different in aperture, and the pore wall surface of channel has hydrophilicity, hydrophobic filling layer is filled in channel and covers the surface of one side of conductive substrate layer, and hydrophobic filling layer has hydrophobicity, wherein, the inner wall surface of channel and the gap between hydrophobic filling layer filled in channel are formed.The present application utilizes the wettability gradient formed by hydrophilic pore wall and hydrophobic filling layer, combines interface gap and multi-level pore array structure, realizes the spatial complementation of liquid water directional discharge and reaction gas efficient permeation, can solve the performance attenuation problem caused by fuel cell cathode waterlogging under high current density, is conducive to improving water management capability and battery output performance.

The invention provides a signal peptide category and cleavage site prediction method and system based on multi-modal characteristics, and belongs to the technical field of biological information analysis. The method comprises the following steps: acquiring an amino acid sequence of a signal peptide sample, and acquiring three-dimensional structure data of the signal peptide by utilizing a protein structure prediction model; obtaining sequence modal input data of the amino acid sequence of the signal peptide, and constructing a structural diagram to obtain structural modal input data; inputting the sequence modal input data into a sequence encoder and a protein language model, and extracting sequence features; inputting the structural modal input data into a structural encoder, and extracting structural features through graph convolution operation; carrying out fusion processing on the sequence features and the structural features to obtain multi-modal feature representation; and outputting a category prediction result and a cleavage site prediction result of the signal peptide through a prediction module. According to the invention, through fusion of the sequence and the structure information, the accuracy of signal peptide prediction and the recognition capability of minority class samples are improved.

This invention provides a method and system for constructing a negative obstacle risk map, relating to the field of robot environmental perception technology. The method includes: acquiring multimodal sensing data of a legged robot in the current operating environment to obtain the types of negative obstacles, multiple candidate regions, and the corresponding confidence and uncertainty of each candidate region; using a risk diffusion algorithm, performing ink blurring processing based on the type, confidence, and uncertainty to generate a continuous risk field corresponding to each candidate region; dividing the continuous risk field to generate a continuous risk gradient distribution corresponding to each candidate region; performing terrain modeling based on laser point clouds, updating the continuous risk gradient distribution using Bayesian fusion to obtain a multi-level grid map, and then performing tactile closed-loop correction on the multi-level grid map to generate a risk cost map. This invention improves the accuracy of constructing a negative obstacle risk map.

This application discloses an auxiliary classification system and method based on multi-view spatiotemporal interaction and difference compensation, relating to the field of auxiliary classification. The method includes: acquiring time-series data from multiple brain regions and calculating Pearson correlation matrices and partial correlation matrices; constructing first and second-dimensional time matrices based on features extracted from the time-series data; generating multiple spatial connectivity matrices based on the Pearson correlation matrix and partial correlation matrix, thereby constructing first and second-dimensional spatial matrices; performing bidirectional cross-attention interaction on the time and spatial matrices of the two dimensions respectively to obtain corresponding graph structure representations and node representations; generating compensation terms based on the differences between the graph structure representations and node representations of the two dimensions; enhancing the graph structure representation and node representation of the second dimension using the compensation terms to obtain fused features; and inputting the fused features into a graph convolutional network to output classification results. This invention improves the accuracy of auxiliary classification through two-dimensional spatiotemporal interaction and difference compensation.

This invention provides a method for identifying the authenticity of wheat flour based on the fusion of Raman and near-infrared spectroscopy, belonging to the field of wheat flour authenticity identification technology. The method includes: acquiring Raman and near-infrared spectral data of the wheat flour sample to be tested, and preprocessing them separately; constructing and training a fusion detection model, which includes a data-level fusion module, a feature-level fusion module, and a decision-level fusion module; the data-level fusion module generates full-spectrum fusion data; the feature-level fusion module concatenates core features into a comprehensive feature set; the decision-level fusion module inputs the comprehensive feature set into a hybrid model; and the trained fusion detection model processes the comprehensive feature set to output the authenticity identification result of the wheat flour sample to be tested. This invention, through a three-level spectral fusion architecture and a dynamic adaptive adversarial mechanism, can effectively achieve high-precision and high-robust identification of wheat flour authenticity.

The invention provides a deep counterfeit face detection method fusing time sequence consistency and space consistency, and belongs to the field of counterfeit detection, and the method comprises the steps: obtaining a target video; for each video frame containing the face in the target video, executing a space-time fusion feature extraction operation to obtain a space-time fusion feature corresponding to the video frame; determining whether the target video is a forged video based on the space-time fusion feature corresponding to each video frame; wherein the space-time fusion feature extraction operation comprises the steps of determining a space consistency feature corresponding to the video frame based on the video frame, and determining a time sequence consistency feature corresponding to the video frame based on the video frame and a previous video frame of the video frame; and performing space-time fusion by taking the space consistency feature corresponding to the video frame as a query vector and taking the time consistency feature corresponding to the video frame as a key vector and a value vector to obtain a space-time fusion feature corresponding to the video frame. According to the invention, the accuracy and reliability of deep fake face detection can be improved.

The invention relates to the technical field of industrial robot load identification, in particular to a six-degree-of-freedom robot end load parameter identification method, which comprises the following steps: establishing a six-degree-of-freedom series robot dynamical model, reconstructing a load minimum parameter set, and determining that identification can be completed only by exciting a third joint, a fifth joint and a sixth joint. Designing four groups of symmetric excitation trajectories combined by Fourier series and quintic polynomials, and performing discrete sampling to generate an observation matrix; optimizing a load parameter set and a regression matrix, and reducing the condition number of an observation matrix; a load minimum parameter set is obtained through difference calculation of the no-load identification result and the on-load identification result; and based on the parameter set, solving the mass, the mass center coordinate, the inertia moment and the inertia product of the load. The method solves the ill-conditioned problem of a regression matrix by reconstructing the minimum parameter set, reduces the number of excitation joints, does not need to depend on a robot body kinetic model, and has the advantages of high identification precision, high stability, low experiment complexity and high efficiency.

This application provides a machine vision-based welding quality inspection method and apparatus, belonging to the field of machine vision and intelligent inspection technology. The method includes: acquiring a target image to be inspected, wherein the target image is an image containing a printed circuit board; inputting the target image to be inspected into a defect detection model for defect detection, and obtaining a defect detection result, which is used to characterize the defect type of the printed circuit board. The machine vision-based welding quality inspection method and apparatus provided in this application can improve the accuracy of welding defect detection.

This utility model relates to the field of integrated energy system technology and discloses a multi-port adjustable photovoltaic-storage-DC-flexible integrated energy power supply system. It includes a 750V DC bus platform, a photovoltaic power generation unit, a grid input terminal, a V2G charging pile, an energy storage unit, and a control cabinet cluster. The photovoltaic power generation unit includes a photovoltaic array, an MPPT controller, and a photovoltaic DC combiner box with a DC 750V output. The photovoltaic power generation unit is electrically connected to the 750V DC bus platform. The control cabinet cluster includes control cabinet one, control cabinet two, and control cabinet three. This utility model, through a multi-port DC power router architecture combined with intelligent control strategies, reduces AC / DC conversion links, significantly reducing energy loss, improving the overall energy utilization rate of the system, achieving energy complementarity and optimized configuration, enhancing the system's power supply capacity in emergency situations, and improving the reliability and stability of the power supply.

The invention discloses a colorless-phase black master batch, a colorless-phase black fiber and a preparation method of the colorless-phase black master batch, and belongs to the technical field of high polymer material coloring. Aiming at the defects of'dead black 'and'splicing black' technical color cast of the existing carbon black colored product, the master batch comprises a polyamide carrier, a coloring system (carbon black, phthalocyanine blue, permanent violet RL and dye blue 60) and a polymeric dispersant; master batches are prepared through twin-screw extrusion (nine-temperature-zone gradient heating), and the master batches and a carrier are further subjected to composite spinning to obtain colorless-phase black fibers. By means of spectrum complementary absorption of a coloring system, a value a and a value b of a product approach to zero (a is smaller than or equal to 0.1, and b is smaller than or equal to 0.1), an L value is 18-20, and neutral hue and high blackness are achieved; the dye blue 60 endows the transparent layering sense, the process is stable, the adaptability is high, and the method is suitable for the field of high-end products.

This invention provides a method and apparatus for rapid and non-destructive digital reconstruction of cultural relics and ancient buildings. The reconstruction method includes: acquiring an image set of the target object and corresponding depth data; performing quality improvement processing on the depth data and performing pose calculation on the image set to obtain pose data; identifying different feature regions on the surface of the target object based on the image features of the image set, and assigning corresponding fusion weights to the processed depth data according to the characteristics of each feature region; using the image set, the weighted depth data, and the pose data as input data, performing three-dimensional reconstruction of the target object through a reconstruction model, and completing scale calibration during the reconstruction process; and outputting a scale-calibrated three-dimensional model of the target object. The technical problem solved by this invention is that limitations in the performance indicators of object acquisition and processing make it difficult to coordinate and optimize the speed, accuracy, and direct usability of the overall reconstruction method, resulting in less than ideal digital reconstruction results.

This invention relates to the field of image steganalysis defense technology, and particularly to an active image steganalysis defense method and system based on signal enhancement. The method involves adding interference noise to the image to be processed based on signal texture features to enhance the stegana signal. The enhanced image is then input into a pre-trained active image steganalysis defense network. This network restores the stegana signal features and recovers the original carrier image from the processed image through an inverse difference operation. The active image steganalysis defense network employs a dual-channel parallel network model to reconstruct the stegana signal distribution by mining the correlation and spatial relationships between the stegana signals. This invention does not require knowledge of the steganalysis algorithm type and embedding rate. Through image noise addition and neural network modeling, it achieves dual destruction of secret information in the image under a heterogeneous balance state, while simultaneously restoring the quality of the original image.

The invention discloses an emotion recognition method based on online cross-modal knowledge distillation. The emotion recognition method comprises the steps that electroencephalogram and electrocardio original signals are obtained and subjected to window segmentation; constructing an electroencephalogram and electrocardio student model, extracting intermediate features from each modal data segment through an encoder, and obtaining non-normalized prediction output through a classifier; teacher probability distribution is constructed through joint encoder fusion; self-adaptive comparison loss alignment cross-modal intermediate features are introduced, and distillation loss is introduced to constrain prediction probability distribution of each modal to align to teacher probability distribution; new student model parameters are optimized synchronously through online cooperative training; and performing actual reasoning prediction based on a student model after training. According to the method, bimodal signals are combined to make up single-modal information defects, modal complementarity is mined, teacher supervision signal dynamic generation and modal real-time collaborative learning are realized through online distillation, test calculation overhead is not increased, identification precision, model robustness and generalization ability are effectively improved, and the application prospect is good.

The invention relates to the technical field of blueberry wine brewing, in particular to a mixed bacteria fermentation blueberry wine brewing process and application thereof, and the mixed bacteria fermentation blueberry wine brewing process comprises the following steps: raw material pretreatment: unfreezing, picking and crushing blueberries to obtain blueberry pulp; performing enzymolysis: adding potassium metabisulfite and pectinase into the blueberry pulp, and performing enzymolysis at 48-52 DEG C for 1-2 hours; sugar adjustment: adjusting the sugar degree of the blueberry pulp to 14-22Bx, and adjusting the pH value to 3.5-4.5; inoculating yeast: inoculating activated mixed yeast strains into the blended blueberry pulp, wherein the mixed yeast strains are RW angel wine yeast and 71B yeast; fermentation and post-treatment: fermenting the inoculated blueberry pulp with residues, filtering after fermentation to obtain the blueberry fruit wine, and through synergistic fermentation of RW angel wine yeast and 71B yeast, generating aroma substances such as esters and alcohols richer than that of single-bacterium fermentation, so that the wine body has strong fruit aroma, mellow wine aroma and soft fermentation aroma, is full in layering sense, and has a good taste. The problems that traditional fruit wine is thin in taste and serious in homogenization are effectively solved.

The embodiment of the utility model discloses a heat pump system. The heat pump system comprises indoor heat exchange equipment, a heat pump device, auxiliary heat exchange equipment and a valve assembly. The heat pump device communicates with a water inlet of the indoor heat exchange equipment through a first water supply pipeline and communicates with a water outlet of the indoor heat exchange equipment through a first water return pipeline. The auxiliary heat exchange equipment is communicated to the first water supply pipeline through a second water supply pipeline and is communicated to the first water return pipeline through a second water return pipeline; the valve assembly is arranged on the second water supply pipeline and / or the second water return pipeline and used for adjusting the flow of water flowing from the second water return pipeline to the second water supply pipeline through the auxiliary heat exchange equipment. According to the embodiment of the invention, heat energy complementation of the heat pump device and the auxiliary heat exchange equipment can be realized, so that a better heat energy effect can be realized.

The application belongs to the field of intelligent medical computer-aided diagnosis application, and relates to a colon polyp segmentation method based on deep learning fusion and collaborative learning enhancement. The method comprises a feature extraction model, a fusion module and multi-view collaborative learning. The feature extraction model is divided into two branches. One branch uses DeiT-Small to extract global feature information and establish the correlation between each pixel. The other branch uses HardNet-MSEG to extract local feature information and obtain more low-level detail information. In order to improve the segmentation accuracy of small target colon polyp images, based on the public colon polyp image dataset and the initial deep learning single-branch segmentation method, a deep learning technology fusion method is proposed, and multi-view collaborative learning is used to enhance colon polyp segmentation. Compared with the feature extracted by a single deep learning method before improvement, the feature is richer, and the information omission defect of a single branch is compensated.

The invention discloses a cooperative control method for a floating type photovoltaic-hybrid energy storage system, and the method employs a flexible connection and semi-tensioning mooring design, and is suitable for a 0-degree head wave long-period working condition. Improving maximum power tracking control, and inhibiting power point oscillation based on a partition variable step size and a prediction power algorithm; hybrid energy storage power is distributed through a low-pass filter, a storage battery and a super capacitor cooperatively process high / low frequency fluctuation, and the service life is prolonged by combining charge partition management; a virtual synchronous generator VSG adaptive control strategy is provided, an improved particle swarm optimization algorithm is adopted to optimize a J / D initial value, and frequency fluctuation is dynamically suppressed; an optical storage grid-connected system is established for verification, and the VSG technology enables the power frequency fluctuation to be quickly stabilized, thereby guaranteeing the stable operation of the system.

The invention provides a method for identifying physiological features of an animal, imaging equipment and a medium. The method comprises the following steps: acquiring image data; identifying an animal target in the image data and a first physiological feature of the animal target; acquiring three-dimensional size data of the animal target; and according to the three-dimensional size data and the first physiological feature, obtaining the physiological feature of the animal target. According to the method, the three-dimensional size data of the animal target is obtained by using the distance measurement tool, a reliable physical scale basis can be provided, the inherent limitation of pure visual analysis is broken through by combining the first physiological feature based on image recognition and the three-dimensional size data obtained by distance measurement, and information complementation is realized. The three-dimensional size data obtained through distance measurement can effectively assist and optimize the physiological features obtained according to the image data, and the accuracy of animal target physiological feature recognition is remarkably improved.

The invention discloses a no-reference image quality evaluation method based on a prediction error graph, which is suitable for the field of image processing, realizes the evaluation of image quality through a deep learning model, and comprises the following steps: constructing a prediction error graph pre-training model based on a Transform encoder and decoder structure; inputting the distorted image into a prediction error graph pre-training model to generate a corresponding prediction error graph: constructing a stepped feature extraction network based on decomposed large kernel convolution, performing layer-by-layer feature extraction on the distorted image, obtaining multi-scale features, and aggregating the multi-scale features to form global features; carrying out element-by-element fusion on the extracted image features and the features of the prediction error graph; and finally, the fused features are mapped into image quality scores. According to the method, the image distortion region and the degradation mode thereof are described by introducing the prediction error graph, and the prediction error graph is combined with the multi-scale features, so that the model can fully utilize the image degradation information, and the accuracy and the stability of no-reference image quality evaluation are improved.