38results about How to "Reduce ambiguity" patented technology

Embodiments of the present application relate to the technical field of image processing, and particularly relate to a contour fusion method, an electronic device and a storage medium. Embodiments of the present application determine the parent-child relationship between multiple contours, directionally constrain the contour points of the parent contour and the child contour, and determine the fusion direction, thereby fusing the nested multiple contours describing the same target region into a complete contour, eliminating the understanding ambiguity of the target region caused by the nested labeling of multiple contours, avoiding the misrecognition of the same defect region as multiple independent target regions by the segmentation network, improving the accuracy of the training data labeling, making the loss value calculation in the supervised training process more accurate, and significantly improving the convergence effect, generalization ability, and detection precision and robustness of the network model on tin hole defects.

This application relates to the field of code generation technology, and discloses a software code generation method and its apparatus, device, and medium. The method includes: responding to a code generation driving event and determining the multimodal information corresponding to the event, wherein the multimodal information corresponds to the original software development requirements; semantically aligning the different modal information in the multimodal information and fusing them to generate software function planning information; extracting each function planning unit as a code generation subtask of the corresponding code generation type according to the code generation type corresponding to the function planning unit described in the software function planning information; calling a large language model to execute each code generation subtask and synchronously triggering the corresponding coding constraint instructions to generate target code. This application, through the semantic alignment and fusion of multimodal information and the synchronous triggering mechanism of coding constraints in the code generation process, highly restores the development requirements of the software code and reduces the subsequent security audit costs and software maintenance risks of the software code.

The application discloses a kind of semantic boundary guided robot operation scene point cloud segmentation methods. Firstly, the three-dimensional point cloud of robot operation scene is obtained;Then, the semantic boundary point in the scene three-dimensional point cloud is predicted using point cloud learning network;The scene three-dimensional point cloud is input into point cloud Transformer model for feature extraction, and the semantic boundary point is used as the extended key point of self-attention mechanism in Transformer model, wherein for the point feature of each boundary region, a contrast learning loss function is added to reduce the semantic ambiguity of boundary point feature;Finally, based on the point feature of each point, the semantic class of each point is predicted to obtain the point cloud semantic segmentation result. The application enhances the learned point feature by introducing semantic boundary information in the point cloud feature extraction process, and improves the misclassification problem caused by boundary ambiguous features, effectively improves the point cloud segmentation quality of robot operation scene, and has good engineering practical value.

The present application belongs to the technical field of geotechnical engineering, and specifically relates to a tunnel underpassing a site adverse geological early warning method and system. A multi-source detection network is arranged in a set region of tunnel construction, original data reflecting geological abnormal conditions are obtained and preprocessed, a multi-dimensional feature index set including wave speed, dielectric constant, rock mass integrity coefficient, etc. is obtained, a three-dimensional perspective model containing spatial distribution and physical properties of adverse geological bodies is generated through a data fusion algorithm, and a geological anomaly grade G is determined accordingly. The structure vulnerability grade V of a target historical site is obtained, and a disturbance influence coefficient S of the site by current construction operation is calculated based on construction parameters. The geological anomaly grade G, the site vulnerability grade V and the construction influence coefficient S are input into a preset risk coupling evaluation model, a comprehensive risk value R is dynamically calculated, and corresponding graded early warning is triggered according to the threshold interval of the R value.

The application discloses a method and system for calculating the physical bit number of SSB of a 5G NR network, and belongs to the technical field of mobile communication. The method comprises the following steps: determining the time domain distribution pattern and the possible position of the starting time domain symbol of SSB according to the subcarrier spacing, carrier frequency, duplex mode and spectrum sharing access scene of SSB, then determining the number of candidate SSB indexes, and combining the high-level protocol parameter ssb-PositionsInBurst to calculate the number of time domain symbols occupied by the actual transmission of SSB; determining the number of frequency domain subcarriers occupied by SSB according to the offset of the starting subcarrier of SSB and the starting subcarrier of the common resource block 0; and finally calculating the total number of time-frequency resource elements and multiplying the modulation order to obtain the physical bit number of SSB. The application can quickly and accurately calculate the physical bit number of SSB based on the physical layer parameters without implementing a complex protocol signal processing algorithm, and significantly improves the efficiency of SSB data generation and test work.

This invention discloses a pedestrian detection method and system based on head center point-assisted feature comparison. The method includes: constructing a new CSP network architecture by adding a head branch, a feature comparison branch, a detection confidence enhancer, and a head-aware NMS module to the original CSP network architecture; constructing a training sample set, where the head center point is used as the positive sample for the head branch, and non-head center points within the head bounding box are used as the negative sample for the head branch; using the training sample set as input to the CSP network for training to obtain a pedestrian detection network model; supervising network learning during the training phase using head branch loss and feature comparison loss; and enhancing the detection results during the training and testing phases using the detection confidence enhancer and the head-aware NMS method. This invention improves the detection accuracy of CSP and reduces the probability of false positives and false negatives by introducing head center point, feature comparison, detection confidence enhancer, and head-aware NMS methods.

The invention belongs to the technical field of computer vision, and relates to a VSLAM feature point matching method and system based on semantic topology constraint, and the method comprises the following steps: generating a current feature point set containing a preliminary semantic tag and a semantic recognition result; generating an initial candidate matching pool; generating a dynamic geometric check threshold for each current feature point; performing forward filtering on the initial candidate matching pool by applying a dynamic geometric verification threshold, and performing initial round matching based on appearance description similarity to generate a high-potential pre-matching point pair list; the geometric structure consistency of the high-potential pre-matching point pair list is analyzed, and geometric structure confidence scores are generated; performing reverse calibration and confirmation on the high-potential pre-matching point pair list to generate a final matching result; and generating a semantic tag correction instruction according to the reverse calibration process. According to the method, the problem that adopted geometric correction is generally based on a fixed fault-tolerant threshold and lacks self-adaptive adjustment capability for credibility of different scenes and feature points is solved.

This application relates to a method for generating character consistency images based on multi-level LoRA training, comprising: performing low-rank adaptation training on multiple preset low-level generation models based on facial feature images and their corresponding structured facial feature semantic labels, generating facial feature LoRA weights that match each low-level generation model; loading the corresponding facial feature LoRA weights through a routing mechanism to generate a reference face image of the target character and extracting the target character's identity identifier; obtaining a multi-view extended map generated based on the reference face image, performing data augmentation on the reference face image and the multi-view extended map to construct a character training set; using the character training set to perform low-rank adaptation training on the low-level generation models to obtain character LoRA weights; and generating a consistency image of the target character based on the character LoRA weights. This method addresses the problems of character identity being easily affected by the inference environment and multi-view identity degradation in existing technologies, improving the stability of cross-view character consistency image generation.

The invention relates to the field of fry counting, in particular to an automatic fry identifying, tracking and counting method based on machine vision. An image enhancement module is introduced before detection and hardware is combined to highlight the overall features of the fry; a CGAF module is introduced into a YOLOv11 network model in the detection and recognition stage, multi-source features are adaptively fused through a multi-attention mechanism, an original detection head is replaced with a dynamic detection head DynamicHead with a self-attention mechanism, the detection and counting precision is improved, and the ambiguity of tracking and matching is remarkably reduced; in the tracking stage, a detection result of a current frame is subjected to time sequence verification through a DeepSORT algorithm by using a historical track, corrected information is fed back to a YOLOv11 network model in real time for training feedback, a closed loop of'detection guidance tracking and tracking supervision detection 'is finally formed, and line passing counting is triggered based on a stable track, so that the detection accuracy is improved. Complex scenes such as rolling, light reflection, shielding, winding and the like are effectively handled, and high-flux accurate counting is achieved.

The application relates to the technical field of oil and gas exploration and development, and is a deep reservoir quality prediction method based on rock facies-geochemical coupling, which comprises the following steps: obtaining geochemical parameters and physical property parameters of a layer section to be predicted; taking the geochemical parameters and the physical property parameters of the layer section to be predicted as inputs, inputting a reservoir quality coupling prediction model, and outputting a porosity calculation value and a reservoir quality grade prediction result of the layer section to be predicted; and the reservoir quality coupling prediction model comprises a multivariate regression model related to the geochemical parameters, the physical property parameters and the porosity and a reservoir quality grade classification model. The application eliminates the multi-solution property of a traditional reservoir quality prediction method from a geological mechanism by means of strong coupling modeling of rock facies and geochemical parameters, and significantly improves the deep and super-deep reservoir quality prediction accuracy. Compared with the prior art, the application can significantly improve the drilling rate of high-quality reservoirs.

The application discloses a method for removing gypsolith based on model forward, and relates to the technical field of oil and natural gas seismic exploration, which comprises the following steps: a data preparation step, a lithology calculation step, a model design step, a model forward step, a matching comparison step and a planar lithology distribution analysis step. The application uses the velocity and density of actual drilling and logging data to make a forward geological model, continuously adjusts the parameters related to the model to obtain different forward results, matches and compares the forward results with actual seismic data, determines the seismic reflection characteristics of different lithologies or lithology combinations, and then combines amplitude attribute analysis to predict the distribution law of different lithologies. The method fully utilizes the actual drilling and logging data, simulates the reflection characteristics of different lithologies, and the obtained result is more in line with geological laws. The application uses model forward to simulate the seismic reflection characteristics of different lithologies and lithology combinations, and can make up for the deficiency of people in the seismic wave propagation law in complex medium.

The application discloses a kind of based on the superdeep layer carbonate rock hillock beach body delineation method of frequency division attribute, the present technical comprehensive application of innovation is used in this method, by the full combination of geological model and seismic prediction method, using the constraint of geological prediction model, reduce the multi-solution of seismic prediction means, solve the limitation of single geology or seismic prediction method, improve the prediction accuracy of hillock beach body plane distribution, can be better used in the exploration and development of hillock beach body reservoir;Traditional method focuses on the conventional attribute delineation in full-band data body, and the prediction accuracy of hillock beach body is limited, and the frequency division attribute based on high-frequency and low-frequency information of seismic data can finely delineate the boundary range of hillock beach body, and also can effectively improve the delineation accuracy of hillock beach body details.

The present specification provides a determination method and device of a multi-task inversion model and a multi-task elastic parameter inversion method. The determination method of the multi-task inversion model comprises: inputting pre-stack seismic data samples into the multi-task inversion model to obtain a plurality of different elastic parameter data; inputting the plurality of different elastic parameter data into a forward model to obtain predicted pre-stack seismic data; determining a pre-stack seismic loss according to the pre-stack seismic data samples and the predicted pre-stack seismic data; determining an elastic parameter loss according to label data corresponding to the plurality of different elastic parameter data and the pre-stack seismic data samples; adjusting parameters of the multi-task inversion model and the forward model according to the pre-stack seismic loss and the elastic parameter loss; wherein the adjusted multi-task inversion model is used to determine a plurality of different target elastic parameter data according to target pre-stack seismic data. The above method can improve the prediction accuracy and efficiency of different elastic parameter data.

The image description method, device, equipment and medium of the present application, wherein the method comprises: acquiring an input image; encoding the region features of the input image to obtain region feature encoding. Encoding the grid features of the input image to obtain grid feature encoding. Fusing the region features and the grid features according to the region feature encoding and the grid feature encoding to obtain image features. Fusing the image features and the preprocessed historical words according to text adaptability features to obtain image-text fusion results. Classifying the image-text fusion results to obtain image description results. By complementing the region features and the grid features, the image features can cover more image content. By guiding the fusion process of the image features and the preprocessed historical words through the text adaptability features, the image-text fusion results can be more accurate, thereby reducing the ambiguity of the image description results.

The present application relates to the technical field of construction surrounding rock stability evaluation, and discloses a tunnel surrounding rock stability evaluation method and system based on multi-physical field parameter inversion, aiming to solve the problems of insufficient data synergy, strong parameter inversion multi-solution, and poor accuracy and timeliness of surrounding rock stability evaluation of the existing method, and the scheme mainly comprises: arranging an acousto-optic-electromagnetic seismic drill multi-modal sensing array and implementing time-space synchronization; establishing an interaction entropy spectrum density model to realize multi-physical field collaborative excitation and collection; obtaining a unified feature vector through data correction, feature extraction and weighted fusion; constructing a joint inversion objective function embedded with rock physical constraints, inverting to obtain a three-dimensional physical property parameter field, and combining with acoustic emission energy to calculate a dynamic permeability field; finally, using ensemble Kalman filtering to dynamically update the model, and based on the updated parameters and seepage-uncertainty coupling correction to obtain the final risk probability. The present application improves the accuracy, real-time performance and reliability of deep-buried tunnel surrounding rock stability evaluation.

The invention discloses a hybrid architecture semantic restoration method and device based on a DFA and a large language model, and relates to the technical field of semantic restoration, the method comprises the following steps: receiving an original power text sequence, loading a terminology benchmark library, and carrying out full scanning based on the DFA to output a candidate anchor point set; locking the candidate anchor point set, extracting a context feature window of each anchor point, and calculating a semantic coupling probability of each anchor point and a context in the context feature window; determining a repair path between each anchor point and a context in the context feature window based on the semantic coupling probability, and performing semantic repair on each anchor point and the context based on the repair path to obtain a corresponding standardized term fragment; and integrating the standardized term fragments to form a complete standardized text sequence corresponding to the original power text sequence. According to the invention, the problem of low accuracy and efficiency during conversion from a spoken text to a standardized term in the prior art is solved.

The application discloses a gravity and magnetic inversion method based on grid merging and BTTB matrix fast forward, relates to the technical field of gravity and magnetic exploration, and comprises the following steps: establishing a basic forward and inversion grid and a basic grid kernel function matrix; initializing a coarse grid model and a conversion matrix; coarse grid inversion; adaptive grid subdivision based on a model gradient and establishment of a new conversion matrix; fine grid inversion; iterative convergence and output. The application has the beneficial effects that the BTTB kernel function matrix is calculated based on the basic grid in advance, the conversion from fine scale model parameters to coarse scale model parameters is realized through the conversion matrix, the iterative closed loop process of coarse grid inversion, adaptive grid subdivision and fine grid inversion is adopted, the criterion of adaptive grid subdivision is to compare the gradient eigenvalue of the current inversion model parameters with a preset threshold value, and in the grid inversion of different scales, the BTTB kernel function matrix generated by the same set of basic grid in advance is used to realize fast forward through the processing of the conversion matrix.

This application relates to the field of computer technology, providing a data processing method and related apparatus to improve the accuracy of traffic analysis. The method includes: acquiring target network traffic; generating target prompt content for the target network traffic using a target prompt template; inputting the target prompt content into a trained content extraction model to obtain target key content, which represents the content in the target network traffic that conforms to the characteristics of abnormal traffic; and finally, performing traffic analysis on the target network traffic based on the target key content to obtain traffic analysis results. Utilizing the model's generalization ability, abnormal traffic can be accurately detected when facing new network attacks, thereby triggering alerts for abnormal traffic.

The invention belongs to the technical field of ocean geophysical exploration, and particularly relates to a deep sea hydrothermal sulfide detection method and system based on a frequency domain electromagnetic method, and the method comprises the steps: driving a transmitting coil to excite a medium to generate a secondary electromagnetic response; at multiple different depths in the drill hole, electromagnetic response signals are collected through an electromagnetic receiving sensing system, and magnetic induction intensity-depth data are formed; a consistent inversion depth division mode is adopted, one-dimensional electrical inversion is carried out respectively, and a conductivity model of dielectric electrical parameters changing along with depth under each emission frequency and mutually registered in space is obtained; based on the conductivity corresponding to each measurement depth, obtaining a polarizability parameter representing the polarization characteristic of the medium, and generating a polarizability profile curve in which the polarizability changes along with the depth; comprehensively judging the occurrence position of the sulfide ore body according to the abnormal threshold value and the conductivity characteristic in the polarizability profile curve, and accurately describing the boundary depth and thickness information of the ore body.

The application discloses an image segmentation method based on a multi-scale space adaptive hole convolution, and comprises the following steps: a multi-scale space adaptive hole convolution model is established, which comprises a hole convolution module, a space scale attention module and a feature recombination module, wherein the hole convolution module comprises n parallel hole convolution layers of different scales, is used for performing convolution on an input feature map to generate n groups of first feature maps with the same size, the space scale attention module is used for synchronously generating m*n channel attention maps according to the input feature map, and the feature recombination module is used for recombining the corresponding first feature maps and the attention maps to obtain a final output result; and the image segmentation network model is used for performing image segmentation on the final output result to obtain a segmentation result. According to the method, the feature semantics extracted by the convolution is more explicit, unnecessary interference information can be removed, and ambiguity is reduced, so that a more accurate segmentation result can be obtained.

The application discloses a digital human modeling method, comprising the following steps: S1, three-dimensional reconstruction data about each frame of image is extracted from monocular video, the three-dimensional reconstruction data comprises original data, high-precision human mask, original audio signal, optimized SMPL parameters and camera parameters corresponding to each frame of image; S2, a three-dimensional structure human body grid in a "T" posture is generated based on the optimized SMPL parameters, after being discretized into a 3D Gaussian point set, the initialization of the 3D Gaussian point set is completed in combination with the original data, the high-precision human mask and the camera parameters corresponding to each frame of image; S3, a facial animation sequence capable of representing mouth shape and emotional dynamics is obtained based on the original audio signal, a text description of an expected facial morphology and a reference image; S4, a Gaussian physical model is added in a 3DGS process to render the initialized 3D Gaussian point set, and the 3DGS rendering result is mixed and rendered with the facial animation sequence to generate a digital human model with facial expressions.

The application provides a kind of side slope fissure passive source seismic wave imaging method, the patent adopts multi-wave multi-component seismograph, simultaneously gathers multiple seismic wave signals on three components, including surface wave, body wave and the like.This design significantly improves the richness of data, so that in data processing, a variety of seismic wave signals can be comprehensively utilized for inversion analysis.By joint processing and constraint of different wave types, the multi-solution problem in geophysical inversion can be effectively reduced, and the stability and reliability of the results are improved.In addition, the acquisition of multi-component data makes the imaging more comprehensive, especially in the characterization of fissure directionality and geometric features, which has obvious advantages.

The invention relates to a multi-modal data-based intelligent VLA data pre-labeling method and device, and the method comprises the steps: obtaining a multi-modal data stream containing visual modal data, language modal data and action modal data, and carrying out the time sequence segmentation of the multi-modal data stream according to a time sequence, and obtaining a plurality of time period data units; for each time period data unit, identifying a current task execution state based on the visual modal data and the language modal data, and generating at least one candidate labeling hierarchy set corresponding to the state according to the task execution state; performing multi-modal semantic representation on each level label in the candidate label level set, and constructing a multi-modal consistency relation graph containing visual nodes, language nodes and action nodes; and executing a consistency propagation operation in the multi-modal consistency relation graph to obtain a multi-modal consistency constraint relation among the nodes. According to the scheme, the stability and reliability of multi-modal labeling in a complex scene can be improved.

The application relates to the technical field of language translation, and particularly discloses a multilingual real-time translation cash register terminal system, which comprises an interactive language recognition module, a translation and ambiguity content recognition module, and an alternative translation screening module. The interactive language recognition module judges the interactive language of a customer based on the first interactive voice data input by the customer. The translation and ambiguity content recognition module translates the interactive voice data input by the customer or a cashier in real time and determines all ambiguous words, all ambiguous sentences and all alternative translation results in the corresponding interactive voice data based on the interactive language of the customer or the interactive language of the cashier. The alternative translation screening module screens the final translation result of the real-time input interactive voice data from all the alternative translation results based on the adjacent context of all the ambiguous words, the adjacent context of all the ambiguous sentences and a preset cash register scenario multi-word meaning weight information base. The application improves the communication efficiency and accuracy in the cash register process and provides more convenient and smooth service experience for the customer and the cashier.

The present application relates to a kind of by floating slurry layer thickness fast evaluation new pouring concrete segregation degree method, belong to concrete quality evaluation field, comprising: the close packing density of coarse aggregate is calculated;The total mass of coarse aggregate in single test piece is calculated;Limit segregation packing volume and limit segregation coarse aggregate height are calculated;The total thickness of concrete in test area is measured, and the thickness of limit segregation state mortar layer is calculated;Limit segregation ratio is defined;Select multiple point positions in test area to carry out actual measurement segregation ratio calculation, take average value to obtain final segregation index I;I is graded, and divided into two grades of qualified and unqualified grade.The present application breaks through the limitation of traditional qualitative evaluation, based on the close packing characteristics of coarse aggregate and the characteristics of test piece component, through multi-step logical deduction, the segregation degree is converted into the segregation index that can be accurately calculated, the subjectivity and ambiguity of naked eye observation are eliminated, the quantitative characterization of segregation state is realized, and the evaluation result is more objective and scientific.

The invention discloses a photovoltaic power generation station construction progress early warning method and system based on a high-resolution remote sensing image, and relates to the technical field of remote sensing and geographic information. A high-resolution satellite and an unmanned aerial vehicle cooperatively collect images, and a multi-task convolutional neural network synchronously executes semantic segmentation and target detection to extract a construction area coverage rate and equipment installation state features; fusing macroscopic attributes and spatial features, screening high-similarity historical items by using weighted cosine similarity to generate weighted historical progress data, dynamically adjusting data weights based on an exponential decay function, inputting the data into a double-LSTM network to construct a time sequence prediction model, and performing iterative updating; and generating multi-dimensional progress information and resource allocation suggestions, comparing real-time and predicted progress detection deviations, and associating historical data to trace reasons to generate early warning. According to the method, the full-cycle progress is accurately predicted, the small ground feature recognition precision is improved, the resource allocation response is accelerated, the construction period deviation is reduced by 15%-20%, and decision support is provided for early intervention.

The application discloses an indoor positioning method and system fusing WiFi fingerprints and space information, the method realizes online positioning through offline construction of a fingerprint map. Wherein, WISP features fusing WiFi features and space features are proposed, the space features are classified and coded according to whether reference points are in straight corridors or corners. In addition, the application also constructs a bag-of-words tree, that is, the bag-of-words tree is used to represent the correlation between reference points, so that in the online positioning stage, firstly, the WISP features of the current position and the bag-of-words tree are used to calculate the similarity between the WISP features of the current position and the WISP features under each cluster; then the product of the similarity and the corresponding TF-IDF weight value is calculated, finally, the cluster with the maximum product value is selected, and the position corresponding to the WISP features under the cluster is used to deduce the final positioning result, effectively reducing the influence degree of the perception confusion of WiFi signals.

The invention relates to a passive source signal acquisition method, system and device for oil-gas exploration and a medium, and belongs to the technical field of oil-gas exploration, and the passive source signal acquisition method for oil-gas exploration comprises the following steps: selecting a work area position where an active source is deployed and acquired as an observation area for oil-gas exploration; deploying a passive source consistent with the acquisition deployment of the active source in the observation area, wherein the passive source is used for acquiring a passive source signal; and constructing a passive source observation system consistent with the observation system of the active source, wherein the passive source observation system is used for processing and recording the passive source signal. According to the invention, passive source oil-gas exploration and active source seismic signal data are combined for use, and can be used as signal supplement and expansion of an active seismic source, so that the identification precision of stratums and faults is effectively improved, and the multiplicity of solutions of a target geologic body is reduced. And data guarantee is provided for ultra-deep layer oil-gas exploration well position optimization and exploration field decision making.

The present application belongs to the technical field of geophysical exploration, and relates to a method and system for improving seismic data resolution through frequency extrapolation, comprising: collecting a seismic signal, and representing the seismic signal by a reflection coefficient sequence and a seismic wavelet; fitting the reflection coefficient sequence to obtain a plurality of candidate fitting functions, selecting a function having the least influence on the reflection coefficient sequence from the fitting functions, and calculating an entropy norm of the reflection coefficient sequence according to the function having the least influence on the reflection coefficient sequence; converting the reflection coefficient sequence into a reflection coefficient in a frequency domain, performing frequency extrapolation on the reflection coefficient in the frequency domain through the function having the least influence on the reflection coefficient sequence, and solving an optimization problem through iteration; making the reflection coefficient obtained through the frequency extrapolation satisfy a coherence constraint condition, and outputting the frequency extrapolation reflection coefficient meeting the coherence constraint condition. The present application can effectively improve the seismic data resolution while maintaining the continuity of seismic events.

The present application belongs to the technical field of radar signal processing and target identification, and specifically provides a multi-target array structure inversion method based on two-dimensional power spectrum imaging, to solve the problem that the array structure in the multi-target scattering image is difficult to be stably recovered under the condition of strong aliasing, strong speckle and strong coherent interference background; the present application converts the collected radar scattering data into a power spectrum image, generates a global candidate point set with a level label through log power spectrum construction, sub-aperture decomposition, hierarchical candidate point extraction, consistency voting and density clustering deduplication, and further forms a closed loop through translation and combination prediction and bidirectional matching verification, and the original multi-target array structure is inferred from geometric evidence, which has good robustness, engineering feasibility and identification accuracy; and the present application does not depend on phase compensation, is suitable for observation scenes under multi-target, and meets the application requirements of complex target structure identification, multi-target imaging identification and target classification.