200 results about "Spatial perception" patented technology

The present disclosure provides a method for calibrating a humanoid robot, comprising obtaining a humanoid robot with original kinematic biasing values, controlling the humanoid robot through predetermined poses, capturing image data of body parts using vision sensors mounted on the humanoid robot while moving through the poses, determining revised kinematic biasing values by processing the image data using a bipedal spatial perception model trained using synthetic image data containing keypoints, and replacing the original kinematic biasing values with the revised kinematic biasing values. The bipedal spatial perception model processes captured image data to generate observed keypoint locations on robot components, which are compared with kinematic-based locations from joint encoder measurements to minimize discrepancies through optimization algorithms.

The application discloses a kind of based on triple perception learning enhanced visual language model's zero sample exception detection method and system, it is related to computer vision field, method includes: extracting global and local visual features from input image;Visual coding process is corrected local feature in deep network by spatial perception attention enhancement module, and fine-grained attribute text description is generated for abnormal visual feature;Through attribute perception guide module, attribute text description and general text prompt are deeply semantically aligned;The similarity of enhanced visual feature and optimized text feature is calculated, and pixel-level exception segmentation map is generated;Inference stage converts segmentation map into spatial attention weight by exception perception reconstruction module, and feedback is generated to visual encoder final global feature representation and calculates exception score.The method of the application significantly improves the accuracy and generalization ability of model in industrial defect detection and other scenarios without target domain training samples.

The invention relates to the technical field of deep learning, in particular to a small sample fine-grained image classification method based on multilayer feature dynamic interactive fusion, which comprises the following steps: inputting a support set and a query set into an image classification model, and outputting a prediction category of a query sample; the image classification model processing steps are as follows: S201, extracting to obtain multi-scale features; s202, obtaining interaction features after interaction optimization of each support sample through a multi-layer feature interaction module; s203, obtaining fusion features of each support sample through a dynamic feature fusion module; s204, projecting the fusion features to a corresponding feature space; s205, calculating category prototype representation of each category in each feature space; s206, calculating the cosine similarity between the query sample and the category prototype representation of each category in each feature space; and S207, determining a prediction category of the query sample based on the cosine similarity. According to the method, efficient cross-layer feature aggregation can be realized, and the capability of capturing fine-grained differences is enhanced through an attention mechanism of spatial perception.

The present disclosure relates to the technical field of audio control. Provided are a computer-implemented method for audio control, and an electronic device and a storage medium. The computer-implemented method for audio control comprises: on the basis of a specified measurement signal, performing acoustic effect testing on a target space, in order to obtain environmental response information, wherein the environmental response information comprises impulse response information and post-separation impulse response information; analyzing the environmental response information, in order to obtain acoustic characteristic information of the target space; on the basis of the acoustic characteristic information, performing compensation processing on audio to be processed, in order to obtain regulated audio; regulating gains of multiple channels corresponding to the regulated audio, in order to obtain re-regulated audio; and performing multi-channel collaborative optimization processing on the re-regulated audio, in order to obtain audio having a target effect. The method achieves precise compensation for spatial acoustic defects, resolves inter-channel crosstalk, improves the acoustic image localization and spatial perception, significantly improves the perceptual audio quality, and optimizes an output.

A humanoid robot system comprises vision sensors for capturing image data, a computing architecture with processing hardware and memory, and a bipedal spatial perception model. The model includes a feature extractor that extracts hierarchical feature maps from input images, a robot data module that detects robot parts, and a robot vector data module that calculates three-dimensional spatial position and orientation data for each detected robot part. The feature extractor uses a feature pyramid network generating multi-scale feature maps through bottom-up and top-down pathways with lateral connections. The robot vector data module predicts 2D-to-3D point correspondences and solves perspective-n-point problems to obtain final position and orientation vectors, enabling real-time robot self-awareness and closed-loop visual servoing for precise object interaction.

The invention discloses an unmanned aerial vehicle autonomous navigation system based on a rasterized world model, and relates to the technical field of unmanned aerial vehicle control. According to the system, observation information of an unmanned aerial vehicle is input into a trained rasterized world model through an observation acquisition module; through cooperative work of a sequence model, a multi-modal self-encoder, a hidden space dynamics predictor, a multi-modal information prediction head and a grid predictor in a rasterized world model, a cyclic variable containing historical information, a prediction hidden state vector and a prediction local grid map are provided for an agent model. And action decision making is carried out based on the data through an intelligent agent model so as to accurately output the actions of the unmanned aerial vehicle. According to the system, by introducing the rasterized world model, multi-modal observation information can be effectively fused, and a three-dimensional structure of a local environment is predicted in real time, so that the spatial perception capability of an intelligent agent model to a complex environment is enhanced, and the decision-making performance of the intelligent agent model is improved.

The invention discloses a stroke motion cognition evaluation system and method based on spatial perception nerve enhancement. The stroke motion cognition evaluation system comprises a data acquisition module, a preprocessing module, a feature extraction module, a weight coefficient generation module, a motion cognition evaluation module and a visualization module. Extracting information of different frequency bands of the brain through a feature extraction module, and obtaining power spectral density, a phase locking value and cross-frequency coupling features; meanwhile, a weight coefficient generation module is used for obtaining weight coefficients for performing weight fusion on different features, and the features extracted by a feature extraction module are fused through a motion cognition evaluation module to completely describe complex brain network interaction, so that cognition and spatial perception state evaluation of the subject in the motion imagination process is completed. In addition, threshold adjustment and personalized feedback are performed based on the real-time change of the evaluation result, so that the quality and efficiency of motor imagery training can be remarkably improved.

The invention relates to an unstructured road scene-oriented data generation and end-to-end sensing method, belongs to the technical field of automatic driving and intelligent traffic, solves the problem that the performance of a sensing algorithm in an unstructured scene is severely reduced in the prior art, and comprises the following steps: S1, generating a long-tail video sample based on a diffusion model, acquiring original data by adopting a multi-modal sensor data acquisition mode, and constructing a training data set by using the long-tail video sample and the original data; s2, establishing a multi-mode diffusion denoising network to carry out data denoising to obtain denoised image data and point cloud data; s3, multi-modal feature extraction and fusion are carried out, and enhanced point cloud BEV feature representation is obtained; s4, unifying BEV feature representation learning, and constructing BEV features of spatial perception; s5, designing a double-branch fusion architecture based on BEV features of spatial perception, and processing to obtain a perception result; and S6, outputting a unified sensing result, and providing the unified sensing result to a downstream path planning system for path planning.

The invention discloses a two-stage multi-modal 3D target detection method based on space-semantic-range multi-dimensional joint modeling, and the method comprises the steps: collecting the multi-modal data of a 3D target, fusing an RGB image and a LiDAR point cloud through an image-guided depth completion technology, and generating a virtual point cloud; frame labeling is carried out on the 3D target in the partial virtual point cloud to obtain full-labeling data, and the full-labeling data comprises a 3D bounding box surrounding the 3D target, a category label and direction information; constructing a dual-stage multi-modal 3D target detection model of dual-stage multi-dimensional joint modeling, wherein the model comprises a spatial perception sampling sub-module, a multi-dimensional importance score sampling sub-module, a sparse convolutional backbone network, a distance perception sub-network and a detection head; the full-annotation data and the virtual point cloud serve as training data, a two-stage multi-mode 3D target detection model is trained, then a 3D bounding box, a category label and direction information in the full-annotation data serve as true values, a loss function is calculated, detection head parameters are trained, and the trained model is the 3D target detection model; and inputting a virtual point cloud generated by performing depth completion on the RGB image and the LiDAR point cloud into the trained 3D target detection model to obtain a 3D bounding box, category information and direction information of the 3D target.

The invention discloses a robot action generation method integrating multi-layer feature bridging and world knowledge prediction. The method comprises the following steps: extracting multi-layer middle layer visual features and action query hidden variables by utilizing a pre-trained visual language model; an initialization strategy based on robot body sensing state guidance is adopted, real-time pose priori is injected for action query, and traditional all-zero initialization is replaced; a spatial perception vector gating mechanism is introduced into the bridging attention module, and fine-grained selection of specific image region features is achieved; meanwhile, a world knowledge prediction task is integrated, and physical common knowledge is enhanced through explicit modeling environment depth, semantics and a dynamic region; and finally, replacing a traditional L1 regression action head with a diffusion model architecture, and generating an optimal action sequence under multi-modal distribution through an iterative denoising process. According to the method, the operation precision of the robot in a long-range complex task is improved, and the problem of control failure caused by an action'averaging effect 'is solved while the light weight of the model is kept.

The invention discloses a scene graph and multi-granularity contrast learning fused multi-modal named entity recognition method. The method comprises the following steps: constructing a text scene graph; constructing a visual scene graph; the method comprises the following steps: dividing an input image into regions, constructing a visual region graph, performing feature enhancement of spatial perception, and obtaining a visual region feature set and a global semantic vector of the visual region graph; constructing a global visual representation through the global semantic vector of the visual scene graph and the global semantic vector of the visual area graph; realizing overall semantic alignment of input images and text sequence input through global comparative learning; fine-grained alignment of text object nodes and image local areas is realized through local comparative learning; performing cross-modal interaction processing on the text object feature set and the unified visual feature set; carrying out fusion processing on the cross-modal embedding representation generated by interaction, and then carrying out aggregation on the fused cross-modal embedding representation and a word-level text vector input by the text sequence to generate a final representation; and executing a named entity recognition task according to the final representation, and outputting a final entity recognition result.

The invention relates to the technical field of geological disaster early warning, in particular to a rainfall classification-based landslide disaster early warning method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the characteristics of each grid node through collection and grid processing of multi-source data such as geology and rainfall; segmenting rainfall events and extracting morphological and statistical features; machine learning rainfall typing is carried out based on the features, and probability vectors of all rainfall types are obtained; then, constructing a graph structure fusing a space and a geological relationship, and fusing static and dynamic characteristics of grid nodes; inputting the spatio-temporal characteristic graph into a graph neural network, taking a probability vector as a condition signal, adjusting attention weight through learnable mapping, and realizing adaptive spatial information aggregation guided by rainfall typing; and calculating a landslide probability and generating an early warning based on the updated grid node representation, so as to deeply couple rainfall typing and a graph neural network, realize the crossing from a static threshold value to dynamic feature modulation, and improve the early warning accuracy, timeliness and spatial perception capability.

The application discloses a multi-period water supplement evaluation method based on big data simulation, and comprises the following steps: taking the Yangtze River estuary and the downstream basin as an object, establishing a hydrodynamic salinity diffusion numerical model, and training a physically constrained Fourier neural operator proxy model; dividing spatial grid nodes to construct a spatial topology connection graph; establishing data interaction relationship between nodes and updating parameters in real time through an improved spatial perception Mamba network; calculating uncertainty based on the salinity simulation results output by the proxy model, and dynamically feeding back and optimizing the Mamba network; reversely transmitting the optimized prediction results to dynamically update the constraint conditions of the proxy model; and determining an ecological water supplement position, time and water quantity scheme. The application realizes high-precision and high-frequency ecological water supplement decision support.

The application discloses a kind of adaptive hearing compensation Bluetooth hearing aid optimization method and system, comprising: S1: collecting user individual hearing curve and the background audio data of left and right ear under each typical scene and respectively pre-processing;S2: extracting the time-frequency fusion feature of each typical scene, generate scene feature vector and construct scene feature vector library;S3: calculate scene perception conversion vector, mutation risk vector, smooth gain vector;S4: calculate binaural gain change vector, binaural synchronization entropy vector, phase alignment factor vector and left and right ear collaborative gain vector;S5: according to left and right ear collaborative gain vector, the background audio data of left and right ear pre-processed under current scene is handled, and output to Bluetooth hearing aid.The application can solve the gain mutation problem generated when existing Bluetooth hearing aid switches scene and the problem of spatial perception distortion caused by binaural collaborative disorder.

This disclosure provides a model training method and apparatus, an autonomous driving method and system, and a mobile device. The model training method includes: acquiring a first BEV feature obtained by a spatial perception module based on a multi-view image and a second BEV feature encoding at a first time moment; performing 3D reconstruction based on the first BEV feature to obtain a first 3D reconstruction result; iteratively training the spatial perception module based on a first training sample including the first 3D reconstruction result and a first reference point cloud; acquiring at least one fifth BEV feature corresponding to at least one fifth time moment predicted by a future decoding module based on a third BEV feature and motion information of the second mobile device at at least one fifth time moment; performing 3D reconstruction based on the at least one fifth BEV feature to obtain at least one second 3D reconstruction result; and iteratively training the future decoding module based on a second training sample including at least one second 3D reconstruction result and at least one second reference point cloud.

The application discloses an implementation method and system of an intelligent earphone interrupt function based on a positioning technology. The method comprises the following steps: establishing a connection between an intelligent earphone and a mobile host terminal and synchronizing a clock, and determining a relative position of multiple parties in space; calculating a distance difference by receiving sound data, and solving three-dimensional coordinates and distance information of a sound source based on a hyperbolic positioning principle; determining an interrupt priority value in combination with a semantic analysis result of the sound data, and sending an intelligent reminder to a user according to the interrupt priority value. The application realizes accurate spatial perception and intelligent importance judgment of surrounding sound events, enables the intelligent earphone to actively provide an interrupt type reminder at a critical moment, and improves the intelligence and practicability of human-computer interaction.

This invention relates to a calibration method for the 0-angle and limit angles of a forklift steering wheel, belonging to the technical field of all-day operation methods in the steel manufacturing industry. The technical solution of this invention is as follows: A map is constructed using a single-steering-wheel forklift in a test scenario; the position coordinates of the forklift during its movement are collected, along with the parameters corresponding to the steering wheel rotation angle output by the position system; the forklift's trajectory is fitted based on the collected position coordinates during its movement, forming new forklift trajectory generation points; a model function is constructed, and the new forklift trajectory generation points are substituted to calculate the 0-angle position of the forklift steering wheel; the limit rotation radius is determined by least-squares fitting, and the limit angles on both sides are calculated. The advantages of this invention are: it eliminates the need for additional calibration tool templates, demanding calibration environments, and complex spatial perception algorithms; the calibration method is simple, has low requirements for the calibration environment, low calibration costs, and is easy to implement in mass production on assembly lines within a workshop.

The invention discloses a flight personnel space orientation capability assessment method based on holographic eye movement characteristics, and belongs to the technical field of flight personnel capability assessment. Extracting multi-dimensional eye movement characteristics such as pupil contraction / expansion tracks, iris stability parameters and proportional change rate; aligning and matching the eye movement characteristics with real-time flight attitude and environmental stimulation data; carrying out collaborative analysis on the features, and generating a fusion eye movement feature vector insensitive to ambient light interference; and finally, inputting the spatial orientation capability into a pre-trained spatial orientation capability evaluation model, and outputting quantitative evaluation values of spatial perception acuity, environmental adaptability and stress response level. According to the method, the problems that an existing evaluation method is high in subjectivity, prone to being interfered by the environment and difficult to comprehensively reflect the complex physiological cognitive state of the pilot are solved, and more objective, stable and comprehensive quantitative analysis on the spatial orientation capacity is achieved.

This invention discloses a visual BEV perception method based on implicit and explicit dual-path height feature collaboration, comprising: real-time acquisition of raw image data from multiple perspectives around the vehicle via an onboard surround-view system, followed by inputting the data into an image encoding network for feature extraction to obtain multi-scale feature images; processing the multi-scale feature images through a vertical height feature modeling unit and a horizontal feature modeling unit to obtain a BEV feature representation containing three-dimensional spatial information; wherein the vertical height feature modeling unit adopts a dual-path parallel architecture with explicit height modeling branches and implicit height modeling branches; and inputting the BEV feature representation into a target detection output module to obtain the target detection result. This invention, without relying on external depth sensors such as LiDAR, effectively improves the three-dimensional spatial perception of pure vision systems through visual input, significantly enhancing the ability to express vertical dimension features, thereby achieving more accurate and reliable three-dimensional target detection in complex scenes.

The application provides a spatial perception enhancement and pathological image classification method for processing class imbalance, aiming to enhance the model's perception of spatial information and process class imbalance. The specific steps are as follows: in the preprocessing stage, the image is segmented, the background is filtered out, and the image block is cut, and the coordinates are recorded; in the feature extraction and fusion stage, the pathological features are extracted through a pre-trained convolutional model, then the coordinates are normalized to form position features, and the pathological features and position features are fused to form comprehensive features with spatial information. In the reasoning stage, a multi-scale feature enhancement network is introduced, which can capture local features through convolution operations and model spatial relationships through attention mechanisms, thus achieving more comprehensive processing of local and spatial information. Finally, the parallel classifier is used to solve the problems of class imbalance and sample shortage, and the final pathological classification result is obtained. This method can better assist pathologists in improving the diagnosis accuracy and reducing subjective differences in the diagnosis of chondroma.

The application provides a city resident population distribution trend prediction method based on a geographic detector, comprising the following steps: acquiring panoramic street view data of a research area; performing semantic segmentation to obtain visual perception elements and establish visual perception factors; establishing spatial perception factors through the accessibility of different facility types under different travel modes; using the geographic detector to calculate the explanation rate of the visual perception factors and the spatial perception factors on the resident population distribution density; obtaining several high-explanation-rate perception factors, obtaining corresponding weights based on a judgment matrix, establishing a resident intention index, and predicting the city resident population distribution trend. The prediction method is based on the perception of people to the real city environment, takes into account the subjectivity and objectivity of evaluation, and has high reference value for city planning related to city resident population distribution.

The invention relates to the technical field of space perception monitoring, in particular to an unmanned aerial vehicle countering analysis method fusing radar-photoelectric data. The method comprises the following steps: performing cross-modal coupling sensing feature analysis of unmanned aerial vehicle target monitoring on airspace unmanned aerial vehicle multi-source monitoring data, and generating unmanned aerial vehicle target monitoring coupling sensing feature data; performing unmanned aerial vehicle flight behavior propagation feature analysis based on the unmanned aerial vehicle target monitoring coupling sensing feature data to generate unmanned aerial vehicle flight behavior propagation feature data; performing unmanned aerial vehicle threat propagation field feature analysis based on the unmanned aerial vehicle flight behavior propagation feature data to generate unmanned aerial vehicle threat propagation field feature data; and performing intelligent control strategy analysis of unmanned aerial vehicle countering based on the unmanned aerial vehicle threat propagation field feature data, and generating unmanned aerial vehicle countering intelligent control strategy data. According to the invention, through fusion analysis of radar monitoring data and photoelectric monitoring data, an accurate and efficient unmanned aerial vehicle intelligent countering strategy is realized.

The application discloses a FAST core array broadband beam effect elimination method based on deep learning, relates to the technical field of radio astronomy image processing, and constructs an end-to-end image restoration system, which comprises the following five steps: acquiring an observation image containing a broadband beam effect; extracting multi-scale features and introducing a spatial perception mechanism; performing feature frequency domain enhancement processing and multi-scale context fusion; and outputting a multi-scale prediction image and a restored image from which the broadband beam effect is removed. The application realizes the function of improving imaging quality by constructing a deep learning model, avoids cumulative errors caused by step-by-step processing, can effectively remove broadband beam distortion, and makes the reconstructed image more close to the real sky brightness distribution in overall fidelity and structural details, and has higher physical consistency.

The invention discloses a surface detection method based on feature selection and a parallel interactive attention mechanism, and belongs to the technical field of image detection. Comprising the following steps that 1, a feature map needing to be detected is sent to a DFS module, the DFS module screens out a feature channel subset most sensitive to abnormity, then the feature channel subset is mapped to a unified space through a feature adapter, dimensionality reduction is conducted, and an adaptive feature map is obtained; 2, the adaptive feature map is sent to a PIA module, the PIA module processes the adaptive feature map through a global context branch and a local detail branch, then outputs of the global context branch and the local detail branch interact, and finally an enhanced feature map is output through residual connection; and 3, testing the enhanced feature map by adopting an exception decision technology to obtain an exception detection result. According to the method, a channel more sensitive to defects can be found, multi-scale feature enhancement is realized, and the spatial perception capability of the model to an abnormal region is enhanced.

This invention discloses a method, device, and storage medium for directional device identification and control based on vision and spatial perception, belonging to the field of intelligent control and human-computer interaction technology. The method is applied to portable terminals, achieving "point-and-shoot" by using a coaxially integrated laser and camera module to acquire target images containing laser pointer spots. To address the identification challenges caused by differences in target object size and distance, the method proposes multi-scale adaptive region extraction, generating and matching multiple candidate regions centered on the laser pointer spot. Simultaneously, to eliminate control ambiguities for similar devices in different rooms, the method innovatively utilizes smart home device nodes in known locations within the environment as Bluetooth beacons, achieving zero-deployment-cost collaborative spatial perception and obtaining the terminal's precise spatial location. Finally, by jointly querying and adjudicating the visual recognition results and spatial location information, the target device is uniquely identified and control commands are generated. This invention achieves intuitive interaction with zero learning costs, solving the problems of complexity and confusion in traditional control methods. It has advantages such as accurate identification, fast response, simple deployment, and privacy security, making it particularly suitable for smart home and smart elderly care scenarios.

The present application relates to a kind of forest leaf instance segmentation method and system based on deep learning, method includes: obtaining vegetation image, vegetation image is input into leaf instance segmentation model, obtains leaf instance segmentation prediction result;Leaf instance segmentation model is trained using training set;Training set includes: vegetation original image;Feature extraction and enhancement are carried out using backbone module in leaf instance segmentation model, and adaptive spatial fusion mechanism in progressive feature pyramid network is integrated to dynamically adjust feature weight, generate dynamic fusion feature;Through the dynamic asymmetric spatial perception mechanism built-in in dynamic anomaly regression head module, the corresponding multi-source deformation feature layer of dynamic fusion feature is obtained, and the feature fusion strategy of top-down cascaded decoding module is used to optimize multi-scale feature, obtain multi-source fusion feature layer, further using multi-source fusion feature layer, generate leaf instance segmentation prediction result.The present application solves the problems of data scarcity, poor adaptability and low efficiency.

The invention discloses a Mongolian handwriting recognition method based on double-branch feature fusion and Transform sequence recognition, and the method comprises the steps: obtaining a Mongolian handwriting image data set, and carrying out the multi-level data enhancement and data preprocessing; constructing a Mongolian handwriting recognition model based on combination of double-branch feature fusion and Transform sequence recognition, extracting global features and detail features of the preprocessed data by using global branches and detail branches, and performing dynamic weighted fusion based on sample characteristics by using a space attention module; and based on strategies of dynamic position coding, space perception sequence conversion and attention deviation optimization, carrying out Transform coding and decoding sequence identification according to the fusion features. The Mongolian handwriting recognition precision and efficiency can be improved through double-branch feature fusion, and a systematic scheme is provided for relieving the technical bottlenecks that in Mongolian handwriting recognition, data set samples are deficient, conjunctions are dense, and similar characters are confused and difficult to recognize.

The invention provides a nasal tracheal catheter and a nasopharyngeal cavity three-dimensional sensing method. The method comprises the following steps: generating a geometric three-dimensional surface model of a nasopharyngeal cavity; associating pressure information of the outer wall of the front end of the catheter to a nearest coordinate point area on the geometric three-dimensional surface model, and marking on the geometric three-dimensional surface model through a preset threshold value to obtain a plurality of high-pressure contact points; determining the extension strength in the surface normal direction of the geometric three-dimensional surface model by taking each high-voltage contact point as the center; performing curved surface reconstruction on all the high-voltage contact points based on the extension strengths corresponding to different high-voltage contact points to obtain a safety boundary curved surface enveloped outside the geometric three-dimensional surface model; and performing spatial position judgment on a short-term predicted movement track of the catheter and the safe boundary curved surface, and generating a three-dimensional sensing path of the catheter in a safe passing area in the nasopharyngeal cavity according to a judgment result. By the adoption of the scheme, three-dimensional spatial perception of interaction force between the catheter and nasopharyngeal cavity tissue can be achieved.

This invention discloses an augmented reality interactive system and method for electric vehicle charging, belonging to the field of intelligent electric vehicles and augmented reality technology. It includes: an AR display terminal module for receiving graphics rendering instructions from a processor, generating a virtual information layer with correct perspective and occlusion relationships, and fusing it with optical or video images from the real world; a multi-source data fusion and communication module for establishing and maintaining data channels, and achieving information aggregation through a vehicle communication unit, a charging pile communication unit, and a cloud service interface; a spatial perception and registration module integrated into the AR display terminal for achieving virtual-real fusion; a natural human-computer interaction module providing diverse and natural control methods; and a central processing and rendering engine module deployed locally on the AR terminal or on an edge / cloud server. This invention enables spatialized and contextualized presentation of information, enhancing user freedom, operational intuitiveness, and the overall charging experience.

This invention discloses an integrated space-air-ground remote sensing monitoring method and system for eco-hydrological elements, belonging to the field of remote sensing monitoring technology. The method includes: acquiring a slope aspect layer and comparing slope abrupt changes; extracting fracture boundary segments; grouping directional changes to screen stable paths; extracting humidity cycle markers for abrupt changes; tracking NDVI growth paths to generate response sequences; mapping and reconnecting layer boundaries to form an updated path set. This invention enhances the spatial perception of topographic boundary changes by guiding slope abrupt changes through the dominant runoff direction; groups paths according to directional vector change trends to improve the continuity and stability of boundary extraction; combines trend offset blocks in the humidity cycle for time annotation to strengthen temporal comparison during the humidity response process; constructs vegetation response chains by continuously splicing NDVI growth paths; and uses path mapping to drive dynamic adjustments to the remote sensing boundary structure, making the boundary update process more closely aligned with the spatiotemporal progression characteristics of ecological disturbances.