24results about How to "Improve target recognition ability" patented technology

The invention relates to a method for removing a mountain shadow in an optical remote sensing image. The removing method comprises the following steps: determining a shadow region and a non-shadow region in the optical remote sensing image according to spectral data of different wavebands; acquiring first spectral data of various ground objects in a non-shadow area, and establishing a non-shadow area spectrum library; according to the similarity between each second spectrum data in the shadow area and the first spectrum data in the non-shadow area spectrum library, obtaining reference spectrum data in the non-shadow area spectrum library, and respectively generating new spectrum data corresponding to the second spectrum data of each pixel in the shadow area; and generating a new shadow area according to the new spectral data, and merging to obtain a shadow-free remote sensing image. The shadow-free remote sensing image is obtained by removing the shadow, so that the image is clearer and brighter, the target recognition capability is enhanced, the information extraction precision is improved, and the application of the remote sensing data is more accurate and reliable.

The invention belongs to the technical field of sonar image processing, and relates to a target identification method based on sonar image segmentation and a computing device.The target identification method based on sonar image segmentation comprises the following steps that 1) a data set of existing sonar images is acquired; 2) building a target identification model composed of an encoder and a decoder; 3) inputting the data set into the target recognition model from the encoder, and performing initial training on the target recognition model; 4) performing sample augmentation on categories with a small number of samples in the data set to obtain an augmented feature domain data set; 5) using the augmented feature domain data set to carry out fine tuning training on the target identification model; and 6) after the target identification model is finely adjusted and trained, inputting a newly obtained sonar image into the target identification model to obtain a segmentation image of the target. The method can achieve the accurate recognition of a target.

The invention provides an inland ship target detection and identification method based on deep learning, and the method comprises the steps: building a multi-layer shallow information network through an improved YOLOv8 algorithm, adjusting a detection layer, adding a shallow detection head, capturing the features of a small target and the semantic information of the small target more finely, enhancing the target identification capability in a targeted manner, and improving the detection efficiency. The function of conveniently and interactively detecting and identifying the inland ship target is realized. According to the method, an EfficientNetV2 lightweight network optimization strategy is combined, a model network structure is improved, and a more effective loss function is adopted, so that the calculation efficiency and precision are improved, and the model training and reasoning efficiency is guaranteed. A dynamic non-monotonic focusing mechanism is adopted to ensure that the anchor frame is more fit with a real bounding box model, and the overall performance of the detector is improved. According to the method, the problems that the detection effect of the small target is not ideal, the detection anchor frame and the actual boundary are not tightly attached, and the system interaction is insufficient in a complex water surface environment are solved.

The invention discloses a millimeter wave radar and monocular camera fused information acquisition device and method, and relates to the technical field of target detection. The device comprises a millimeter wave radar, a monocular camera and an adjusting bracket, wherein the adjusting bracket can adjust the positions and angles of the radar and the camera. And the data fusion module performs time synchronization and space alignment on the radar information and the image information, and obtains a target recognition result through a preset target recognition model. The acquisition method comprises the steps of timestamp matching, time compensation, joint calibration, coordinate transformation, image distortion correction and the like. Further, radar physical features are extracted by using a CFAR algorithm, image features are extracted by using a deep learning model, and effective features are obtained through PCA dimension reduction. And finally, an accurate target recognition result is obtained through feature splicing and optimization by using a support vector machine (SVM). According to the method, efficient fusion and target identification of millimeter wave radar and monocular camera information are realized, and the method has a wide application prospect.

The invention discloses an unmanned aerial vehicle target tracking method based on feature fusion and online template updating, and the method comprises the following steps: constructing a deep network model based on a Resnet residual network and cavity convolution, and effectively extracting target features and enhancing the effective receptive field of the features; meanwhile, detail features of a shallow network and semantic features of a deep network are efficiently fused, and the expression ability of the features is enhanced; a template branch and a detection branch of a Siamese structure are formed by a ResNet-50 network and a target fusion network, two branch feature maps are sent to a cascade cross-correlation module, and a target position is determined; and constructing a template library, calculating a response graph score, and carrying out online updating on template branches according to a threshold value. According to the method, the tracking success rate and the tracking accuracy both reach a high level, and the target tracking performance of the unmanned aerial vehicle is effectively improved.

The invention discloses a metric learning method and system for intelligent detection and recognition of small sample ship targets, particularly relates to the field of SAR image target recognition, and combines multi-scale features with feature vectors generated by metric learning, thereby improving the detection precision of different-scale targets. In order to solve the problem of unbalanced sampling of positive and negative samples, a dynamic weight control resampling strategy is adopted, the sampling probability is dynamically adjusted in the training process, oversampling of minority classes is gradually reduced, and sampling of majority classes is increased at the same time; the method is obviously superior to a baseline YOLOv5 model in the aspects of robustness and detection accuracy, and has good applicability and practical value.

The invention relates to the technical field of image processing, and discloses an image acquisition method and device based on intelligent high-speed machine vision, and a storage medium. The method comprises the following steps: performing cache allocation processing on trigger time sequence data to obtain memory pre-allocation data; performing region division processing on the trigger time sequence data and the memory pre-allocation data to obtain multi-region configuration data, and performing line acquisition processing on the multi-region configuration data to obtain original multi-region image data; performing row enhancement processing on the original multi-region image data to obtain enhanced image data, and performing spectrum fusion processing on the enhanced image data to obtain multi-spectrum composite image data; and performing feature extraction processing on the multispectral composite image data to obtain multi-level feature data, and performing hierarchical compression processing on the multi-level feature data to obtain compressed feature data and feature index data. According to the invention, the image acquisition efficiency based on intelligent high-speed machine vision is improved.

The application discloses a kind of intelligent park target detection methods and systems for incomplete image.The method comprises: inputting first image into teacher model, obtaining teacher model target detection result, teacher model intermediate feature;Mask image corresponding to the first image is input into student model, and teacher model target detection result, student model intermediate feature are obtained;The teacher model intermediate feature, the student model intermediate feature are extracted respectively, and teacher model network feature map, student model network feature map are obtained;Detection task loss value is calculated;Global feature alignment loss value is calculated;The student model is optimized according to the detection task loss value, the global feature alignment loss value, the teacher model is updated, and new teacher model is obtained;Repeat training until reaching first preset condition, and obtain trained teacher model.The method has the characteristics of high target detection accuracy when facing incomplete image.

The invention provides an on-rail robot inspection image processing method, and relates to the field of image enhancement. According to the electric power system defect and fault image recognition and segmentation process, defect image data of a wire, an insulator and a pole under different environment conditions are collected, a filtering enhancement module is constructed, noise in an input image is eliminated through pooling and up-sampling, and beneficial features are enhanced; constructing a multi-order cross-scale module, and extracting high-level global semantic information and low-level detail texture information through adaptive average pooling; the method comprises the following steps of: constructing a decoder, fusing and restoring feature information of different levels, and finally using a Sigmoid activation function to output a segmented image, so that an electric power system defect image shot by an inspection robot is accurately processed, and the target recognition capability of a model when the model faces recognition targets of different scales and complex scenes are changed is improved.

The invention relates to the field of antennas, and particularly discloses a dual-polarized antenna, which comprises a slot antenna layer for radiating a first polarized wave and a microstrip antenna layer for radiating a second polarized wave, wherein the slot antenna layer comprises a waveguide structure, the waveguide structure comprises a plurality of radiation units arranged along a first direction, and each radiation unit comprises a plurality of radiation slots extending along a second direction; the micro-strip antenna layer comprises a plurality of micro-strip oscillator units arranged along a first direction, and each micro-strip oscillator unit comprises a plurality of micro-strip oscillators arranged along a second direction; the microstrip antenna layer is arranged on the slot antenna layer, at least one microstrip array unit is arranged between the radiation units, and projections of the microstrip arrays and the radiation slots on the waveguide structure are not overlapped; the antenna is simple in structure, has the characteristics of low profile, miniaturization and light weight, can obtain the electromagnetic scattering characteristics of a target in different polarization directions, and is suitable for a high-frequency dual-polarized antenna.

The invention provides an image data acquisition processing method and system based on computer vision, and relates to the technical field of computer vision, and the method comprises the steps: obtaining original acquisition images of a target recognition object at different angles; performing image de-noising on each original acquisition image to obtain a de-noised image; performing image enhancement on the de-noised image to obtain an enhanced image; performing feature extraction on the enhanced image to obtain image features corresponding to different angles; performing feature fusion on the image features to obtain fused features; and inputting the fusion feature into a trained multi-view target recognition model to obtain a processing result. According to the invention, the target identification capability and the image data acquisition and processing efficiency can be improved.

The SAR image matching identification method comprises the following steps: carrying out azimuth angle estimation on an SAR image, and screening to obtain various training sample feature data in an angle threshold range; fusing the screened multiple training sample feature data by adopting a parallel strategy to obtain a feature representation vector; and training an SVM classifier by using the obtained feature representation vector, and realizing matching identification of the SAR image by using the trained SVM classifier. The method is suitable for multi-dimensional feature matching recognition of the object in the SAR image, meanwhile, it is guaranteed that the method has the high operation efficiency, the high speed and the high-precision object matching capacity, and object classification and identity recognition tasks can be well completed.

The invention provides a plant station illegal behavior intelligent monitoring method and system based on an improved YOLO algorithm, and belongs to the technical field of industrial plant station safety management. The monitoring method comprises the steps of video stream acquisition and preprocessing, improved YOLO feature extraction and detection, specific illegal behavior identification, illegal behavior decision, alarm linkage, platform interaction and the like. The core improvement lies in that a lightweight backbone network is constructed by adopting GhostNet, and model parameters and calculation amount are reduced; a small target detection effect is optimized by newly adding a high-resolution detection head, introducing a BiFPN fusion mechanism and adopting a Focus loss and attention mechanism; color band channel enhancement, time sequence difference features and optical flow detection are fused, and flame / smoke recognition robustness is improved; and multiple types of illegal behavior recognition are integrated in the same model system, so that unified monitoring is realized. According to the method, the problems of insufficient model lightweight, unstable small target detection, unreliable fire identification, complex deployment and the like in the prior art are solved.

The invention discloses a marine mammal underwater acoustic target recognition method based on small sample deep learning, and relates to the field of underwater acoustic signal processing and recognition. In order to solve the problems of scarcity of samples, strong environmental interference and limited traditional single feature representation capability in underwater target identification, the method comprises the following steps: firstly, carrying out resampling and framing processing on underwater acoustic signals of marine mammals, and simulating a real marine environment based on a probabilistic random data enhancement strategy; three types of complementary acoustic features including a Gammatone frequency cepstrum coefficient, a time delay-Doppler feature and a wavelet time frequency feature are extracted and fused, and a multi-channel three-dimensional joint feature is constructed through standardization and early fusion; a ResNet-18 network is adopted for training, and dynamic learning rate scheduling and L2 regularization are introduced to enhance the generalization ability and robustness of the model. The recognition accuracy on a typical marine mammal audio data set is obviously better than that of a single feature and double feature fusion scheme, and the method is suitable for underwater acoustic target high-precision recognition under the small sample condition.

The present application discloses a method for identifying small parts faults in power transmission lines, including: obtaining a field image containing a target small part category on the power transmission line, performing multi-scale processing on the field image to obtain a multi-scale image feature map and a multi-scale sub-region feature map; performing enhancement processing on the multi-scale sub-region feature map, performing feature fusion on the processed multi-scale sub-region feature map and the multi-scale image feature map to obtain a multi-scale target region enhancement feature map, and then processing it to obtain a feature-optimized multi-scale fusion feature map; using the created image data set as a training set to train the constructed initial model, and using the trained model as a small part fault feature analysis model; analyzing the feature-optimized multi-scale fusion feature map based on the small part fault feature analysis model to obtain small part fault identification information. The method can realize accurate identification of small part faults in power transmission lines and meet the inspection needs under different terrains and landforms.

The utility model discloses a sighting telescope with a damping and buffering mechanism, and belongs to the technical field of sighting telescopes. Comprising a sighting telescope body, a precision adjusting knob, a multiple amplification auxiliary lens, an auxiliary lighting module, a lighting LED lamp, a lighting module switch, an energy storage module charging hole, a damping air cylinder, a sighting telescope connecting rod, an electric wire, auxiliary lens connecting threads, a sighting telescope objective lens, a sighting telescope eyepiece, an auxiliary lens objective lens, an external threaded sleeve and an energy storage module. By arranging the damping air cylinder and the sighting telescope connecting rod, recoil vibration transmitted to the sighting telescope can be reduced through the air cylinder structure, meanwhile, an auxiliary lighting module, a lighting LED lamp, a lighting module switch, an energy storage module charging hole and an energy storage module are arranged, and a sufficient light source can be provided for the sighting telescope; and meanwhile, the auxiliary lens connecting threads are arranged, the multiple amplification auxiliary lens and the external threaded sleeve can be externally connected with the sighting telescope according to needs, and the amplification multiple of the sighting telescope is increased.

The invention belongs to the technical field of nucleic acid aptamers, and particularly relates to a nucleic acid aptamer-loaded ruthenium (II) complex nano net as well as a preparation method and application thereof. The aptamer-loaded ruthenium (II) complex nano net provided by the invention can specifically recognize cancer cells in a targeting manner, and can reduce cytotoxicity. The ruthenium (II) complex is combined with the nucleic acid aptamer, so that the defects that the nucleic acid aptamer is unstable in metabolism and easy to remove by nuclease are overcome, the structural stability and the biological activity of the nucleic acid aptamer loaded ruthenium (II) complex nano net are improved, the compound can specifically recognize nucleolin in tumor cells, G4DNA conformation is formed, and the tumor cells can be specifically recognized. And targeted recognition of tumor cells is realized.

This invention discloses a method for obstacle recognition on expressways based on the cascaded use of Haar features and depth estimation features. The method obtains depth features from road images from depth data and performs shadow segmentation on the acquired road images using these features to obtain segmented images. Haar features and depth estimation features are extracted from the segmented images and subjected to parallel matching processing. A similarity-based attention mechanism is introduced to fuse the matching results of the Haar features and depth estimation features, resulting in a fused feature vector. The fused feature vector is then thresholded to obtain a depth estimation result. Based on the depth estimation result, a 3D model of the obstacle is reconstructed, and GPS is used to achieve obstacle recognition on expressways. This invention effectively solves the problems of low obstacle recognition accuracy due to illumination effects and poor target recognition capability for dynamic objects in road obstacle target detection.

The invention discloses a smart park target detection method and system for incomplete images. The method comprises the following steps: inputting a first image into a teacher model to obtain a teacher model target detection result and teacher model intermediate features; inputting the mask image corresponding to the first image into a student model to obtain a student model target detection result and student model intermediate features; respectively extracting the teacher model intermediate features and the student model intermediate features to obtain a teacher model network feature map and a student model network feature map; calculating to obtain a detection task loss value; calculating to obtain a global feature alignment loss value; optimizing the student model according to the detection task loss value and the global feature alignment loss value, and updating the teacher model to obtain a new teacher model; and the training is repeated until a first preset condition is reached, and a trained teacher model is obtained. The method has the characteristic of high target detection accuracy when facing the incomplete image.

The invention relates to the technical field of bird probing radar data processing, in particular to a bird probing radar data processing system and method for small target optimization, the system comprises a data acquisition module, an adaptive filtering module and a graph neural network processing module, the data acquisition module acquires radar data and expected data; the adaptive filtering module dynamically adjusts the step length and the weight based on the current error in the filtering process by setting the initial step length and the weight, and generates a radar feature vector; and the graph neural network processing module generates a target node according to the radar feature vector, defines an edge weight through an adjacent error, calculates an attention weight, and finally updates the radar feature vector of the target node, so that the precision and stability of small target signal extraction are improved, and the problems of signal noise and parameter fixation in a traditional method are solved.

The invention discloses a Conditional DETR image detection method based on graph attention weighting enhancement. The method comprises the steps of preprocessing a data set, constructing a Conditional DETR network, training the Conditional DETR network, storing a model, and testing the Conditional DETR network. The Conditional DETR network model is formed by sequentially connecting a backbone network, an encoder-decoder feature aggregation module and a prediction network in series, the backbone network takes ResNet50 as the backbone network to extract image features, and the image features are used for input of an encoder in the feature aggregation module; the encoder-decoder feature aggregation module is formed by connecting a self-attention mechanism module and a cross attention mechanism module in series; and the prediction network performs classification and regression prediction by using encoder-decoder output features, so that accurate detection of a target is realized. The method has the advantages of high detection precision, strong robustness and the like, and can be used for target detection of images.

The present invention relates to the technical field of bird-detecting radar data processing, and specifically to a bird-detecting radar data processing system and method optimized for small targets. The present invention includes a data acquisition module, an adaptive filtering module and a graph neural network processing module, wherein: the data acquisition module acquires radar data and expected data; the adaptive filtering module dynamically adjusts the step size and weight based on the current error during the filtering process by setting an initial step size and weight to generate a radar feature vector; the graph neural network processing module generates a target node according to the radar feature vector, defines edge weights through adjacent errors, calculates attention weights, and finally updates the radar feature vector of the target node. This scheme improves the accuracy and stability of small target signal extraction and solves the signal noise and parameter fixation problems in traditional methods.

The application discloses a sea surface target tracking method based on visual guidance radar under high sea conditions, and the method comprises the following steps: collecting millimeter wave radar point cloud, image, positioning, direction and speed data synchronously when the unmanned ship sails; using a YOLOv8 model optimized by slideloss weighting training and model distillation to detect the image target and obtain a target detection frame; preprocessing the radar point cloud; performing point cloud proposal and classification based on the image detection frame, respectively generating a clustering cluster through DBSCAN clustering, and fusing visual angle information of the proposed cluster; converting the clustering cluster to a world coordinate system, realizing double matching with a historical target through a Hungarian algorithm, initializing an unmatched target, and updating the state of a matched target and predicting the future position of the matched target. The application fills the short-distance perception blind area, significantly improves the precision and stability of sea surface target tracking under high sea conditions, and provides reliable protection for safe sailing of the unmanned ship.

The invention belongs to the technical field of express code detection, and particularly relates to an improved RT-DETR-based express code identification method in a complex environment, A0 and A1 are spliced along a channel dimension by introducing a CGFM module, and information of two feature maps can be combined; according to the fusion mode, the model can consider features from different sources or different levels at the same time, an attention mechanism is combined, and the ability to understand complex scenes and the ability to recognize small targets are enhanced; the false detection condition of the baseline network can be effectively avoided; the condition of incomplete target identification generated by bar code detection is avoided, meanwhile, the identification model can be combined with a neural network module of context information fusion and an attention mechanism, the ability of understanding a feature map is enhanced, and the position and the size of a prediction frame are effectively optimized; the detection precision is improved.