7results about How to "Simplify data volume" patented technology

This disclosure relates to the field of image processing technology, and provides an image deduplication method and apparatus for equipment security inspection. The method includes: acquiring a group of images to be deduplicated, the group comprising multiple pairs of images to be deduplicated; extracting spatial feature vectors for each pair of images in the group of images to be deduplicated based on a preset Siamese network model; calculating a similarity value between the spatial feature vectors of each pair of images in the group of images to be deduplicated; extracting color data feature vectors of images in each pair of images to be deduplicated based on the similarity value and a preset image threshold; and determining whether duplicate image pairs exist in the group of images to be deduplicated based on the color data feature vectors. This disclosure can quickly and in real-time find and filter out duplicate images taken at the same angle in the same scene, saving time and effort, increasing work efficiency, and improving the efficiency of subsequent equipment security inspection.

The invention discloses and provides a power module parallel digital current sharing method which is simple in data volume and high in communication quality. Each module collects current sharing frame IDs and current data of all modules on a CAN bus, calculates total current, the number of parallel machines and average current, sends the data to the CAN bus through current sharing frames, and selects a main module by comparing local addresses of all the current sharing frame IDs of the CAN bus. The module with the minimum local address is used as a main module, other modules are used as slave modules, the main module controls output voltage through a voltage loop, and the slave modules control output average current through a current loop. When the main module detects that the parallel output voltage drops to the set minimum voltage, current limiting and current sharing enabling closing occur in the module, and all the slave modules receive current sharing enabling closing signals and release a current loop, so that the stability of the parallel output voltage of the module is ensured. The method is applied to the technical field of the power module parallel digital current sharing method.

Methods and data processing systems are described for efficiently enriching dynamic vehicle data for software applications using static or semi-static vehicle data stored in a database, wirelessly transmitting, by the vehicle, a data record having dynamic vehicle data including static vehicle identifications to a data processing system external to the vehicle based on event triggering, the data processing system including a database having a plurality of static and semi-static vehicle data for the plurality of vehicles and a plurality of rich instances, each rich instance has only local access to a part of the vehicle data stored in the database, and the received data record is transmitted to at least one rich instance of the plurality of rich instances on the data processing system side on the basis of the vehicle identification, which has local access to the vehicle data relating to the corresponding vehicle in the database, each rich instance that has passed the received data record enriches the received data record with vehicle data to which it is locally accessible, relating to the respective vehicle, and forwards the enriched data record to the software application.

The invention relates to the technical field of data processing, can be applied to business system platforms of medical health, financial science and technology and the like, and discloses a text-guided visual data reconstruction method, device and equipment and a medium. Respectively coding the text description data and the visual data into a text embedding vector and a visual embedding vector; calculating an attention matrix according to the text embedding vector and the visual embedding vector; performing importance evaluation on each visual token in the visual embedding vector according to the attention matrix to obtain a visual token importance score; and screening effective text embedding vectors in the text embedding vectors according to the similarity between a preset threshold value and the text embedding vectors and the visual embedding vectors. Through redundancy trimming, recovery and reconstruction of the visual data, directional deletion of redundant data in the visual data and retention of core visual data, the completeness of the visual data is balanced, and the efficiency of subsequent calculation of the visual data is improved.

The embodiment of the invention provides a battery parallel capacity grading method and system, and relates to the technical field of battery capacity grading. The capacity grading method comprises the following steps: establishing a mathematical model of a battery parallel structure of a first-order RC equivalent circuit; obtaining detectable voltage and current data in the battery according to the mathematical model; obtaining an internal parameter value of each battery according to the voltage and current data; k-means clustering is carried out on each of the internal parameter values; and according to the classification number, carrying out comprehensive parameter capacity grading on the clustering result according to a preset rule. According to the method, only testing is needed in the middle section of the SOC, the cut-off voltage of the battery does not need to be judged, the data size can be reduced, dynamic capacity grading of the battery is achieved, the capacity grading cabinet only needs to be arranged in a constant-voltage mode, a complex working mode is not needed, and the capacity grading time and cost can be reduced.

The invention relates to an irradiation plan optimization method based on novel sparse sampling, belongs to the technical field of radiation irradiation dose optimization, and solves the problems that invalid voxel data volume cannot be effectively simplified, the processing data volume is large and the result is not accurate enough in the prior art. Receiving an initial irradiation plan and generating a full dose deposition matrix corresponding to each radiation field; constructing a gradient image based on the full dose deposition matrix; performing dual-condition voxel screening on the constructed gradient image to obtain a final sampling voxel set, and constructing a sparse dose deposition matrix; and performing irradiation plan optimization and dose normalization on the sparse dose deposition matrix to obtain an optimized irradiation plan.

The invention discloses a strain clamp point defect identification method, device and equipment and a storage medium. The method comprises the following steps: acquiring all scale and proportion combinations; combining different scales and proportions one by one, constructing a filter adaptive to the scale and proportion combination, and processing the to-be-detected image to obtain a feature response result of each pixel; selecting a target pixel according to a feature response result of each pixel under all scale and proportion combinations, and determining an optimal scale and proportion combination, an optimal dimension number, and a first local feature response calculated for the target pixel by a central region of a filter adapted to the optimal scale and proportion combination; obtaining a target scale and proportion combination of the adjacent dimension number, and determining a second local feature response, which is calculated for the target pixel, of the central region of the filter matched with the target scale and proportion combination; and determining that the target pixel is the center of the point defect according to the first local feature response, the second local feature response and a preset judgment condition.