7 results about "Background distribution" patented technology

The application discloses a small sample point cloud semantic segmentation method based on difference enhancement and related equipment, and the method comprises the following steps: obtaining initial prototypes of various categories according to support set features and corresponding support set masks; calculating the similarity between the initial prototypes and query set features to obtain the probability distribution of the query set features belonging to foreground and background categories; taking the inverse of the foreground distribution according to foreground and background difference enhancement, combining the current background distribution to obtain accurate background probability distribution, and integrating to obtain reliable probability distribution of the entire category; the query set features are subjected to transformation processing through a feature aggregation module to obtain aggregated query set features; the reliable probability distribution and the aggregated query set features are subjected to aggregation processing through a prototype correction module to obtain pseudo prototypes; the support set data is expanded according to the pseudo prototypes through the prototype correction module to obtain corrected prototypes, and the segmentation result of the point cloud is obtained according to the corrected prototypes. The application obtains better segmentation effect.

This invention relates to the field of defect detection technology, specifically to a PCB defect detection method for industrial incremental scenarios. It includes a feature fusion method based on a lightweight segmentation strategy that combines random pruning with separate processing of strong and weak features. This strategy simplifies redundant computation by using a feature selector, and differentiates and recombines defect features of varying saliency during the feature fusion stage. This significantly improves inference speed while maintaining the precision of segmentation, solving the problems of high computational cost and difficulty in capturing minute defects in existing methods. The method also includes an incremental learning approach employing a background classifier adaptation mechanism and local semantic distillation. Class-specific regularization and spatially weighted logical alignment distillation work synergistically. By dynamically calibrating the background prediction logic and constructing a pixel-level semantic relevance matrix, it achieves deep alignment between new and old knowledge and background distribution. This effectively solves the catastrophic forgetting problem caused by the evolution of PCB background texture in existing incremental learning methods, significantly enhancing detection stability and the adaptability of the enhanced model.

The application discloses a kind of intelligent behavior detection method and system based on motion feature enhancement, construct based on adaptive mixture Gaussian model, the Mahalanobis distance of pixel point and background distribution is calculated and model weight is dynamically updated, realize the accurate decoupling and separation of foreground motion region in video frame;Then, the foreground mask obtained initially is based on the connected domain analysis of area threshold to filter out discrete noise, and the morphological closing operation of structure element adaptation is used to fill target internal cavity and repair edge fracture, to generate the high signal-to-noise ratio of motion region binary graph;Again, the purified binary graph is mapped as high visual saliency color heat map, it is weighted linearly fused with original RGB video frame, the fusion feature map is used to guide deep learning model to pay attention to motion region, and through double threshold detection strategy, the perception and identification ability of system to small target and hidden uncivilized behavior under complex dynamic scene is greatly improved.

The invention discloses a vehicle-mounted underway high-precision greenhouse gas two-dimensional distribution monitoring system and method, and relates to the technical field of environment monitoring. According to the vehicle-mounted underway high-precision greenhouse gas two-dimensional distribution monitoring method, gas is alternately collected through double-path sampling probes arranged at different heights of a vehicle roof, and instantaneous interference of exhaust emission of motor vehicles on a road is removed by combining an underway position and real-time meteorological data which are synchronously obtained and based on concentration difference and meteorological information by utilizing a specific algorithm, so that the real-time distribution of the greenhouse gas is realized. Therefore, the real environment background greenhouse gas concentration is extracted. According to the method, the high-precision two-dimensional greenhouse gas concentration distribution diagram of the underway area is inverted through spatial interpolation processing according to the purified background concentration data, credible quantification of regional scale greenhouse gas spatial distribution is realized, a key technical means is provided for accurate measurement and traceability of carbon emission, and the method is suitable for popularization and application. The core problem that the existing mobile monitoring technology is difficult to accurately reflect regional background distribution due to self-path pollution is effectively solved.

The invention discloses a ship detection method based on Warsestein distance fusion statistics, and relates to the field of polarized synthetic aperture radar ship target detection and identification. The method comprises the following steps: S1, acquiring polarized synthetic aperture radar data of an original ship and carrying out filtering processing; s2, performing feature extraction on the filtered polarized synthetic aperture radar data to obtain three polarized features; s3, using the three polarization features to calculate a statistical diagram of the Warisstein distance of each pixel of the image to be detected; and S4, screening a sea surface background on the statistical magnitude map by using a feature prior mode, estimating the distribution of the sea surface background by using a kernel density method in combination with a constant false alarm rate detection method, solving a threshold value according to a preset false alarm rate, and completing pixel-by-pixel judgment on the sea surface background. According to the method, the target / background contrast is remarkably enhanced, and high recall, low leak detection and good robustness can still be realized in complex scenes such as strong side lobes, stripe artifacts and dense targets.

The invention discloses a three-dimensional space-time kernel density estimation method based on adaptive bandwidth, and the method comprises the steps: determining the boundary contour of a research region in a geographic space, determining the lower bound and upper bound of the occurrence time of a target event, constructing an effective space-time cube, and collecting a target event sample in the interior, thereby obtaining a target event data set. And inputting the data set to calculate and fit the real background distribution of the target event, namely a joint probability density model. And carrying out time-by-time slicing processing on the joint probability density model to obtain a space condition probability density model. And respectively carrying out qualitative analysis and quantitative analysis on the two models according to requirements, and analyzing a spatial distribution rule and a time evolution characteristic of a high-incidence area (hot area) of a target event. The adaptive bandwidth change is introduced into the traditional algorithm, so that the algorithm provided by the invention has better fitting result, more accurate prediction performance, more concentrated hot area identification capability and more robust extreme event impact absorption capability.

The present application relates to the technical field of carbon footprint uncertainty analysis, and particularly relates to a carbon footprint uncertainty calculation method and electronic equipment, the method comprising: automatically generating a plurality of rules from unstructured text comprising carbon footprint historical data; fuzzy weighted synthesis of the plurality of rules in a scenario to generate a credibility distribution; extracting a global range and a scenario range according to a credibility threshold, and truncating the probability density function of the obtained truncated normal distribution on the global range to obtain a background distribution; equally dividing the scenario range into three scenario sub-intervals, and calculating the scenario weights using the background distribution; conditionally truncating sampling for each scenario sub-interval, and independently running Monte Carlo simulation to output the corresponding mean and variance; and weighting and synthesizing the Monte Carlo simulation results of each scenario through the corresponding scenario weights to obtain the uncertainty of the carbon footprint. Compared with the prior art, the present application has the advantages of efficient and accurate analysis of carbon footprint uncertainty, etc.