70 results about "Feature extraction" patented technology

Described herein are systems and methods for performing multi-stage detection and classification of cancer regions from digitized images of biopsy slides. Novel methods for processing the digitized images to improve feature extraction and structure identification are disclosed, including but not limited to the use of quaternions, logarithmic mappings of color channels, and application of wavelets to logarithmic color channel mappings. The extracted features are utilized in improved machine learning algorithms that are further optimized to analyze multiple color channels in multiple dimensions. The improved machine learning algorithms include techniques for accelerating the training of the algorithms, making their application to biopsy detection and classification practical for the first time. The performance of the described systems and methods are further improved by the disclosure of a novel multistage machine learning scheme, in which additional classifiers are utilized to choose among the classes proposed by other classifiers in close cases.

The invention provides a full-automatic three-dimension characteristic extracting method. The method comprises the steps of firstly converting the three-dimension point clouds of all single view fields obtained by a stereoscopic vision measuring system into depth images of a single view field; splicing the point clouds of the single view fields; carrying out automatic searching and extracting on the corresponding depth images according to the known positioning characteristic of a workpiece to obtain the positioning characteristic of the workpiece, thus automatically converting the spliced three-dimension point clouds into the workpiece coordinate system; subsequently inputting the ideal coordinates of all characteristics to be detected of the workpiece in the workpiece design model; dividing the characteristics into a plurality of types such as aperture, tendon, plane and the like; automatically extracting the depth images respectively and obtaining the two-dimension coordinates of depth images; searching the three-dimension points of two-dimension depth image coordinates under the corresponding workpiece coordinate system; and obtaining the three-dimension information of each characteristic by calculation. The method has the characteristics of full automation, good flexibility and good expandability, and can be used for automatically extracting three-dimension characteristicsof various workpieces with known positioning characteristics.

The invention discloses an obstacle feature extraction method based on a millimeter wave radar and a laser radar. The method includes the following steps that: step 1, radar raw data are acquired; step 2, the radar original data are solved according to a radar protocol, so that obstacle position information can be obtained; step 3, the solved millimeter wave radar data are screened; step 4, time and space synchronization is performed on the millimeter wave radar and the laser radar; step 5, regions of interest (ROI) are established in a laser radar coordinate system according to millimeter wave radar information and are denoted as Ri, wherein i is equal to 1, 2, ... n, wherein n is the number of the effective targets of the millimeter wave radar; step 6, the point cloud data of the i-th ROI are processed according to a scattered point cloud data simplification processing method; and step 7, a two-dimensional image is generated on the basis of the point cloud data, edge detection is performed on the image, and obstacle features are extracted. With the method adopted, obstacle edge information can be obtained, and therefore, more references can be provided for a driverless decision-making layer.

The invention discloses vehicle identification method and system based on multilayer descriptors. The method comprises the following steps of: (1) extracting the global features of a vehicle; (2) extracting the singular area extraction features of the vehicle; (3) constructing the multilayer descriptors; (4) carrying out vehicle identification based on the multilayer descriptors, and acquiring the final vehicle identification result by virtue of various matching strategies, in case of comprehensively considering matching accuracy and matching speed. The invention develops a vehicle identification system based on multilayer features on the basis of the method; some specific vehicle identification systems are further invented on the infrastructure of the system. According to the vehicle identification method disclosed by the invention, the vehicle is described by the multilayer descriptors, and the matching accuracy and matching speed among the multilayer descriptors are ensured by the various matching strategies, thus the method is capable of obtaining good effects in vehicle identification applications in different fields.

The invention discloses a polarized SAR image classification method based on a K mean value and a depth SVM. The polarized SAR image classification method mainly solves the problems of low classification accuracy and classification efficiency of an existing polarized synthetic aperture radar (SAR) classification method. The polarized SAR image classification method comprises the steps of (1) inputting images; (2) performing filtering; (3) extracting features; (4) establishing a misclassification set; (5) establishing a nearest neighbor sample set; (6) establishing a final training set; (7) establishing a depth support vector machine classifier; (8) performing classification; (9) calculating accuracy. By means of the polarized SAR image classification method, polarized SAR images can be classified accurately, the classification time of the polarized SAR images can be shortened effectively, and target recognition and tracking of the polarized SAR images are achieved.

The invention discloses an electroencephalogram signal feature extraction method combining with a public space mode algorithm and an EMD. Firstly, electroencephalogram signals of a subject are selected as a training set and a testing set, and the signals of the single subject in two channels C3 and C4 are pre-processed; then, experience mode decomposition is conducted on the prepressed EEG signals to obtain a series of intrinsic mode functions IMFi, and energy spectrum diagrams of all the intrinsic mode functions are drawn; then, previous three order IMF components of the channels C3 and C4 undergoing a single pass test are merged to form a N * T matrix Xi, wherein N represents IMF number, T represents the number of sampling points undergoing one test, the whole experiment process includes G groups of tests, and G groups of vector matrixes are obtained and are divided into G1 groups of test vector matrixes and G2 groups of training vector matrixes which are respectively subjected to public space mode decomposition. CSP filtering is conducted on the intrinsic mode functions decomposed by utilizing the three channels and the EMD, frequency domain information of the EMD is added on the basis of the CSP, and the problem that the CSP is lack of the frequency domain information is solved.

The invention discloses a bidirectional LSTM network-based vehicle behavior identification method and system; the method comprises the following steps: invoking offline acquired traffic video data; detecting and tracking vehicles in the traffic video data, and extracting vehicle driving tracks; extracting features from preprocessed vehicle driving tracks, and building a training data set and a test data set; using a bidirectional LSTM recurrent neural network to make model training for the training data set, and forming a vehicle behavior identification model; inputting the test data set intothe vehicle behavior identification model to make precision assignment; inputting the traffic video data acquired online in real time in the vehicle behavior identification model with the precision assignment, identifying vehicle behaviors, and outputting an identification result. The method uses the vehicle behavior identification model based on the bidirectional LSTM network to identify the vehicle driving tracks in the traffic video data, thus determining the vehicle driving behaviors with a high identification precision.

The invention provides an abnormality detection method and device based on log graph modeling. The method and the device are applied to a nonsocial network. The method particularly includes: buildinga bipartite graph according to a key field of abnormal data annotated in advance in the nonsocial network, wherein left-side nodes of the bipartite graph correspond to multiple user accounts while right-side nodes of the same correspond to parameter combinations during business interface requesting; extracting features from the bipartite graph, and splicing the extracted features to form a featurevector; performing k-means cluster processing on the basis of the abnormal data and the feature vector to acquire optimal cluster number; according to the optimal cluster number, using a Gaussian mixing model to fit dark industry feature probability distribution; when receiving incoming data, calculating dark industry probability of the data according to feature vector of the incoming data and the Gaussian mixing model, and judging whether the data are abnormal or not according to the dark industry probability and the dark industry feature probability distribution. When the data are judged abnormal, access behaviors of users can be intervened timely, so that network attacks by hackers can be avoided.

The invention relates to a handwriting signature identification equipment and a method thereof. The equipment comprises a video collection module, an initialization module and a feature extraction module, wherein, the video collection module is used for collecting a group of handwriting signature visual images; the initialization module is used for obtaining high-quality images to be detected by utilizing the handwriting signature visual images collected; and the feature extraction module is used for extracting the features of the images to be detected so as to provide convenience for classifying and judging the effectiveness of the images. By combining video collection, initialization process and image feature extraction, processing modules are embedded into a camera, and the equipment and the method of the invention realize a handwriting signature identification system with low cost, convenient use and small network burden.

The invention discloses a method for counting targets in a remote sensing image based on an attention mechanism, which fuses the characteristics of the attention mechanism, a scale pyramid and deformable convolution on the basis of VGG16, and consists of three cascaded stages: feature extraction of a front-end network, scale fusion of a middle-end network and generation of a density map of a rear-end network. Through the technical scheme of the invention, the problems of target scale diversity, complex and disordered background interference and target direction randomness in a dense target counting task in the remote sensing image can be well solved.

The invention discloses a rolling bearing multi-fault coupling mechanism and early fault feature extraction method in the technical field of rolling bearing research. The method comprises the following specific steps: S1, analyzing a vibration mode of a rolling bearing based on nonlinear dynamics; S2, performing fault evolution analysis of the rolling bearing based on the bifurcation theory; S3, extracting fault features of the rolling bearing based on a chaotic feature quantity; S4: carrying out the simulation analysis and fast testing of the fatigue life; S5, carrying out the nonlinear dynamic coupling analysis and early fault feature extraction of the rolling bearing. The method employs a nonlinear dynamics method to analyze the multi-fault coupling features of the rolling bearing in afatigue life test, employs the bifurcation theory to analyze the change rule of a vibration signal in the evolution process of a single fault and the path to chaos, employs the characteristic that anarray subsystem is extremely sensitive to small parameter disturbances for achieving the early fault feature extraction and diagnosis of the rolling bearing.

The invention relates to a fraud detection method and device for video data, computer equipment and a storage medium. The method comprises the following steps: acquiring to-be-detected video data; extracting image data of each video frame from the to-be-detected video data, and dividing the image data into a plurality of image data sets according to the time sequence of each video frame, the imagedata sets which comprises image data corresponding to continuous video frames; inputting each image data set into a pre-trained image feature extraction model to obtain an image feature vector; extracting voice data from the to-be-detected video data, and obtaining a voice feature vector of the voice data; performing cascade splicing on the image feature vector and the voice feature vector to obtain a multi-modal feature vector; and inputting the multi-modal feature vector into a pre-trained fraud detection model to obtain a fraud detection result corresponding to the to-be-detected video data output by the fraud detection model. By adopting the method, the characteristic information amount can be increased, the comprehensiveness and diversity of the characteristic information are improved, and the accuracy of video data fraud detection is effectively improved.

The invention discloses an odor identification method based on gas sensors and deep learning, which is characterized in that a response curve cluster of odor to be detected is obtained through a gas sensor array, original data is directly used as an input sample of an odor identification deep neural network, data preprocessing and data amplification are performed on the input sample, time sequenceresponse data hierarchical features are automatically extracted by adoption of deep learning, global feature extraction and long-range dynamic feature extraction are performed at the same time, and an odor label is output through a classifier, so that high-sensitivity and specific odor identification is realized. The odor identification method of the invention has high sensitivity and high reliability, and can be widely applied to the fields of industrial production, medical treatment, environment, safety and the like.

The invention discloses a principal component analysis collaborative filtering method for image multi-direction morphological structure grouping. The principal component analysis collaborative filtering method for the image multi-direction morphological structure grouping includes subjecting an image to overlapping partition, and subjecting image blocks to multi-direction morphological structure grouping according to variance, gradient and singular value information of the image blocks to obtain a smooth block group, a random block group, a direction edge block group and a direction texture block group; then subjecting the grouped image blocks to characteristic adaptive principal component analysis, utilizing hard threshold shrink method to subject image block transformation coefficients to filtering and then to grouping reconfiguration; and finally aggregating the grouping reconfiguration blocks into a full-size filtering image. The principal component analysis collaborative filtering method for the image multi-direction morphological structure grouping fully considers the multi-direction morphological structure of image blocks and non local similarity information of the image, has good structure and texture maintaining characteristics in image filtering process, is strong in noise elimination capacity, and is widely applicable to the image characteristic extraction and preprocessing process of target detection.

The invention discloses an image segmentation method for flaw detection of a cell tail end. Firstly, an initial BGR image at the cell tail end is converted into a gray scale image by considering image characteristics of intelligent flaw detection of the cell tail end, then the gray scale image is subjected to binaryzation processing so as to obtain a binaryzation image, and then coordinates of four key pixel points on the outer contour of the cell tail end are determined on the binaryzation image. The size of an ROI rectangular area where the cell tail end exists is determined according to the coordinates of the four key pixel points on the outer contour of the cell tail end, therefore the ROI rectangular area where the cell end tail exists is marked off in the initial BGR image, and a target area is not damaged. High accuracy in image dividing is ensured, operation is simple, and an early-stage guarantee is provided for accuracy of follow-up processing algorithms (such as the image characteristic extracting and matching algorithm). In addition, the ROI rectangular area is marked off in the initial BGR image, so that the processing of the whole image comprising the background with the follow-up processing algorithms is avoided, detecting efficiency of a cell is improved, and detecting cost is reduced.

The invention discloses a super key distributed searching method, which is information retrieval for intelligently matching full text of a text. The full test is used as the searching input, and intelligent fusion is performed on a search engine framework in multiple stages. A system processes the whole text to obtain representative keywords comprising keyword itself, keyword disambiguation result, keyword weight and relation among the keywords. An intelligent process of the system uses the following modules: a full text matching algorithm merging module (FMM), a multi-feature extraction module (FE) and a global feature extraction module (WFE), wherein the FMM is distributed in a basic server and an advanced server; the FE is distributed in an inverted extraction process of the basic server and a query analysis process of the advanced server; and the WFE is distributed in a searching module. The system keeps the advantages of high data throughput rate, high reliability, concurrent service and the like of the distributed searching, causes no additional complexity in searching and improves relativity of searching.

An automated transaction device and an automated transaction system avoiding the misbehaviors while registering the organism information for the personal authentication are provided. In the automated transaction system, the network can connect the customer terminal of the customer operation in communicating way and the central terminal of the operation by the operator. The hand side facial photography portion shoots the face and hand side of the customer while obtaining the organism data and the operating the customer terminal. In addition, while obtaining the organism data in order to process the personal authentication, the invention also shoots the face and hand side of the customer. Then the characteristic comparing portion obtains the organism data to confirm whether the picked customer hand side characteristic is accordant; if the customer hand side characteristic is not accordant, the central operator uses the central terminal to compare the faces of the customers of the organism data; if the situation can not be confirmed, the situation is set as no transaction to avoid the misbehaviors.

The invention provides an information processing device, a server, an information processing method and a program. The information processing device includes a playback processing unit for playing back media-recorded content recorded on an information recording medium, a feature extraction unit for extracting a feature amount of the media-recorded content, and an information processing unit for obtaining a comment which is set to similar content having a feature amount analogous to the feature amount of the media-recorded content extracted by the feature extraction unit and for performing control in a manner that the obtained comment is superimposed on a playback image of the media-recorded content.

The invention belongs to the technical field of digital forensics, and particularly relates to a malicious code fragment intelligent forensics method and system, and the method comprises the steps: constructing a code fragment training set and a code fragment test set for training and testing through extracting the underlying data features of a storage medium; training the set full-connection neural network model by using the data in the code fragment training set, the input being a feature vector, and the output being a normal or malicious prediction result; for the code fragment test set, performing test output by utilizing the trained full-connection neural network model to judge whether model input is a malicious code fragment; and performing feature extraction on the target code snippets, and inputting the target code snippets into a fully-connected neural network model generated through training and testing to obtain an intelligent malicious code recognition result of the targetcode snippets. According to the method, malicious code fragments in storage media such as computers, mobile phones and tablets and evidence containers such as RAW, E01 and AFF can be recognized, and the method has a good application prospect in the field of digital evidence collection such as crime event evidence underlying data automatic analysis.

The invention provides a pipeline magnetic flux leakage detecting data automatic analysis system. The pipeline magnetic flux leakage detecting data automatic analysis system comprises a pipeline magnetic flux leakage detector, a client end and a server, and the pipeline magnetic flux leakage detector collects pipeline magnetic flux leakage detecting data and transfers to the server. A data generator, a trainer, a reasoning machine and a predefined self-learning model are arranged on the server, feature extraction is automatically carried out and mode recognition, defect classification and defect quantization are automatically carried out through deep learning artificial intelligence technology, and labor investment is greatly reduced.

The invention discloses an industrial big data-based equipment residual life prediction method and system and a medium, and the method comprises the following steps: collecting the real-time operation data of equipment, carrying out the feature extraction and feature selection of the real-time operation data, and obtaining the data features; performing data drift detection according to the data features, and determining the health stage of the equipment; and obtaining a corresponding residual life prediction model according to the health stage, constructing a health factor according to the data features, inputting the health factor into the obtained residual life prediction model, and outputting an equipment residual life prediction result. According to the operation data generated in real time in industrial actual production, the health factor of the equipment is acquired, and the residual service life of the equipment at the current moment is predicted by using the health factor, so that the complete life cycle of the equipment can be more utilized, the service life waste and the time waste caused by sudden shutdown are reduced, the equipment health management is automatically carried out. The method can be widely applied to the field of prediction of the residual life of the equipment.

The invention discloses an identity recognition method based on multi-modal information. The identity recognition method comprises the following steps: 1, making a multi-modal video data set with a label; 2, respectively constructing and training a face detection model and a head detection model; 3, constructing and training a feature extraction model of the face, the head and the sound; 4, performing feature extraction on the face, head and sound information through the trained feature extraction models; 5, constructing and training a classification model to respectively classify the three extracted features; 6, performing result prediction by using the three features through the classification model; 7, performing information fusion on the classification result according to the formulated multi-modal information fusion strategy; 8, arranging the fused result and then outputting an identity recognition result. Based on the identity recognition network model based on multi-modal information provided by the invention, the identity recognition network model has a wide application prospect in the fields of human-computer interaction, information security, security monitoring and the like.

The invention relates to an information pushing method and device based on a block chain, computer equipment and a storage medium. The method comprises the steps of obtaining each block of a block chain; analyzing the block chain, and obtaining multi-dimensional characteristics of a user recorded in each block on the block chain and prediction probabilities of a plurality of pieces of to-be-pushed information corresponding to the multi-dimensional characteristics of the user; and pushing the to-be-pushed information to a user according to the prediction probability. Distributed storage of the operation information and the push information probability of the user is realized through the block chain; multi-dimensional features of the user and the probability of to-be-pushed information are extracted according to the multi-dimensional features; calculation of comparative consumed resources such as user preference feature extraction and the like is directly carried out at the user side, distributed tracking storage is realized, decentralization of feature comparison calculation is realized, the pressure of the server side is greatly reduced, accurate pushing and user requirement mining are realized, and information pushing is more accurate and timely.

The invention provides a steel coil end face defect distinguishing method and system and electronic device. The method comprises: obtaining an adversarial network; training the adversarial network byusing a real steel coil end face image without defects; obtaining a steel coil end surface counter-resistance network; equally dividing the end face image of the steel coil to be measured into a plurality of first image blocks, extracting a first feature vector corresponding to each first image block, outputting a plurality of corresponding reconstructed images of each first image block based on the to-be-detected steel coil end face reactance network; performing feature extraction on each reconstructed image; obtaining a second feature vector, sequentially judging whether a first loss value between each first feature vector and the corresponding second feature vector exceeds a first threshold value or not. Therefore, the confrontation network only needs to train a defect-free steel coil end face image, feature extraction is carried out on the reconstructed image, whether the defect exists or not is judged through the loss value of the feature vector, the calculation amount is reduced,and the recognition efficiency is improved.

Compared with the prior art, the astronomical mass observation data hybrid retrieval method based on the image content and the metadata comprises the following steps of acquiring a to-be-retrieved observation image; extracting an image file header from a to-be-retrieved image file header to obtain metadata A1; performing image feature extraction processing on the to-be-retrieved image to obtain image feature vector data A2; according to the metadata A1, retrieving from a metadata database by utilizing a metadata index to obtain a first hit data set D1; according to the image feature vector data A2, retrieving from an image feature database by using an image feature index to obtain a second hit data set D2; and performing hit data set intersection processing on the hit data set to obtain afinal detected image data set. Compared with the prior art, the retrieval accuracy of massive astronomical image data can be improved, the retrieval quality and effect can be improved, the massive data query cost can be reduced, and the working efficiency of astronomical workers can be further improved.

The invention discloses a data recommendation method and device based on deep learning, and a terminal device. The method comprises the steps of obtaining character portraits of a plurality of applicants; wherein the character portraits of the applicants and the character portraits of the applicants are in one-to-one correspondence; wherein the character portrait of the applicant comprises multi-dimensional information of the applicant; performing feature extraction on the obtained character portraits of the plurality of applicants; inputting the extracted first features into a trained training model to obtain scores of a plurality of applicants respectively; and the score of each applicant in the plurality of applicants being in one-to-one correspondence with each applicant, and determining a target applicant whose score is greater than or equal to a preset threshold value from the plurality of applicants. By the adoption of the method and device, preliminary screening of a large number of applicants can be rapidly and accurately completed, the working efficiency is improved, and the labor cost is saved.

The embodiment of the invention discloses a maturity evaluation method and device and a storage medium. The method comprises the steps: acquiring target knocking sound generated by knocking target melons and fruits in a current scene; performing feature extraction on the target knocking sound to obtain target sound features; and analyzing the target sound features by using a sound analysis model to determine the maturity of the target melons and fruits, wherein the sound analysis model is a model with a function of evaluating maturity by utilizing sound information.