3201results about How to "Robust" patented technology

Hashes are short summaries or signatures of data files which can be used to identify the file. Hashing multimedia content (audio, video, images) is difficult because the hash of original content and processed (e.g. compressed) content may differ significantly. The disclosed method generates robust hashes for multimedia content, for example, audio clips. The audio clip is divided (12) into successive (preferably overlapping) frames. For each frame, the frequency spectrum is divided (15) into bands. A robust property of each band (e.g. energy) is computed (16) and represented (17) by a respective hash bit. An audio clip is thus represented by a concatenation of binary hash words, one for each frame. To identify a possibly compressed audio signal, a block of hash words derived therefrom is matched by a computer (20) with a large database (21). Such matching strategies are also disclosed. In an advantageous embodiment, the extraction process also provides information (19) as to which of the hash bits are the least reliable. Flipping these bits considerably improves the speed and performance of the matching process.

A method of bonding of germanium to aluminum between two substrates to create a robust electrical and mechanical contact is disclosed. An aluminum-germanium bond has the following unique combination of attributes: (1) it can form a hermetic seal; (2) it can be used to create an electrically conductive path between two substrates; (3) it can be patterned so that this conduction path is localized; (4) the bond can be made with the aluminum that is available as standard foundry CMOS process. This has the significant advantage of allowing for wafer-level bonding or packaging without the addition of any additional process layers to the CMOS wafer.

Systems and methods are provided that empowers various parties to combine transactional data and local non-transactional data using the collective intelligence gathered from a variety of sources to help the parties make more intelligent decisions relating to consumers. For example, the system can help select consumers based on the probability that the consumers will take advantage of an offer, coupon, or other item. In some embodiments, the present invention can be deployed as a part of a system that processes transactions. In this system, information associated with the transactions is analyzed in conjunction with non-transactional data in order to probabilistically determine whether a further action should be taken with the consumer.

The invention discloses an attention mechanism-based in-depth learning diabetic retinopathy classification method comprising the following steps: a series of eye ground images are chosen as original data samples which are then subjected to normalization preprocessing operation, the preprocessed original data samples are divided into a training set and a testing set after being cut, a main neutralnetwork is subjected to parameter initializing and fine tuning operation, images of the training set are input into the main neutral network and then are trained, and a characteristic graph is generated; parameters of the main neutral network are fixed, the images of the training set are adopted for training an attention network, pathology candidate zone degree graphs are output and normalized, anattention graph is obtained, an attention mechanism is obtained after the attention graph is multiplied by the characteristic graph, an obtained result of the attention mechanism is input into the main neutral network, the images of the training set are adopted for training operation, and finally a diabetic retinopathy grade classification model is obtained. Via the method disclosed in the invention, the attention mechanism is introduced, a diabetic retinopathy zone data set is used for training the same, and information characteristics of a retinopathy zone is enhanced while original networkcharacteristics are reserved.

A container for an aerosol generating device is disclosed. The container has a body comprising a first portion configured to receive an aerosol generating device and one or more retention means externally accessible from the body, each retention means being configured to releasably retain a module.

The invention discloses a pedestrian re-identification method based on multi-attribute and multi-strategy fusion learning. The method of the invention includes the steps of in an offline training phase, firstly selecting pedestrian attributes which are easy to be judged and have a sufficient distinguishing degree, training a pedestrian attribute identifier on an attribute data set, then labeling attribute tags for a pedestrian re-identification data set by using the attribute identifier, and next, by combining the attributes and pedestrian identity tags, training a pedestrian re-identification model by using a strategy fused with pedestrian classification and novel constraint comparison verification; and in an online query phase, extracting features of a query image and images in a database by using the pedestrian re-identification model, and calculating the Euclidean distance between the feature of the query image and the feature of each image in the database to obtain the image with the shortest distance, which is considered as the result of pedestrian re-identification. In terms of performance, the features in the invention are distinguishable and high accuracy is obtained; and in terms of efficiency, the method of the invention can quickly search for the pedestrian indicated by the query image from the pedestrian image database.

Indirect airflow measurement through breath-operated device is accomplished by incorporating an airflow sensor into the inhaler device along a low resistance channel disposed away from the exhaust chamber of the device and having an input port in airflow communication with a low resistance channel and output input port and low resistance channel are formed in the main housing body of the device, and further incorporating an output port formed near the exhaust changer near the mouthpiece assembly, the output port also in airflow communication with the low resistance channel. A method of measuring airflow in an inhalation device is also described that measures air flowing through the low resistance channel. Another aspect of the invention provides a method that allows for the closure of the devices' airflow ports, by allowing for the rotation of the mouthpiece assembly from open to closed positions relative to the inhaling device's main housing body and towards handle assembly.

The invention discloses a method for the fault diagnosis of wind turbines on the basis of a genetic neural network, in particular to a method for modeling the fault diagnosis of wind turbines on the basis of the genetic neural network capable of learning the operating data of the wind turbines in the history, more particularly a method for judging the probability that faults occur to a gearbox, a generator and a yaw system by reading the real-time operating data of the wind turbines online and calling the diagnosis model of the genetic neural network to carry out the analysis on the real-time data, thus judging the fault state of the wind turbines. Based on the method combining the genetic algorithm with the neural network, the invention can achieve the algorithm complementation, improve the model convergence and diagnostic capacity and ensure higher robustness; and the method capable of carrying out the online monitoring and fault diagnosis on the operating state of the wind turbines on a real-time basis to take maintenance measures as soon as possible can improve the reliability of the wind turbines and reduce the maintenance cost.

The invention relates to one efficient wireless sensor network topological control method, which combines topological control cluster division method and power control method and comprises the following steps: a, dividing the network clusters and selecting cluster head point; b, forming cluster inner network; c, processing power control on cluster point to form final network topological structure; after failure, re-establishing network topological structure.

An optical scanner with plural scan beam generation. Scan angle stress is configured in the beam generation.

The invention discloses a method for improving a BP (back propagation) neutral network and based on a genetic algorithm. The method includes coding the BP network to determine structure of the neutral network, wherein the structure includes the number of hidden layers and the number of units of each layer; adopting real-number coding to code by taking weight and threshold as genes, wherein each neutral network corresponds to a chromosome after coding; using the genetic algorithm to perform selection optimization on the network, wherein selection optimization includes the steps of selection, crossing and variation; training the BP network to acquire a final result; decoding an optimal individual selected by the genetic algorithm to generate a new neutral network, and training the new network by applying a BP training algorithm to acquire a final result. The method combines the genetic algorithm with the BP network, thereby being capable of fully utilizing advantages of the both, the problem that initial weight and threshold of the BP network are difficult to determine can be solved, searching range can be narrowed, training speed of the BP network can be increased, and the problem of local minimum can be improved.

The invention discloses a pile-up noise reduction own coding network bearing fault diagnosis method based on particle swarm optimization. The bearing fault diagnosis method provides an improved pile-up noise reduction own coding network SADE bearing fault diagnosis method, SDAE network hyper-parameters, such as cyber hidden layer nodes, sparse parameters, input data random zero setting ratio, are selected adaptively by particle swarm optimization PSO, a SADE network structure is determined, top character representation of malfunction inputting a soft-max classifier is obtained and a classification of defects is discerned. The bearing fault diagnosis method has better feature in learning capacity and more robustness than feature of learning of ordinary sparse own coding device, and builds a SDAE diagnostic model having multi-hidden layer by optimizing the hyper-parameters of noise reduction own coding network deepness network structure with the particle swarm optimization, accuracy of the classification of defects is improved at last.

The invention discloses a multi-gesture and cross-age oriented face image authetication method. The method comprises the following steps of: rapidly detecting a face, performing key point positioning, performing face alignment, performing non-face area filtration, extracting the face features by blocks, performing feature dimension reduction and performing model prediction. The method provided by the invention can perform the face alignment, realize the automatic remediation for a multi-gesture face image, and improves the accuracy rate of the algorithm, furthermore, the feature extraction and dimension reduction modules provided by the invnetion have robustness for aging changes of the face, thus having high use value.

The invention relates to a binocular stereoscopic vision-based stereo matching method. The method includes the following six stages: Gaussian pyramid construction; cost calculation matching and cost aggregation; cost fusion matching; disparity computation; disparity map repair and void filling; and disparity refinement. Laplacian pyramid transformation is additionally adopted in the cost aggregation stage. An edge-protection-based interpolation algorithm is used in the disparity map repair and hole filling stage. A weighting and bilateral filtering combination-based disparity refinement method is additionally adopted in the disparity refinement stage, so that a high-accuracy disparity map can be obtained. The calculation amount of the method of the invention is moderate; matching results at different scales are fused, improvement is made in the cost aggregation stage and the disparity refinement stage, and therefore, a better disparity map can be obtained; and the method has certain robustness to illumination, external noises and the like.

The invention discloses a character detection method and device based on deep learning. The method comprises the steps: designing a multilayer convolution neural network structure, and enabling each character to serve as a class, thereby forming a multi-class classification problem; employing a counter propagation algorithm for the training of a convolution neural network, so as to recognize a single character; minimizing a target function of the network in a supervision manner, and obtaining a character recognition model; finally employing a front-end feature extracting layer for weight initialization, changing the node number of a last full-connection layer into two, enabling a network to become a two-class classification model, and employing character and non-character samples for training the network. Through the above steps, one character detection classifier can complete all operation. During testing, the full-connection layer is converted into a convolution layer. A given input image needs to be scanned through a multi-dimension sliding window, and a character probability graph is obtained. A final character region is obtained through non-maximum-value inhibition.

The invention relates to the fault diagnosis and forecast filed, in particular to a failure prediction method facing to a numerically-controlled machine tool. The failure prediction method comprises the following steps of adopting a hierarchical-type hierarchical structure model to divide the numerically-controlled machine tool to be a plurality of core subsystems and analyze typical gradual failures; reducing a data set of sensor parameters to obtain a data set of failure foreboding parameters and relative relevance degree between the parameters and the failures; using a failure occurrence point to serve as a limit, diving each failure foreboding parameter historical data set according to time series, and corresponding to failure foreboding state series; adopting wavelet analysis technology to extract failure foreboding feature vectors of the data in different time intervals, conducting counter propagation neural network training, and obtaining a failure foreboding judgment model of each parameter; and adopting a dynamic confidence coefficient matching algorithm to monitor an accumulated confidence coefficient of each failure foreboding parameter on line, fusing state dynamic matching results of each failure foreboding parameter, and forecasting probability and time of failure occurrence. The failure prediction method has the advantages of high forecast accuracy, small forecast time difference, low false alarm rate, strong robustness, wide application prospect and the like.

The invention relates to a deep convolutional neural network-based traditional Chinese medicine tongue image automatic segmentation method and belongs to the computer vision field and traditional Chinese medicine tongue diagnosis field. According to the method of the invention, a convolutional neural network structure is designed; collected sample data are adopted to train the network, so that a network model can be obtained; and the model is adopted to automatically segment a traditional Chinese medicine tongue image. The method includes an offline training phase and an online segmentation phase. The method can be applied to both closed type and open tongue image acquisition environments and can effectively improve the accuracy and robustness of the automatic segmentation of the traditional Chinese medicine tongue image. The method of the present invention specifically relates to deep learning, semantic segmentation, image processing and other technologies.

The invention discloses a mixed control method based on trace tracking of a wheeled mobile robot. A kinematic virtual velocity controller, a sliding mode torque controller and a disturbance observer are involved in the mixed control method, wherein the sliding mode torque controller and the disturbance observer are based on dynamics. The virtual velocity controller is used for designing the linear velocity and the angular velocity of the robot; the sliding mode torque controller is used for designing a sliding mode face and a sliding mode control law, and the disturbance observer is used for observation of the external disturbance of a system to reduce the control quantity of the sliding mode controller and is introduced as a feedforward term. By means of the mixed control method, control over the trace tracking of the robot is achieved by the system under the condition that external change and external disturbance happen to a parameter. It is shown upon simulation experiments that by means of the mixed control method, chatter output by sliding mode control and output of the control quantity can be effectively reduced, and good robustness is achieved.

The present invention discloses a facial emotion recognition method based on a depth sparse self-encoding network. The method comprises the steps of 1, acquiring and pre-processing data; 2, establishing a depth sparse self-encoding network; 3, automatically encoding / decoding the depth sparse self-encoding network; 4, training a Softmax classifier; and 5, finely adjusting the overall weight of the network. According to the technical scheme of the invention, sparseness parameters are introduced. In this way, the number of neuronal nodes is reduced, and the compressed representation of data can be learned. Meanwhile, the training and recognizing speed is improved effectively. Moreover, the weight of the network is finely adjusted based on the back-propagation algorithm and the gradient descent method, so that the global optimization is realized. The local extremum and gradient diffusion problem during the training process can be overcome, so that the recognition performance is improved.

The invention provides a motor servo system jitter-free sliding mode position control method based on disturbance compensation. According to the method, the nonlinear friction characteristic, external disturbance and other modeling uncertainty of a system are considered, continuous and smooth friction compensation is made for nonlinear friction, and the low-speed servo performance of the motor position servo system is further improved; uncertainty such as non-modeling disturbance is estimated through an extended state observer, feedforward compensation is made when a controller is designed, and the robustness of the actual motor position servo system to external disturbance is improved; jitter and singularity will not be caused to voltage output of the designed terminal sliding mode controller, the controller can guarantee that the state of the system tends to be balanced in a limited time, and the tracking performance of the system is greatly improved; the designed terminal sliding mode controller is simple, has certain robustness to system parameter variation and is more favorable for being applied to engineering practice.

The invention relates to a method for identifying abnormal microblog users. The method comprises the steps of obtaining a plurality of users' microblog data, storing the microblog data into a database, taking statistical distribution of time intervals of user behaviors as behavior time characteristics of the users according to the microblog data of the users, generating behavior time characteristic vectors and defined parameters, calculating Kullback-Leibler divergence between the behavior time characteristic vectors of the normal users and the behavior time characteristic vectors of the users to be detected, judging the users to be detected with the calculated Kullback-Leibler divergence exceeding the defined parameters as the abnormal users, and extracting and showing keywords of contents of the abnormal users. The invention further provides a system for identifying the abnormal microblog users corresponding to the method. According to the method and system, the keywords of the blog article contents of the abnormal users can be extracted quickly, promulgators of junk information such as marketing and advertisements can be identified accurately, and the method and the system are applicable to detection of multiple microblog service platforms, and has the advantages of high accuracy and efficiency and wide applicability.

The invention discloses a static gesture recognition method based on a finger contour and decision-making trees. The method comprises the steps that a Kinect depth image is used as a data source at first, the approximate coordinates of the palm are positioned through the Kinect skeleton tracking function, and a square area containing the palm is cut out with the coordinates as a center; the self-adaptive adjacent value method is used for conducting foreground detection on the area, and the palm contour is detected after appropriate image morphology processing is conducted on the foreground image; a circumference sequence curve is used for conducting modeling on the palm contour, and the extreme point pair method is utilized for accurately distinguishing each finger contour and a wrist contour and building gesture feature sets; at last, the decision-making trees are used for training and recognizing the gesture feature sets with different finger numbers.

The invention relates to the technical field of PMSMs, solves a coupling problem during online identification of multiple parameters of a surface-mounted type PMSM, and achieves online decoupling identification of PMSM inductance, stator resistance and rotor flux linkage. Accordingly, the technical scheme adopted by the invention is that an online decoupling identification method of multiple parameters of the PMSM comprises the steps as follows: 1) identifying and coupling analysis of parameters of the PMSM; 2) a decoupling identification strategy, wherein voltage deviation before and after D shaft current injection is used for increasing the order of a motor mathematical equation, so that the decoupling identification of multiple parameters of the surface-mounted type PMSM inductance, the stator resistance and the rotor flux linkage are achieved; 3) neural network identifier design, wherein according to a parameter online identification problem of the PMSM, online identification is performed on motor parameters by adopting a self-adaptive neural network structure and a weight convergence algorithm based on a least mean square algorithm. The method is mainly applied to the design and manufacture of the PMSM.

The invention provides an image detection method and device, electronic equipment and a computer readable medium, and relates to the field of image recognition. The method comprises the steps: carrying out the multiscale processing of a to-be-processed image through a target skeleton network, and obtaining a multiscale feature image; processing the multiscale feature image through a first networkbranch based on an anchor and a second network branch which is not based on an anchor to obtain a first processing result and a second processing result, wherein each of the first and second processing results comprises: the position information of a target detection frame and / or the probability that the target detection frame comprises a target object; carrying out the fusion of the first processing result and the second processing result, and determining a target object detection result of the to-be-processed image based on a fusion result. The method solves a technical problem that the conventional image detection technology is lower in detection performances when the conventional image detection technology is used for detecting an image with the large-scale changes.

The invention relates to a multi-time-scale power system robustness scheduling system design method. The method includes: dividing a scheduling frame into three time scales: a day-ahead plan, a within-day rolling plan, and a real-time plan, and building a scheduling system; drawing the day-ahead plan based on next-day load prediction data and short-term prediction data of new energy on the basis of a known set initial state, a tie-line exchange plan, and the switch state of the day; drawing the within-day rolling plan based on ultra-short-term load prediction and ultra-short-term new energy power prediction with higher prediction precision on the basis of the day-ahead plan; and drawing the real-time plan based on the within-day rolling scheduling, and further correcting deviation of the scheduling plan and a predicted result. The method is advantageous in that the scheduling mode with multiple time scales is employed to gradually reduce the impact on the power grid by the uncertainty of the new energy, the robustness scheduling mode is employed in the day-ahead plan and the within-day rolling plan, and the robustness of the scheduling scheme is high.

The invention provides a nonlinear MIMO (multiple input multiple output) system-based decoupling control method and a nonlinear multiple input multiple output system-based decoupling control device. The device comprises an input module, a neural network module, a neural network inverse module, a control module and a delay module, wherein an output signal of the neural network inverse module is input to the control module and the neural network module; when the control module and the neural network module have the same input signal, output signals of the control module and the neural network module are input into an output module; the output module generates a disturbing signal according to the output signals of the control module and the neural network module; and after being delayed, the disturbing signal is input into the neural network inverse module, and after being processed by the control module, the disturbing signal is input into the output module. By the method and the device, the speed and the stability of decoupling control on a nonlinear MIMO system are improved.

The invention discloses a partition-based 3D printing filling path generation method. The method includes the following steps: determining an SLC file for entity processing, and a filing path spacing; generating contour offset paths, and saving the innermost contour offset path; determining a scan line angle; carrying out primary partition on a processing area according to the number of the intersection points of the scan line and the offset polygon of the innermost offset contour; carrying out second partition on the primary partitioned result according to the number of intersection points to obtain subarea sets; generating subarea paths according to the intersection points of the scan line and the offset polygon of the innermost offset contour and a relationship among the subareas; adjusting the subarea paths; and connecting remaining subpaths by using a spline curve to obtain the final inner filling path. The partition-based 3D printing filling path generation method has the advantages of strong robustness, adaptation to models with various shapes, easy realization, good versatility, easy embedding into 3D printing equipment, and commercialization realization.

The invention relates to a method for distinguishing false iris images based on robust texture features and machine learning, which comprises the following steps: preprocessing true iris images or false iris images; extracting the partitioned statistical features of a robust weighted partial binary pattern; and carrying out training and sorting of a support vector machine, and judging whether thetest images are false iris images or not according to the output result of a sorter. The method of the invention combines SIFT descriptors and partial binary pattern features to extract the robust texture features, the description of textures is more stable because of the robustness of the SIFT to brightness, translation, rotation and scale change, and the support vector machine enables the method to have better universality. The invention can be used for effectively distinguishing the false iris images, has the advantages of high precision, high robustness and high reliability, can be used for distinguishing false irises such as paper printing irises, color printing contact lenses, synthetic eyes and the like, and can improve the safety of the system when being applied to the applicationsystem in which iris recognition is used for carrying out identification.

The invention discloses a machine learning-based wireless sensing motion identification method which comprises the following steps: a step of data collection, a step of data denoising operation, a step of feature extraction and a step of SVM model training and identifying operation. During data collection operation, absolute value of a group of CSI data collected on each sampling point is obtained and is read into a 30*Nr*Nt matrix form. A PCA mode is mainly adopted for the data denoising operation. Feature extraction operation can be conducted based on discrete wavelet transformation. To make SVM model training convenient, training samples are subjected to Kmeans clustering operation via the machine learning-based wireless sensing motion identification method, n clustering centers can be used as word bags, and voting operation is performed based on feature vectors and best matching items of all the word bags; when the matrix form feature vectors are converted into column vectors, SVM model training can be realized conveniently. The machine learning-based wireless sensing motion identification method is a human body behavior identification method which is high in identification accuracy and high in robustness for environment change.

The present invention discloses a high-resolution depth map acquisition method based on active and passive fusion. Firstly, a TOF low-resolution depth map is subjected to sparse up-sampling and parallax map calculation, the low-resolution depth map generated by a TOF camera is mapped to the world coordinate system of a 3D scene, and a three-dimensional point is projected to a left color camera or a right color camera to form a TOF parallax lattice; the three-dimensional matching and TOF depth fusion are carried out, and the parallax map of a weak texture region, the parallax map of a texture region, and the parallax map of other regions are calculated. Compared with the prior art, the algorithm has a certain robustness, the advantages of a TOF depth camera and the matching algorithm of the color three-dimensional camera are integrated to make up the disadvantages of each algorithm, the good effects of real world scene and a standard data set can be displayed, the algorithm has good performance, and the algorithm has good application prospects in the fields of computer vision and robot application.