74results about How to "Efficient perception" patented technology

A media collection and management tool for collecting, managing, and configuring media and media related properties. The tool configures media along with related media properties as a program, wherein a program is a collection of properties. Configurable media may include video, text, images, 3D content, web page content and audio. In addition to media content, a program may be configured to include interactive regions in the form of hot spots, interactive mapping regions, and other interactive features. The media collection and management tool utilizes intuitive interfaces to configure programs. The media and programs can be exported to other media components, such as a digital document authoring tool.

The invention discloses a service flow aware system and method combining flow detection and package detection in the SDN, and relates to the field of the SDN. The service flow aware method includes the steps that when an initial novel service flow enters the SDN, table items not matched with the novel service flow in an equipment flow table are forwarded, the novel service flow is forwarded to a controller, a flow detection module carries out flow detection, a package detection module carries out package detection to recognize the service type and the service characteristics of the novel service flow, the controller triggers the awareness of a specific service according to the characteristics of the service flow, flow signs, flow statistics and package statistics, for current continuous service flow, the controller samples a real-time service flow, the dynamic variation of the real-time service is discovered through service awareness, and network strategies are adjusted correspondingly. The service flow aware system and method combining flow detection and package detection in the SDN combines the flow detection technology and the package detection technology, expands the southing port of the controller in the SDN, and can completely obtain the states of network flows of different industries and different applications in the current network.

A method and apparatus for improving the audibility of sound from a loudspeaker located close to an ear, comprises the steps of detecting ambient acoustic noise arriving from other sound sources using a transducer that provides corresponding ambient sound signals, inverting the polarity of said ambient sound signals and combining them with the signals being fed to the loudspeaker to reduce the audibility of said ambient acoustic noise, and passing said ambient sound signals through a filter having a predetermined average transfer function that compensates for the spectral modification of sounds travelling from the loudspeaker to the ear caused by the proximity of the ear of a listener in use.

A method of modifying perception of a malodor, such as the stench associated with putrefying flesh, by selectively affecting specific olfactory receptors in an individual is disclosed. Disclosed are articles of manufacture and especially innovative compositions effective in implementing the method of the present invention including at least three plant extracts.

The invention discloses a license plate positioning identification method based on a support vector machine and a convolutional neural network. The method comprises the following steps of using the weighted grayscale algorithm to carry out graying processing and histogram equalization on an original color license plate image so as to acquire an equalized grayscale image; carrying out edge detection on the equalized grayscale image and acquiring an edge image; carrying out binarization on the edge image; using a morphology to operate the license plate image so as to process and acquiring a license plate candidate area image; using a SVM discrimination model to classify and acquiring a license-plate accurate positioning image; using a vertical and horizontal projection combination method tocarry out image segmentation and acquiring a single character image; using a normalized function to carry out normalization processing on each license plate character; and using a convolutional neuralnetwork model license plate character identification model to identify and outputting and displaying an identification result. In the invention, the robustness of a license plate character identification algorithm is effectively increased, and the method has a high identification rate and is suitable for the license plate positioning and identification under a complex background.

An active semiotic system that is able to create implicit symbols and their alphabets from features, structural combination of features, objects and, in general sense, patterns; create models with explicit structures that are labeled with said implicit symbols, and derive other models in the same format by means of diagrammatic- and graph transformations. The invention treats vision as a part of larger system that converts visual information into special knowledge structures that drive vision process, resolve ambiguity and uncertainty via feedback projections, and provide image understanding that is an interpretation of visual information in terms of such knowledge models. Mechanisms of image understanding, including mid- and high- level vision are presented as methods and algorithms of said active semiotic system, where they are special kinds of diagrammatic and graph transformations. In the invention, the derived structure and not the primary view is a subject for recognition. Such recognition is not affected by local changes and appearances of the object from a set of similar views, and a robot or unmanned vehicle can interpret images and video similar to human beings for better situation awareness and intelligent tactical behavior.

A digital document comprising a multi-channel interface is provided that achieves improved user interaction. The digital document includes a plurality of content channels providing primary content continuously in a looping manner and at least one supplementary channel on a single page. The supplementary channel is configured to provide supplementary content upon the occurrence of an event during playback of the document. Channel content may include video, text, images, web page content and audio. The media content can be managed in a spatial and temporal manner. In addition to media content, a channel may contain interactive regions in the form of hot spots and interactive mapping regions. An authoring tool is also provided for generating a multi-channel interactive digital document.

The present invention discloses an on-line near-field payment system and method based on cell phone numbers. The system comprises a user terminal, a register server, a payment server, an IP address assignment authentication unit, an information collection unit and an identity server. A user first registers on the register server and the identity server, a payment request is sent to the register server when payment is required, payer identity and paying party authentication processes are completed through cell phone numbers, so as to carry out payment and realize on-line near-field payment. Users do not need to browse merchant websites, individual-to-individual and individual-to-merchant payment can be conveniently carried out, and the system and method are high in safety, simple to operate, and suitable for various consuming and account transferring environments.

A method for simultaneous performing visual servo and adaptive depth identification through a robot belongs to the computer vision and mobile robot technology field. The method comprises: obtaining an open loop kinematical equation with stable error according to a robot pose position polar coordinate representation method; and designing an adaptive updating rule capable of identifying depth information according to a concurrence learning strategy, and constructing a vision stabilization control rule of a mobile robot. The parameter adaptive updating rule designed by the invention can perform learning at the initial stage of the robot performing stabilization motion and perform online identification of the depth information in the robot motion process. The method for simultaneous performing visual servo and adaptive depth identification through the robot can prove the simultaneous convergence of the pose posture errors and the depth identification errors according to a Lyapunov method and a LaSalle invariance principle. The method for simultaneous performing visual servo and adaptive depth identification through the robot can accurately and reliability identify the depth information while the mobile robot completes the vision stabilization control so as to prove the simultaneous convergence of the controller and the identification module.

The invention provides a terminal switching method and a base station. The terminal switching method is used for intra-base-station terminal switching in an MME mode and comprises the steps that a BPOE of the base station receives a switching notification message; a target BPOE sends an ARP message to a service server to inform the service server of active updating of a saved ARP table; after the updating of the ARP table is accomplished by the service server, the target BPOE receives a message, sent to UE by the service server and cached in a source BPOE, from the source BPOE, and a message sent to the UE by the service server through the target BPOE is cached; the target BPOE sends the message received from the source BPOE to the UE firstly after receiving a switching completion message and then sends the message sent to the UE by the service server through the target BPOE to the UE. According to the terminal switching method and the base station, a transmission channel for switching can be provided fast, interruption of service during switching is avoided, and certain transmission rate is maintained.

A method for vision field computing may comprise the following steps of: forming a sampling system for a multi-view dynamic scene; controlling cameras in the sampling system for the multi-view dynamic scene to perform spatial interleaved sampling, temporal interleaved exposure sampling and exposure-variant sampling; performing spatial intersection to the sampling information in the view subspace of the dynamic scene and temporal intersection to the sampling information in the time subspace of the dynamic scene to reconstruct a dynamic scene geometry model; performing silhouette back projection based on the dynamic scene geometry model to obtain silhouette motion constraints for the view angles of the cameras; performing temporal decoupling for motion de-blurring with the silhouette motion constraints; and reconstructing a dynamic scene 3D model with a resolution larger than nominal resolution of each camera by a 3D reconstructing algorithm.

The invention discloses a face state recognition method and a face state recognition device based on deep learning. The method comprises the steps: collecting a face image of a recognized person, recording the collection time, carrying out the processing of the image of the recognized person through a face recognition algorithm, and outputting a face recognition result. Inputting the face recognition result into a pre-trained deep neural network and a face feature point model for processing to obtain an expression recognition result and a fatigue recognition result; wherein the expression recognition result comprises an expression type and a predicted value thereof; taking the expression recognition result and the corresponding acquisition time as expression data, and sequentially recording the expression data into an expression database; and recording the fatigue data into the fatigue database according to time; acquiring a plurality of data from the expression and fatigue database and analyzing the data to obtain a state recognition result of the recognized person. The emotion and fatigue degree of the recognized person can be efficiently sensed and analyzed. The method and the device can be widely applied to the automation fields of fatigue driving detection, robots and the like.

Poor visibility at night is a tiring and dangerous situation in traffic, feared by many drivers. As a result of poor visibility, the frequency of accidents at night is significantly higher than during the day with good visibility conditions. Cars are thus prospectively fitted with night vision systems, in order to increase safety in traffic. A night vision system used for this purpose generally comprises an illumination unit for illuminating the environment surrounding the vehicle, an image recording unit for collecting environmental data, and an image processing unit for evaluating the environmental data. So that the operation of the night vision system can be adapted to different situations, it must be designed in a flexible manner. To this end, the individual components of the night vision system must be able to be operated in different combinations. In order to achieve this, the components of the night vision system can be individually controlled by means of at least one control signal.

The invention provides a safety production index abnormity rapid sensing method based on Internet of Things and artificial intelligence, and the method comprises the steps: carrying out the hierarchical division of production processes based on a production process, and each layer comprises a plurality of production processes; calculating a domino index corresponding to each production process based on the accumulated deviation of the production processes and the level, and calculating the influence degree of each production process on the production result based on the domino index and the number of the production processes directly connected with the production processes; and selecting an effective production process from all the production processes based on the influence degree, performing production process abnormality sensing based on the actual deviation at the effective production process, and obtaining the position and the abnormality type of the abnormal production process. According to the method, the abnormity in the production process can be quickly and efficiently sensed only by monitoring related data of individual production processes, and the abnormity types and the abnormity positions of the production processes are judged.

The invention discloses a method for determining a beamforming weight. The method comprises the following steps: obtaining a beamforming weight corresponding to UE according to channel information ofthe UE; weighting first CSI-RS sent to the UE according to the beamforming weight to obtain a beamforming weight of the first CSI-RS; sending a second CSI-RS to the UE, wherein the second CSI-RS is obtained according to the beamforming weight of the first CSI-RS; receiving a precoding matrix index PMI fed back by the UE according to the second CSI-RS; according to the beamforming weight of the first CSI-RS and the PMI, determining the beamforming weight of the downlink service data of the UE. The invention also discloses a device and equipment for determining the beamforming weight.