53results about How to "Fix alignment issues" patented technology

The invention proposes a load ownership network evidence-obtaining method and system based on Bloom filters, and the method comprises the steps: grabbing a prepared evidence-obtaining network data flow, and carrying out the preprocessing of the network data flow; partitioning a load, and substituting partitioning contents into specified Hash functions, wherein the obtained result of each Hash function is a serial number of the corresponding Bloom filter; enabling the partitioning contents to be respectively stored in the corresponding Bloom filter according to the serial number of the corresponding Bloom filter; obtaining a field which is ready to be queried, carrying out the Hash operation of all partitioned blocks in the field one by one; judging whether all partitioned blocks in the field are respectively mapped to the position of the corresponding Bloom filter or not; and judging that the load comprises the field if all partitioned blocks in the field are respectively mapped to the position of the corresponding Bloom filter. The method can support the query of a wildcard character better, solves a problem of head block offset, a problem of alignment and a problem of continuity, reduces the false positive false alarm ratio at an acceptable data compression ratio, and improves the query speed and verification accuracy.

The invention provides an OAM entanglement modulation key distribution network system with real-time tracking compensation. The OAM entanglement modulation key distribution network system comprises anOAM-OAM entanglement generation unit, a first multiplexing module, a second multiplexing module, an OAM modulation unit, a Bob client and a coincidence measurement unit, wherein the OAM-OAM entanglement generation unit is used for generating an orbital angular momentum and an orbital angular momentum mixed entanglement quantum state; the first multiplexing module and the second multiplexing module are used for separating any OAM state; the OAM modulation unit is used for carrying out phase deflection modulation on the orbital angular momentum and loading coding information; and the coincidence measurement unit is used for performing coincidence measurement on signal light and idle light to decode the quantum bit information. According to the invention, the security of the quantum key distribution system is improved; an adaptive optical system is introduced, phase distortion caused by atmospheric turbulence is reduced, and the mode consistency during OAM light interference is guaranteed; an M-Z interferometer is used for separating any OAM, the quantum coding capacity is expanded, and an efficiently coded quantum key distribution system is obtained; and the engineering applicationrange is wide, the structure is simple, and the expandability is high.

The invention discloses a method for manufacturing a three-dimensional flexible stacked encapsulating structure of an embedded ultrathin chip. The method comprises choosing a flexible substrate with a single-layer metal layer, etching the single-layer metal layer on the flexible substrate to form a plurality of metal electrodes, conducting laser grooving on one side of the flexible substrate without the single-layer metal layer to form a plurality of grooves, conducting flipped thermocompression bonding with a plurality of chips on one side of the flexible substrate, provided with the plurality of grooves, thermally compressing the side of the flexible substrate, provided with the plurality of chips through thermocompression bonding and a flexible medium layer to embed the plurality of chips into the flexible medium layer, thinning one side of flexible medium layer embedded with the plurality of chips of the flexible substrate to obtain a thinned flexible medium module embedded with the plurality of chips, bending a part of the flexible medium module without the chips to stack the chips, encapsulating and curing a three-dimensional flexible encapsulating module with the plurality of stacked chips, and routing or reballing the encapsulated three-dimensional flexible encapsulating module.

The invention provides a sparse representation based incremental face recognition method which comprises the steps of: obtaining face key point information through face detection, determining face local block positions, extracting HOG features of each local block, constructing a local face dictionary for each local block according to a sparse representation classification method, and dynamically selecting part features according to a test picture during testing to generate a global face dictionary. For the face test picture, local block features are extracted, a test is carried out on the corresponding local face dictionary, the quality of face parts is judged according to an obtained local result, qualified parts are selected out according to a quality result, the features of the qualified parts in a training set are connected to construct the global face dictionary, global face features of the test picture are constructed, and an obtained global result is a final result. The whole training set is divided into N mutually exclusive subsets by type in incremental processing, processing is carried out in each subset to obtain a judgment result, and competition is carried out in all the subsets to obtain a final result.

The invention belongs to the technical field of power distribution network scheduling automation, and particularly relates to a method for fast state estimation of a power distribution network based on three pieces of measurement data. The method comprises the following steps of setting the taking time of SCADA measurement to be Eb+Tk in order to enable the sampling time of the SCADA measurement data to be consistent with the time of PMU data, wherein Eb is an expectation of SCADA measurement delay, and Tk is the occurrence time of the PMU data; adopting a linear extrapolation method to supplement high-precision pseudo measurement data in order to fuse PMU, SCADA and AMI measurement data; and establishing a mixed measurement state estimation model, computing a Jacobi equation H matrix, adding the active power and the reactive power for compensation decoupling, and obtaining a state estimation result after iteration solution. According to the method, a problem of data alignment of time-scale-free measurement and long-delay measurement is solved, the observability, the estimated accuracy and the computation convergence of the system are improved, and the problems of high mixed measurement dimension and slow convergence are solved.

The invention discloses a reflective photoplethysmography simulation calibration method and device. According to the method and device, simulation synthesis parameters are dynamically adjustable, standard simulation photoplethysmography containing multi-parameter physiological information such as pulse rate, respiration, pulse rate variability and oxyhemoglobin saturation serves as a detection envelope, and problems in calibrating the pulse rate, respiration, pulse rate variability and oxyhemoglobin saturation at the same time are solved; a pair of optical filters is used to separate infraredlight from red light emitted by a reflective photoplethysmography monitor to solve problems in identification of the infrared light and red light; an infrared light and red LED and an infrared light and red light PD of the calibration device are mounted in the same position to solve the problem in aligning the LED with the PD between the calibration device and the monitoror; and amplitudes of redlight and infrared light pulses emitted by a light emitting device in the calibration device are modulated to the standard simulated reflective photoplethysmography signal envelop according red lightand infrared light driving pulse waveform of the reflective photoplethysmography monitor itself, and the problem that different monitors are different in light emitting frequency and driving sequenceis solved.

The invention relates to a face aligning method based on a double-layer cascaded neural network. The method comprises the following steps that a double layer neural network model which comprises a first-layer network for detecting face contour characteristic points and a five sense organ area and a second-layer network for detecting characteristic points of the five sense organ area, and the fivesense organ area comprises an eyebrow and eye area, a nose area and a mouth area; and 2) and the double layer neural network model carries out 68 characteristic point detection on the picture to be detected, and face alignment is realized. Compared with the prior art, the method has the advantages of high detection precision and high adaption to complex background.

The invention discloses an end-to-end voice intention recognition method, and relates to the technical field of voice intention recognition. Most of existing voice intention recognition applications firstly acquire texts through voice recognition and then perform intention recognition, the accuracy of the intention recognition mode based on the texts seriously depends on the accuracy of translating the texts through voice recognition, and texts and pictures with inaccurate voice intention recognition exist. In order to solve the problem, the key points of the technical scheme are that the method comprises the steps: inputting a to-be-recognized voice, carrying out the noise reduction and feature extraction of the to-be-recognized voice, converting the to-be-recognized voice into a featurevector containing sound information, inputting the feature vector into a voice intention recognition model, and outputting a voice intention. The voice intention recognition model adopts a pre-training model thought of a deep learning network. According to the invention, the effects of reducing information loss caused by voice recognition and improving the voice intention recognition accuracy areachieved.

A liquid crystal display unit can realize optimization of contrast ratio and degree of saturation for penetration region and reflection region on colour liquid crystal display by adjusting thickness of colour filter for two transparent base plates at top and bottom. It comprises the first base plate with multifilm transistor for limiting reflection and penetration region, an insulation layer on the first base plate, the first colour filter on the insulation layer, a padded high insulation layer on part of the first colour filter, the first electrode covering the padded high insulation layer and part of the first colour filter, a liquid crystal layer on the first electrode, the second colour filter, the secone electrode and the second base plate.

The invention discloses a mobile positioning mechanism for a photovoltaic cell interconnected electrode, which comprises a lower positioning body and an upper positioning body, wherein the lower positioning body comprises a lower limiting groove, lower suction holes and a lower air hole, the lower limiting groove is arranged in the lower positioning body, the lower limiting groove is internally provided with the lower suction holes, the lower suction holes are communicated with the lower air hole arranged in the lower positioning body, and the depth of the lower limiting groove is smaller than the height of a flat welding strip of the photovoltaic cell; and the upper positioning body comprises an upper limiting groove, upper suction holes and an upper air hole, the upper limiting groove is arranged in the upper positioning body, the upper limiting groove is internally provided with the upper suction holes, the upper suction holes are communicated with the upper air hole arranged in the upper positioning body, and the depth of the upper limiting groove is larger than the height of the flat welding strip of the photovoltaic cell. The mobile positioning mechanism for the photovoltaic cell interconnected electrode has the advantages that the operation is simple; the stability is high; the applicability is strong; and problems of grasping and aligning for various welding strips with over small cross sections and strong flexibility can be effectively solved.

The invention discloses a parallel corpus aligning method. The method comprises the following steps: converting all the original text sentences in an original text and all the translated text sentences in a translated text into characters with a same encoding manner; carrying out word segmentation on the converted original text sentences in the original text and removing stop words therein so as to obtain content words; obtaining all the translated items of each content word of each original text sentence; matching all the translated items of each content word of each original text sentence in the converted translated text sentences of the translated text so as to obtain the similarity between each word content of each original text sentence and the translated text sentences; matching each original text sentence with the translated text sentences according to the similarity between each word content of each original text sentence and the translated text sentences so as to obtain the similarity between each original text sentence and the translated text sentences; and matching and aligning the translated text sentences having the highest similarity with the original text sentences with the original text sentences. The invention furthermore discloses a parallel corpus aligning device. According to the method and device, the problem of aligning between the original texts and the translated texts is solved.

The invention provides a character segmentation and recognition method based on a CTC deep neural network. The method comprises the following steps: a1, extracting features of an input image by usinga CNN; a2, carrying out the CELL segmentation on the features extracted in a1, fixing the height and width of CELL, and determining the number of CELL by the length of the image; a3, directly segmenting and classifying each CELL of the determined features, and outputting segmentation signals; a4, calculating the loss between the real segmentation signal and the segmentation signal output by the model by using CTCLOSS, feeding back the loss condition and training the whole model; a5, segmenting the text by using the segmentation signal output in the step a3, carrying out CNN + softmax classification identification on a single character, mapping a real segmentation signal from the annotated text, and automatically solving the text alignment problem by using the CTCLOSS. According to the invention, the OCR recognition speed is improved, and the recognition optimization is targeted after the characters are cut into single characters, so the final precision is improved; a recognition framework is improved, and a recognition process is separated into character segmentation and single character recognition, so optimization can be separately carried out in a targeted manner.

The invention provides a method and an apparatus for locating translation errors. The method includes: marking webpage labels of an original text and a translation text; differentiating words among all sentences in the marked original text, removing stop words among the sentences in the marked original text, and obtaining notional words; obtaining all translation items of each notional word among sentences in the original text; matching all translation items of each notional word of each sentence in the original text with all translation sentences in the marked translation text, and obtaining similarities between each notional word of each sentence in the original text and the translation sentences in the translation text; based on the similarities between all notional words of each sentence in the original text and the sentences in the translation text; and matching and aligning the sentences in the translation text that have the highest similarities with the sentences in the original text with the sentences in the original text. The method for locating translation errors of the invention addresses the problem of alignment between original text and translation text.

The invention relates to a wireless charging device based on an electric vehicle, and belongs to the field of wireless charging application of new energy electric vehicles. The wireless charging device comprises a vehicle position positioning layer and a charging unit, wherein the vehicle position positioning layer comprises a pressure sensor array used for positioning an electric vehicle parkingposition, the charging unit comprises a primary side coil, a horizontal driving mechanism and a controller, the controller is in control connection with the horizontal driving mechanism, the controller is in sampling connection with the pressure sensor array, and the horizontal driving mechanism is used for driving the primary side coil to move in the horizontal plane. According to the wireless charging device, a specific parking position of the vehicle is acquired according to the pressure distribution of the vehicle through the pressure sensor array, an actual position of the vehicle-mountedsecondary side coil is obtained through combination with a vehicle file, and then the controller controls the horizontal driving mechanism to achieve rapid and accurate matching alignment between theoriginal and secondary coil.

The invention discloses a method for forming an insulating structure and a semiconductor structure, in particular a method for forming a self-alignment structure of a passing gate. The method for forming the insulating structure comprises: a first step of providing a substrate, in which deep trenches filled with a silicon material and shallow trench isolators adjoined with the deep trenches are arranged inside the substrate, and a patterned pad oxide layer and a hard mask which define openings of the deep trenches together are arranged on the substrate sequentially; a step of oxidization, in which first oxide layers are formed on the surfaces of the silicon material in the deep trenches to serve as the insulating structure of the passing gate; a step of forming first silicon layers in the openings to cover the first oxide layers; and a step of removing the hard mask.