607results about How to "Strong correlation" patented technology

The objective of the present invention is to provide a method and an apparatus for automatically generating a poem. The method comprises: acquiring a poem genre corresponding to a to-be-generated poem and a subject heading that is selected for each to-be-generated line in the to-be-generated poem; according to the poem genre, determining a poem model based on a concentration mechanism; according to the subject heading of the to-be-generated line and all historical reference lines that have been generated before the to-be-generated line is generated, generating the to-be-generated line by using the poem model; using the generated to-be-generated line as a historical reference poem; and repeating executing the operations of generating the to-be-generated line and using the generated to-be-generated line as the historical reference line, until generating the to-be-generated poem. Compared with the prior art, the subject heading is allocated for each to-be-generated line by the method, and the subject heading is involved in the generation process of the whole to-be-generated poem, and when generating each to-be-generated line, all generated lines in the to-be-generated poem are used to improve logical associations between lines and enhance usage experience of users.

A cine imaging filter and method of use that includes a denoising image-filter based on the Karhunen-Loeve transform along the temporal direction to take advantage of the high temporal correlation among images. The cine imaging filter may further include the application of a simple formula describing the quantitative noise reduction capabilities of the KLT filter as a function of eigenimage cutoff. Additionally, the filter may validate its accuracy in numerical simulation and in in-vivo real time cine images. Furthermore, exemplary embodiments of the cine imaging filter may employ a technique to automatically select the optimal eigenimage cutoff to maximize noise reduction with minimal effect on image information.

The invention discloses an audio lossless compression encoding and decoding method and belongs to the field of information source encoding and decoding. In the method, a signal is framed adaptively according to related conditions of the former frame and the later frame of the signal, wherein a frame signal after framing is a combination of signals with similar signal characteristics, so that an encoder can achieve higher compression efficiency and is good for later integer wavelet transform and linear predictive encoding. For lossless compression encoding, a signal can be completely reconstructed, and the complete reconstructed characteristic of the signal is guaranteed by adopting integer lifting wavelet transform. Compared with the prior art, after a related adaptive framing-based module and an integer lifting wavelet-based de-correlation module are introduced, redundant information in an original signal can be well de-correlated, and the generated compressed data contains less redundant information. Therefore, the audio lossless compression encoding and decoding method can greatly improve the compression ratio by low computation complexity.

The invention provides an e-commerce search scoring and ranking method. The method includes the steps of 1 building a keyword lexicon of e-commerce platform commodities, 2 calculating net scores of significance factors of the commodities, 3 extracting attribute information in the commodities and storing the attribute information into indexes, 4 building series query blocks according to relevance and significance, 5 selecting corresponding search results according to paging parameters and conducting ranking. According to the e-commerce search scoring and ranking method, according to the characteristics of e-commerce search, central keywords of the commodities are recognized, and preferential matching is executed during query; after query and search keywords input by a user are received, semantic analysis is conducted, and query statements of different relevance and significance are built due to the fact that query domains are different and indexes such as commodity page view are different. When the user searches for commodities, different queries are executed according to difference of query relevance and significance, and therefore search results are returned rapidly.

In an image forming apparatus, in accordance with formation of paper, there are controlled parameters; that is, a color conversion parameter pertaining to a color conversion section 23 of a drawing section 22; a parameter pertaining to a correction curve of an TRC.gamma correction section 24 or a parameter pertaining to the number of screens or a growth pattern of a screen processing section 25 provided in a printer engine control section 12; parameters for use in adjusting electrification, exposure, and a voltage to be used during a developing operation, which are employed in a printer engine 13; a parameter for adjusting a voltage or current during a transfer operation; and a parameter for adjusting a fixing temperature or pressure. Parameters are controlled so as to reduce inconsistencies, which would otherwise arise in paper of poor formation. Thus, picture quality of an image is improved while inconsistencies are made less noticeable.

The invention discloses a content popularity prediction method based on depth learning under SDN architecture. The method comprises the steps that each node calculation resource and link in an SDN network to construct a reconfigurable and distributed depth learning network; the SDN network has an SDN controller and a number of SDN switch nodes; each SDN switch realizes the calculation function of a number of neurons; the neurons are connected with each other through the SDN switch link; each node in the SDN network collects space-time co-distribution data with requested content in real time and uses the data as the input of the depth learning network; a stack-type self-coder is used to carry out characteristic learning on the space-time co-distribution data; and a Softmax classifier is used to predict the content popularity. According to the invention, the reconfiguration of the depth learning network is realized based on the characteristics of programmable and global view centralized control SDN; the number of hidden layers and the nodes of each layer of neuron can be adjusted; and the method is very critical to an ICN analysis system, and can help dynamic routing and caching decisions in the system.

The invention discloses a method and system for evaluating the health state of a battery. The method comprises: S1, acquiring life decay data of a plurality of batteries of the same type; S2, according to the life decay data, constructing target characteristic parameters of the batteries and calculating a capacity data sequence; and S3, on the basis of a least square method, fitting of the capacity data sequence and the target characteristic parameters is carried out to construct a health state assessment model for calculating effective capacities of the batteries. The target characteristic parameters include at least one of the following parameters: an equal-voltage-drop discharging ampere-hour value, an equal-ampere-hour discharging voltage difference, an equal-voltage-drop charging ampere-hour value, and an equal-ampere-hour charging voltage difference. According to the invention, no complicated testing condition or model as well as no complete charging and discharging voltage curves on the constant current condition of the battery is needed; and the health state of the battery can be predicted accurately.

The invention discloses a constructing method and a dynamically updating method of a GNSS network difference positioning reference station network. In the method, each GNSS reference station is projected to a two-dimensional plane according to the known and accurate coordinates of the GNSS reference station; a reference station network taking a triangle as a basic structure is formed by the discrete points of the two-dimensional plane according to the network constructing rule of a Delaunay triangle network, and the shape of the network is unique. When a new reference station is increased or a certain reference station is not used, the real-time construction of a new network is finished through operation of point insertion or point deletion. The invention can keep the unique optimized network shape of the reference stations and meet the precision requirement when a movable station is positioned.

A cine imaging filter and method of use that includes a denoising image-filter based on the Karhunen-Loeve transform along the temporal direction to take advantage of the high temporal correlation among images. The cine imaging filter may further include the application of a simple formula describing the quantitative noise reduction capabilities of the KLT filter as a function of eigenimage cutoff. Additionally, the filter may validate its accuracy in numerical simulation and in in-vivo real time cine images. Furthermore, exemplary embodiments of the cine imaging filter may employ a technique to automatically select the optimal eigenimage cutoff to maximize noise reduction with minimal effect on image information.

An information retrieval apparatus for searching a set of information items. The apparatus comprises a mapping processor, a graphical user interface, a user control and a search processor. The mapping processor is operable to generate data representative of a map of information items from a set of information items. The map provides the information items with respect to positions in an array in accordance with a mutual similarity of the information items, similar information items mapping to similar positions in the array. The graphical user interface displays a representation of at least some of the information items, and the user control is provided for selecting an information item. The search processor is operable to perform a related search with respect to the user selected information item by identifying information items which correspond to neighboring position in the array with respect to the array position corresponding to the user selected information item. Since the search processor is arranged to identify information items in the search from the array rather than by searching the information items for some characterizing information feature, such as a keyword, searching for information items of interest can be effected with a reduced complexity.

An X-ray image analyzing system, including:an X-ray source for radiating an X-ray; an X-ray detector for detecting an X-ray image radiated onto an x-ray image detection surface, wherein phase contrast X-ray simple imaging is capable of being performed determining a trabecular bone index computing region, computing a trabecular bone index indicating a state of a trabecula from image data in the trabecular bone index computing region, determining a bone-flesh boundary index computing region by a second region determination method different from the first region determination method, and computing a bone-flesh boundary index indicating smoothness of a bone-flesh boundary from image data in the bone-flesh boundary index computing region.

The invention provides a mixed recommendation system and method for recommending application programs to an intelligent device. The mixed recommendation system based on a distributed architecture of a cloud platform comprises a log leading-in and analysis device, a user modeling device and a mixed recommending device, wherein the log leading-in and analysis device leads in an application program access log from a database, analyzes the leading-in application program access log and extracts information about the application program use situation, the user modeling device is used for creating an application program list comprising a plurality of application program items, using the information extracted by the log leading-in and analysis device, and conducting user modeling for users in a distributed mode based on the MapReduce programming model, and the mixed recommending device generates item-based first candidate recommendation data of the collaborative filtering algorithm, second candidate recommendation data recommended by Top-N, and content-based third candidate recommendation data of the collaborative filtering algorithm for each user, and then generates mixed recommendation data according to the preset sum of recommended items, the first proportion, the second proportion and the third proportion.

The invention relates to a Bayer image compression method based on YUV conversion, which belongs to the technical field of an aerospace optical remote sensor and is a Bayer image compression method used for an aerospace optical remote sensing CMOS camera. The method comprises the following steps: carrying out space conversion on a Bayer-format image from a three-color space to a brightness-chroma space, eliminating the relativity among the three color components to be ready for different compression ratios used by brightness and chroma, self-adaptively determining the compression ratios based on the complexity of the image, carrying out line conversion and encoding. The invention greatly improves the quality of the compressed image by carrying out space conversion to eliminate the relativity among the three color components, and further enhances the adaptive capacity of the camera by using different compression ratios for images of different complexities.

The invention provides a comment information display method and device, electronic equipment and a computer readable medium. The comment information display method comprises the following steps: displaying preview information of a first multimedia resource; in response to a first trigger operation of a user on the preview information, sending a calling request for the first multimedia resource toa server; receiving the first multimedia resource returned by the server, and receiving comment information associated with the first multimedia resource returned by the server; screening out one or more pieces of comment information from the plurality of pieces of received comment information; and displaying the first multimedia resource, and displaying the one or more pieces of comment information in a bullet screen form on a display interface of the first multimedia resource. Compared with the prior art, the method has the advantage that the user can check the comment information related tothe multimedia resource more conveniently, quickly and visually.

The invention provides a feature extraction and matching method in visual navigation. According to the method, a feature region is extracted by adopting a two-dimensional maximum entropy threshold segmentation method; an external square region of the feature region is subjected to image normalization processing; each external square region subjected to the image normalization processing is used as a feature region to be matched; a feature vector of a feature point is obtained; each feature point in each feature region in a first image is matched with feature points in each feature region in a second image, and the number of matching points is obtained; and two feature regions with the most matching points are selected as matching regions, wherein matched feature points are used as matching feature points. According to the scheme, the number of the matching points is obtained through feature point matching; the feature regions with the most matching points are obtained by using the number of the matching points as a screening condition; the matching of the feature regions is completed; the practical matching pair is further obtained; and the matching points of navigation images with fine, dense and single textures under the condition of generating affine transformation such as shear transformation is increased.

The embodiment of the invention provides a method and a mobile terminal for searching keywords in mobile terminal information. The method comprises the following steps: receiving a message; dividing the message into words, and acquiring the words in the message; matching the acquired words in the message with the search keywords in a search engine; after successful matching, marking the keywords; and selecting the marked keywords to enter a search result interfaces of the keywords. The embodiment of the invention has the technical effect that the keywords in the received message can be directly selected to display the search result interfaces of the keywords in the message. Users can directly click information browsing interfaces to search needed or interested contents, the relevance is very strong, and the operation is very convenient.

The invention provides an ancient poetry generation method based on image content. The method comprises the steps that 1, target detection is conducted on image content objects based on a single-time multiple-frame target detection frame, and the names of the objects are obtained; 2, word separation is conducted on a set number of Tang poetries by using the Chinese lexical analysis tool, and a glossary is obtained, feature learning is conducted on each word in the glossary by using the word2vec tool, and each word is mapped to a vector space; 3, the obtained object names are inputted into the word2vec tool, a mapped vector of the object names is obtained, and the cosine similarity between the mapped vector and an ancient poetry vector is calculated, after a threshold is set, part of the object names selected are used as the corresponding subject term of the objects; 4, the subject term is utilized to expand the keywords, after the keywords are inputted and the Tang poetries are learned, a RNN model is obtained, and the ancient poetries are generated. According to the ancient poetry generation method, the method of target detection and the method of cyclic neural network algorithm are combined to generate ancient poetries, and the correlation between ancient poetries and image content is strong, and better coherence is achieved.

An object of this invention is to get a circuit board and a semiconductor module with high endurance against thermal cycles, and which is hard to be broken under thermal cycles, even if thick metal circuit board and thick metal heat sink are used, corresponding to high power operation of semiconductor chip. This circuit board comprises, an insulating-ceramic substrate, a metal circuit plate bonded to one face of the insulating-ceramic substrate, a metal heat sink bonded to another face of the insulating-ceramic substrate, wherein (t12−t22) / tc2 / K<1.5, here, thickness of the insulating ceramics substrate is tc, thickness of the metal circuit plate is t1, thickness of the metal heat sink is t2, and internal fracture toughness value of the insulating ceramics substrate is K.

Disclosed is a ferritic stainless steel sheet with excellent thermal fatigue properties, including, by mass %, 0.03% or less of C, 1.0% or less of Si, 1.5% or less of Mn, 0.6% or less of Ni, 10˜20% of Cr, 0.05˜0.30% of Ti, 0.51˜0.65% of Nb, 0˜less than 0.10% of Mo, 0.8˜2.0% of Cu, 0˜0.10% of Al, 0.0005˜0.02% of B, 0˜0.20% of V, and 0.03% or less of N, with the balance being Fe and inevitable impurities, having a composition satisfying the following equations (1) and (2) and having a structure in which ε-Cu phase grains each having a long diameter of 0.5 μm or more are present in a density of 10 or less per 25 μm2: Nb−8(C+N)≧0 . . . (1), 10 Si+20 Mo+30 Cu+20(Ti+V)+160 Nb−(Mn+Ni)≧100 . . . (2). The ferritic stainless steel sheet, having a relatively inexpensive component composition, has excellent thermal fatigue properties, and is suitable for use in an automotive exhaust-gas path member, including an exhaust manifold, a catalyst converter, a front pipe, or a center pipe.

The embodiment of the invention discloses a compressive sensing based dynamic clustering wireless sensor network data collecting method. The compressive sensing based dynamic clustering wireless sensor network data collecting method comprises the steps of obtaining one or more event source positions in a wireless sensor network, determining a sensor node nearest to each event source position as a cluster head and clustering the sensor nodes in the wireless sensor network with each cluster head as center; collecting data in each cluster, wherein the data sensed by the sensor nodes in each corresponding cluster are collected by means of a compressive sensing method. The embodiment of the invention discloses further provides a compressive sensing based dynamic clustering wireless sensor network data collecting device. The compressive sensing based dynamic clustering wireless sensor network data collecting device and method can use an event source as the center to perform clustering, utilize the compressive sensing technology to collect the data in the clusters, perform dynamic clustering on the wireless sensor network according to position change of the event source and accordingly enable the data collected in the nodes in the clusters to be high in dependency, transmission times can be fewer at the same accuracy, and accordingly the service life of the network is prolonged.

The present invention discloses a time correlation learning neuron circuit based on a resistive memristor and an implementation method thereof. The present invention utilizes switching characteristics of the resistive memristor. When two terminals of the resistive memristor are selected synchronously by two excitation signals, the voltage drop between these two terminals will change the resistance value of memristor, thereby achieving the on-off of a synapse connection and achieving the correction of the two excitation signals. Meanwhile the device also has a memory characteristic. Also, the previous excitation signal can be repeated. That is, the purpose of learning is achieved. Since the resistive memristor has a simple structure and a high degree of integration, it can achieve large-scale physical synapse connection in order to achieve more complex learning and even logic functions. The present invention has a good application prospect in a neuron cell computation.

The invention provides a searching method, a server and a system. The searching method comprises the steps as follows: a search target document is received; at least one subject key word in the search target document is extracted; related documents are searched according to the at least one subject key word, and at least one related document of the search target document is acquired; and at least one related document is returned to. The searching method, the server and the system can realize related document searching taking documents as searching conditions.

The present invention is applicable to the field of network information processing and provides a field subject-oriented Web news dynamic aggregation method. The method comprises the following steps of: for a field site list pre-defined by a user, according to a subject provided by the user, acquiring a search record list by utilizing a vertical search engine and a meta search engine; performing deduplication on the search record list and identifying a Web news webpage to obtain a news webpage search record list; with a Web information extraction method, acquiring a structured news list from the news webpage search record list; and performing ranking on the structured news list according to a field model to obtain an ordered structured news list, and returning the ordered structured news list to the user as a dynamic aggregation result. According to the field subject-oriented Web news dynamic aggregation method, a multi-source associated Web news set is acquired in real time according to a field and a subject provided by the user, an interactive mechanism where the ranking of the Web news depends on the popularity of the Web news is provided, and the purpose is to provide a convenient and efficient internet information acquisition and sharing mode.

The invention discloses a medical standard term ontology management system and method, a device and a storage medium, belonging to the technical field of life and health. The system comprises a disease terminology library, a terminology inquiry module, a terminology customization module and a foundation platform construction module, wherein the disease terminology library displays the collected data in a tree structure, and simultaneously displays the details of the terminology, synonyms and parent-child node information; the term query module is used for creating term index, constructing termquery system, and realizing term fast positioning. The terminology customization module is used for establishing the terminology system belonging to different departments by relying on the disease terminology database and combining the practical application of different departments' business systems. The basic platform construction module is used for managing different organizations in the medical union to manage users of different organizations. The system has a uniform classification definition, can meet the needs of the development of medical and health field and industry in our country, and can be maintained and updated in a timely manner.

The invention belongs to the image processing field and relates to a stereo image quality evaluation method based on GSSIM. The method includes the steps: (1) for left image and right image, gradient structure similarity values are respectively solved, the average of the two values is solved, so as to obtain stereo image quality evaluation value QE; (2) the following method is adopted to carrying out image stereoscopic perception objective evaluation: absolute difference image of the original image and processed image is calculated, Mu1 and Mu2 of the absolute difference image are solved; Sobel operator is used for solving the gradient amplitude image of the absolute difference image; filtering is carried out on the absolute difference image of the original image, and original image binocular parallax distribution situation is calculated; Dl(x, y), Dcg(x, y) and Dsg(x, y) values at binocular parallax position are solved; DSSIM value at the binocular parallax position is calculated; and image gradient structure similarity namely image stereoscopic perception objective evaluation value DE is calculated. The invention can be well applied to stereo image quality evaluation, and correlation of objective evaluation result and subjective evaluation result is strong.

The invention provides a method for detecting a hearth radiation energy signal and a method for controlling boiler combustion by using the same, and belongs to the field of combustion monitoring and controlling. The method for detecting the hearth radiation energy signal comprises steps of analog-to-digital conversion, color value extraction and hearth radiation energy signal calculation. In the method for controlling air quantity of hearth combustion, the optimum oxygen quantity calculated by using the hearth radiation energy signal, set oxygen quantity and set offset are added, and the sum is sent to an oxygen quantity regulator to control the air quantity of the hearth combustion. The method for controlling superheater attemperating water comprises the steps of combining the hearth radiation energy signal with a heat signal to form a heat value signal; and introducing the heat value signal to a heat regulator to control the superheater attemperating water. The radiation energy signal detected by the method has the advantages of stability, reliability, capability of accurately reflecting the combustion situation in the hearth space in real time, and strong correlation with unit operation parameters; and when the new heat value signal constructed by the radiation energy signal is introduced to air supply and attemperating water control loops of the unit, the control effect of the unit can be optimized, and the efficiency of the unit can be improved. Therefore, the method for controlling the boiler combustion is suitable for combustion optimization control of a pulverized coal fired boiler of a thermal power generating unit.

The invention provides a voiceprint recognition method based on RNN. According to the voiceprint recognition method based on RNN, after MFCC features of denoised voice data and first-order and second-order differences are obtained, the advanced features of a speaker in the MFCC features are extracted by using a recurrent neural network, the extracted features are classified by using a softmax classifier, and finally, the speaker is recognized by using a naive bayes method. Different from mute elimination of a traditional method, the method provided by the invention has the advantages that a mute section in the voice data is reserved, the features related to the context can be extracted on the basis of the recurrent neural network, the advanced features of the voice, such as the speaking mode and the rhythm of the speaker can be extracted aiming at the voice data, so that the feature information is complete, and the speaker can be well represented. Compared with an existing voiceprint recognition method based on Gauss, the method provided by the invention has the advantages that the requirements for voice data are relatively low, the accuracy is higher, the accuracy is still kept ata very high level even when big data is aimed, and the running speed is not reduced.

The invention provides a wireless communication voice quality assessment device, a system and a method thereof, wherein the device comprises: a radio link parameter obtaining unit which is connected with a comprehensive assessment unit to output the obtained radio link parameter to the comprehensive assessment unit; a voice coding-decoding parameter obtaining unit that is connected with the comprehensives assessment unit to output the obtained video coding-decoding parameter to the comprehensive assessment unit; and a radio process parameter obtaining unit that is connected with the comprehensive assessment unit to output the obtained radio process parameter to the comprehensive assessment unit, wherein the comprehensive assessment unit can implement a voice quality comprehensive assessment and output assessment result according to the input radio link parameter and the input voice coding-decoding parameter and / or radio process parameter. In the invention, the assessment is carried out based on more comprehensive influencing factors, and the correlation to PESQ assessment is stronger.

The invention discloses a method for classifying texts on the basis of nave Bayes. The method includes steps of 1, forming feature vectors of the to-be-classified texts by the aid of word segmentation tools, comparing the feature vectors to common words, removing meaningless words in the to-be-classified texts and setting weights w for each word s which appears in the to-be-classified texts; 2, acquiring probability sets Q (w<1>, ..., w<n>) of P (w<1>, ..., w<n>) which appear in training text sets D, and multiplying attributes in the Q (w<1>, ..., w<n>) to obtain prior probability P (w|D) of the P (w<1>, ..., w<n>) which appear in the training text sets D; 3, dividing the quantities of files in the training text sets D by the total number of integral training text sets to obtain prior probability P (D), multiplexing the P (D) by P (x|D) to obtain posterior probability P (D|w) of the P (w<1>, ..., w<n>) in the training text sets D; 4, repeatedly carrying out the steps 2 and 3 and computing all posterior probability; 5, comparing results obtained at the step 4 to obtain the maximum posterior probability P (D). The P (w<1>, ..., w<n>) belong to D categories. The method has the advantages that the method is good in performance and has excellent practical application values in classifying appeal texts of electric power users.