660results about How to "Improve search accuracy" patented technology

Disclosed is a speech search device including a recognizer 2 that refers to an acoustic model and language models having different learning data and performs voice recognition on an input speech, to acquire a recognized character string for each language model, a character string comparator 6 that compares the recognized character string for each language models with the character strings of search target words stored in a character string dictionary, and calculates a character string matching score showing the degree of matching of the recognized character string with respect to each of the character strings of the search target words, to acquire both a character string having the highest character string matching score and this character string matching score for each recognized character strings, and a search result determinator 8 that refers to the acquired score and outputs one or more search target words in descending order of the scores.

The present invention is directed to systems and methods for encoding and retrieving information from a variety of sources using novel search techniques. The systems and methods of the invention are capable of extracting all types of structural and relational information from a query or a source data allowing for the recognition of subtle differences in meaning. The capability of discerning subtle differences in meaning that are beyond the search systems and methods presently available, the invention described herein is capable of repeatedly providing accurate and meaningful responses to a diverse set of queries. Particular embodiments of the present invention include walkie-talkie-type telephone interfaces where the user may speak directly to the present invention and receive a spoken response relevant to any provided enquiry.

The present invention is directed to systems and methods for encoding and retrieving information from a variety of sources using novel search techniques. The systems and methods of the invention are capable of extracting all types of structural and relational information from a query or a source data allowing for the recognition of subtle differences in meaning. The capability of discerning subtle differences in meaning that are beyond the search systems and methods presently available, the invention described herein is capable of repeatedly providing accurate and meaningful responses to a diverse set of queries.

The present invention is directed to systems and methods for encoding and retrieving information from a variety of sources using novel search techniques. The systems and methods of the invention are capable of extracting all types of structural and relational information from a query or a source data allowing for the recognition of subtle differences in meaning. The capability of discerning subtle differences in meaning that are beyond the search systems and methods presently available, the invention described herein is capable of repeatedly providing accurate and meaningful responses to a diverse set of queries.

The invention provides a search method and device based on a query word. The method comprises the following steps: searching a query word which is input by a user through a client, and selecting a search engine suitable for the query word from multiple search engines by combining the query word with a query word attribute library, wherein the query word attribute library is used for characterizing the correlativity between each query word or query word class and each search engine; and acquiring the search result information of the selected search engine, processing the search result information and displaying the processed search result information. In the invention, when multiple search engines exist, by means of carrying out intention comprehension and analysis on the query word of the user, the search engine related to the search user requirement can be selected or the search engine with higher accuracy can be searched based on the intention of the user, and pertinent link search is carried out, thus the system efficiency can be improved, and the search accuracy of the user is also increased.

The present invention is directed to systems and methods for isolating sentence boundaries between sentences in text. Sentences of the normalized document feeds or source text are separated by determining boundaries between individual sentences, by a Bayesian algorithm, that has been seeded with rule frequencies, developed from a previous training phase, that employed a text of sentences with marked boundaries between the sentences.

A search system can include a server, a token assignment unit for assigning types of tokens based on different kinds of character string analysis methods, an index generating unit for generating an index list that associates the tokens assigned with the token assignment unit, a type identification value for identifying a type of the character string analysis, and information, a search unit that receives a search word for referencing the information to combine types of search tokens generated from the search word to generate a single search command for parallel inquiry of the information to search for the information, and a search result generating unit for displaying information extracted in relation to the search word through parallel inquiry with the search unit and search tokens so as to identify the tokens.

The invention is directed to the provision of a moving image data playback method, and more particularly, an MPEG data playback method that can randomly access MPEG data stored at a VBR. The playback method comprises: storing in a table an average bit rate computed from time information at the beginning and end of stored moving image data; detecting and storing the time information and position information of randomly accessible picture groups one at the beginning of the moving image data and the other at the end thereof, and computing a picture group unit time from the number of frames between adjacent picture groups and storing the result in the table; computing a point to be accessed, based on the average bit rate and other information stored in the table; and starting playback from the computed point to be accessed, and storing in the table the time information and position information of the picture groups detected during the playback and the average bit rate between each picture group computed from the detected time information and position information.

The invention discloses a keyword based evaluation expert intelligent search and recommendation method. The keyword based evaluation expert intelligent search and recommendation method specifically comprises step 1, segmenting an expert information main text into substring sequences, performing ICTCLAS word segmentation of Chinese academy of sciences and performing stop word filtering on the result of the word segmentation to obtain the word collection; step 2, extracting feature words of the expert information according to fields; step 3, building an expert knowledge representation model based on the fields and the weight of the feature words and establishing an expert information index database; step 4, performing automatic prompting according to a search term thesaurus when a user inputs keywords and meanwhile performing real-time update on the search term thesaurus through a search term counter; step 5, calculating the search relevance between the keywords and the expert information based on the semantic information and the like; step 6, listing relevant experts from high to low according to the matching degree. According to the keyword based evaluation expert intelligent search and recommendation method, the intelligent full-text search and recommendation of the expert information can be achieved through the keyword input and accordingly the experts which are matched with a pended science and technology project can be searched out accurately.

The invention relates to the field of optical imaging, and discloses a method and device for focusing a camera. In the method and device for focusing the camera, the sizes, positions and priorities of two focusing windows can be set to solve errors of a lens structure, so that the searching speed and accuracy of the optimized focusing position can be improved; the step length of a motor is reduced to a clear zone in a gradient way in the focusing search, so that the focusing speed is improved; rolling focusing is carried out at least twice in a microstep speed reducing way in the clear area, so that the focusing accuracy is completed in the clearest position of an image; and focusing is started at a regular moment when the focusing is fuzzy, so that focusing is effectively ensured to be not focused once again in a scene with contrast ratio. When the optical lens is in an adaptive step length with different magnifications and corresponding settings, different focusing modes can be selected for quick focusing, so that the method can be used for effectively determining whether to start focusing and focusing direction or not in various zooms and scenes, identifying whether the image is in a fuzzy state or not and quickly finding out the clearest point.

The invention discloses an objective tracking system and an objective tracking method in a camera-distributed map. The object tracking system comprises a monitoring operation client and a video calculating server. The monitoring operation client inputs an image of a suspected object and sets a to-be-searched range through the camera-distributed map. The video calculating server performs processing on a video and an image which are obtained through sampling. The object tracking method comprises the steps of extracting color characteristic of the image of the suspected target, detecting and tracking the target in a video scene, recording motion information of the target, and then transmitting a corresponding calculating result to the monitoring operation client; on the map in which camera nodes are marked by the monitoring operation client, acquiring the path track of the suspected object on the map according to the camera nodes where the suspected object appears, and simultaneously collecting information such as advancing speed and state of the target, thereby facilitating analysis by safety monitoring personnel. The objective tracking system and the objective tracking method have advantages of greatly reducing work difficulty in observing a screen for searching, making personnel concentrate on effective video segments, and realizing relatively high application value at a scientific police improvement aspect.

The present invention discloses a keyword associated information generating system and method, and includes a user interface and an associated information server, the input keyword of the user can be timely obtained through the user interface, and the associated information fed back by the associated information server can be displayed; the index file in the associated information server for storing keywords and the searching result data can be queried according to the timely obtained input keywords, so as to retrieve the keyword whose front part is matched with the input keyword and the searching result data. The matched keyword and the searching result data are used as the associated information of the input keyword to feed back to the user interface. The present invention is utilized to feed back the searching result data of keyword which is used as a part of associated information to the user, so as to enable the system to associate the searching intention of the user automatically and more correctly to improve the searching accuracy.

The invention discloses a searching method and searching equipment. The searching method comprises the following steps: receiving a search keyword input by a user, acquiring at least one search document relating to the search keyword based on the search keyword, and determining the semantic vector of the search keyword; according to the semantic vector of the search keyword and the semantic vector of each search document, calculating the semantic correlation between each search document and the search keyword; according to the semantic correlation, sequencing the search documents and outputting the search documents to the user according to the sequencing result. The semantic vector of the document of a search engine is acquired according to the semantic vector of a document tile, the semantic vector of a document identifier and the semantic vector of a document category, so that the semantic information of the document can be expressed accurately, the precision of the semantic correlation between the document acquired by calculation and the search keyword is improved, the search precision of the search engine is improved, and the user experience on the search engine is improved.

The invention provides a method for realizing searching by utilizing a client and a search client, wherein the method comprises the following steps: the search client obtains a user inquiry request; the search client calls at least two search engines to search the inquiry request; the search client obtains the search result information of each called search engine; and the search client processesthe search result information so as to obtain a final search result, and then displays the final search result. The invention has the advantages that trough realizing operations such as user intent analysis, search engine selection, linking and downloading, and analysis process, and the like on the client, complicated calculations and broadband resource occupation are transferred to the operatingterminal of the user, thereby effectively utilizing the terminal resource of the user, reducing the occupation on server resources, and improving the searching speed and efficiency.

The present invention is directed to systems and methods for encoding and retrieving information from a variety of sources using novel search techniques. The systems and methods of the invention are capable of extracting all types of structural and relational information from a query or a source data allowing for the recognition of subtle differences in meaning. The capability of discerning subtle differences in meaning that are beyond the search systems and methods presently available, the invention described herein is capable of repeatedly providing accurate and meaningful responses to a diverse set of queries.

The invention discloses a cloud environment task scheduling method based on an improved ant colony algorithm. On the basis of the conventional ant colony algorithm, a rest-life-pheromone (RLP) concept is introduced; by comparing the RLPs of ant brothers in the searching process, repeated search can be effectively eliminated, and searching accuracy is improved; and by adoption of a method for computing the quantity of the ant brothers, the quantity of ants can be dynamically changed under the specific conditions of loads and tasks of cloud environment, so a system is not overloaded, and the searching accuracy is ensured. Compared with the prior art, the cloud environment task scheduling method based on the improved ant colony algorithm has the advantages that requirements of a user on task execution time can be met to the greatest extent, and the allocation efficiency and execution efficiency of the tasks are improved.

The invention relates to the technical field of communication, in particular to a method, device and terminal for searching information. The method includes the steps of collecting webpage information from a webpage collecting source, establishing an information database according to the collected webpage information, conducting search in the information database according to a search request sent by a client side, and pushing the search result to the client side for displaying. According to the method, device and terminal for searching information, the webpage information being paid attention by users is collected in advance and used for establishing the information database, and when the search request sent by the client side is received, search is conducted in the information database, so that search efficiency and search accuracy can be improved.

The invention belongs to the field of unmanned ship global path planning and particularly relates to an unmanned ship global path multi-target planning method based on the improved ant colony algorithm. The method comprises steps of establishing a marine environment map model by utilizing a Maklink graph theory; improving a path heuristic information strategy to obtain a path average value; designing an ant pheromone volatilization adaptive adjustment strategy; designing a strategy combining local pheromone updating and global pheromone updating; improving the state transition probability of anext node searched by an ant colony through a heading angle deviation factor of an unmanned ship; and designing an evaluation function by integrating the requirements of the shortest global path length, the least optimization iteration times of the improved ant colony algorithm, the lowest path smoothing coefficient and the like. The method is advantaged in that multiple targets such as a globalpath distance of marine navigation of an unmanned surface vehicle, the iteration frequency of optimizing the global path by improving the ant colony algorithm and the smoothness coefficient of the planned global path are comprehensively considered, the optimal global path of marine navigation of the unmanned surface vehicle is finally planned, and the method has relatively high safety.

The invention provides a method and system for recommending search terms. The method for recommending the search terms comprises the steps that header information of information issued by an issuer is extracted to obtain one or more subject terms and one or more descriptors used for limiting the subject terms; the combinations of the subject terms and one or more descriptors are used as the search terms; first evaluation values of the search terms are respectively obtained by multiplying the term frequency of the search terms by the inverse document frequency of the search terms; the inverse category frequency of the search terms in specific categories is respectively calculated and used as second evaluation values of the search terms; the evaluation values of the search terms are respectively added in a weighing mode to obtain issuer recommendation degree evaluation values of the search terms; recommendation terms are selected from the search terms according to the issuer recommendation degree evaluation values. The method and system can improve the information matching degree of the issuer and a finder.

Demographic information can be manually input into fields of a user interface. The demographic information can be for a patient of an eye care provider. One or more likely eye care insurers of the patient can be automatically determined from the demographic information. The eye care insurers that are accepted and that the patient is covered under can be unknown initially. A set of insurer maintained data stores corresponding to the set of one or more likely eye care insurers of the patient can be automatically determined. The set of insurer maintained data stores can be automatically searched for insurance information specific to individuals matching the demographic information. Results of the searching can be provided via the user interface, which identifies at least one eye care insurer of the patient and which provides an insurance number specific to an insurance policy of the patient.

The invention discloses a multi-feature fusion human face image searching method. The method comprises the following steps of: acquiring an image and text information surrounding the image from the Internet, storing the image and the text information respectively into an original image library and a text library; filtering non-human face images and irregular human face images from the original image library, saving regular human face images and storing the human face images into an human face image library; extracting the bottom layer features of the human face images in the human face image library; reducing the dimensions of the bottom layer features and storing the bottom layer features into a feature library; establishing feature indexes for the bottom layer features and storing the feature indexes into a feature index file; establishing text indexes for the text information in the text library and storing the text indexes into a text index file; receiving a text and an image which are submitted by a user; and indexing the text from the text index file, searching the image from the feature index file according to the searching result of the text, and feeding back the searching result of the image to the user. The method has the characteristics of high searching precision, high reliability and high efficiency.

The invention discloses an intermediate view synthesizing method based on block matching disparity estimation. In order to carry out disparity estimation on two images selected from the same scene, the same time and different angles, a virtual view at any position on a straight line (baseline) between original views is obtained through image interpolation according to the acquired disparity image. The method comprises the following steps: firstly carrying out disparity estimation by taking a left view and a right view as target images respectively, and computing two disparity views from the left view to the right view as well as from the right view to the left view by adopting a disparity estimation algorithm based on the block matching; then synthesizing a preliminary intermediate view according to the disparity views; and finally carrying out hole filling on the synthetic view to obtain a final result. Aiming at the characteristic that the left view and the right view only have parallax in the horizontal direction because shooting is carried out by a camera on a horizontal line, search is only carried out in the horizontal direction, the amount of calculation is reduced, and thealgorithm running time is shortened; and the experimental result indicates that a favorable synthetic result can be obtained under the condition of a farther distance of view foreground to the cameraor smooth image gradation changes.

The present invention discloses a cross-scene pedestrian searching method based on depth learning. The method comprises a step of carrying out preprocessing on each image in a sample library, a step of constructing and training a convolutional neural network, a step of extracting an upper half body local feature vector set and a lower half body local feature vector set from two groups of preprocessed image sets, and then the two local feature vector sets are fused to obtain a global feature vector, a step of carrying out preprocessing on an image to be searched, extracting an upper half body local feature vector and a lower half body local feature vector and fusing the two vectors to obtain a global feature vector, a step of orderly comparing the global feature vector corresponding to the image to be searched and the global feature vectors corresponding to the sample library images through a cosine similarity, outputting a group of similarity values, and sorting the similarity values according to a sorting algorithm. The method has the advantages that with the pedestrian images obtained in a monitoring video as the sample library, the design of features is not needed, the robustness is high, and the accuracy rate of actual searching is high.

The invention discloses a block-matching motion estimation method for the coding of H.264, which comprises the following steps: motion vectors of four macroblocks which are adjacent to the current macroblock and comprise a left macroblock, a right macroblock, an upper left macroblock and an upper right macroblock in a video image are obtained; according to the motion vectors obtained, a searching center is determined; a searching scope is adjusted according to the average value of the motion vectors; and motion searching is carried out in the searching scope, and the minimum absolute error SAD rule is utilized to carry out matching and stop judgment ahead of schedule. The motion estimation method utilizes the strong associativity between the current macroblock of the image and the motion vectors of the adjacent macroblocks and can greatly improve searching accuracy, lower calculation complexity and increase calculation speed on the premise of basically no signal to noise ratio loss.

Disclosed is an object search apparatus, including: an imaging unit to acquire image data of a point group; a radiation unit to scan the imaging area; a three-dimensional analysis unit to calculate three-dimensional data indicating position coordinates of the point group, from the image data; a master data storage unit to store a master data which is each of the three-dimensional data of a search object to be searched in the imaging area in a plurality of directions; and a judgment unit to position one point of the master data in each direction to each point of the three-dimensional data, to judge whether position coordinates of points of the positioned master data in any direction approach to position coordinates of corresponding points of the three-dimensional data of the imaging area, and to perform a search of the search object based on a result of the judgment.

The invention belongs to the technical field of three-dimensionalpath planning for underwater vehicles, particularly relates to animproved fireflyalgorithm based AUV (autonomous underwater vehicle) three-dimensional track planning method and provides an improved fireflyalgorithm based AUV three-dimensional path planning method. The method comprises the steps as follows: performing modeling and fireflypopulation initialization on the three-dimensionalpath planning for the underwater vehicles; calculating objective function values; calculating self-adaptive parameters; comparing thebrightness among fireflies, and updating positions of the fireflies; adding auxiliary planning operators; outputting optimal pathswhen meeting algorithmiteration stop conditions, ending the three-dimensionalpath planning for the underwater vehicles, and outputting an optimal path of the last iteration. Themethodis more flexiblethan a traditional path search algorithm, the auxiliary planning operators are added, and quick planning of AUV three-dimensional paths can be realized.

The invention provides a self-adapting parallax estimation-based virtual viewpoint drawing method, which is used for carrying out parallax estimation on two images simultaneously shot by a horizontal camera set and obtaining a virtual view at a random position among input viewpoints through interpolation and inverse mapping according to solved parallax views. The method comprises the steps of: firstly, respectively taking a left image and a right image as target images, carrying out self-adapting parallax estimation on a searching window so as to solve two parallax views which take the left image and the right image as the target images and adjusting a size of the window according to texture similarity of the two views; and then, carrying out virtual view drawing on appointed viewpoints according to solved parallax to obtain a final result. In the parallax estimation process of the block matching parallax estimation-based middle view synthesizing method, only horizontal searching is carried out, the amount of calculation is greatly reduced; and a method for self-adapting the searching window is provided, thus the accuracy for the parallax estimation is enhanced. The block matching parallax estimation-based middle view synthesizing method provided by the invention can obtain an excellent synthesizing result under the condition that the camera set is horizontally arranged and the distance from a foreground of a view to cameras is farther.

The invention discloses a method and an apparatus for obtaining augmented reality operation guidance and a mobile terminal. The method comprises the steps of identifying a target object in an image acquired by the mobile terminal; searching for related augmented reality contents according to the target object, wherein the augmented reality contents are an operation method related to the target object; and displaying the augmented reality contents. With the adoption of the scheme, an existing search process can be optimized, so that the search accuracy is improved.

The embodiment of the invention discloses a voice search method, a set top box, a storage medium, a server and a system. The method is applied to searching video resources, and comprises the steps of converting received voice information into text information; resolving a keyword from the text information according to a preset resolving rule; determining a search word corresponding to the keyword according to the keyword and a preset corresponding rule; and submitting the search word to a search server, thus allowing the search server to search a preset video resource library according to the search word and a preset search rule. In the technical scheme of the method provided by the embodiment of the invention, the keyword is resolved from the voice information has been converted into the text information according to the preset resolving rule, the search word corresponding to the keyword is determined according to the preset corresponding rule and submitted to the search server to search the video resources, so that improvement on a voice recognition rate and search accuracy can be achieved, operation is simple, practicability is strong, and user experience on voice search is enhanced.