838results about How to "Reduce memory consumption" patented technology

Some embodiments of the invention are directed to, among other things, systems, computer readable media, methods and any other means for using a handheld device to present information to a user. The handheld device and server(s) can be configured to differentiate among inexpensive and expensive information. For example, the handheld device can be configured to present media listings that represent media stored on the server(s). Each media listing can include text (inexpensive information) and thumbnails (expensive information). While the user scrolls through the media listings, the new listings can only include text. Once the scrolling stops, the missing thumbnails can be downloaded and integrated into their corresponding listings. In response to a listing being selected, the corresponding media file can be downloaded and / or played back to the user by the handheld device.

A light source estimation method of this invention estimates from the sensor response the color characteristics of an unknown light source of an image-pickup scene, in order to improve white balance adjustment and other aspects of the quality of color reproduction; a projection conversion portion 6 projects sensor response values 5 into an image distribution 9 in an evaluation space not dependent on the image-pickup light source 2 using parameters obtained by operations which can be calorimetrically approximated from spectral sensitivity characteristics of image-pickup unit 4, which are known, and from spectral characteristics of an assumed test light source 1; an evaluation portion 10 evaluates the correctness of a plurality of the test light sources 1 based on the distribution state of sample values of the projected scene; and accordingly, the correct image-pickup light source 2 is estimated.

A system extracts translations from translated texts, such as sentence translations from translated versions of documents. A first and a second text are accessed and divided into a plurality of textual elements. From these textual elements, a sequence of pairs of text portions is formed, and a pair score is calculated for each pair, using weighted features. Then, an alignment score of the sequence is calculated using the pair scores, and the sequence is systematically varied to identify a sequence that optimizes the alignment score. The invention allows for fast, reliable and robust alignment of sentences within large translated documents. Further, it allows to exploit a broad variety of existing knowledge sources in a flexible way, without performance penalty. Further, a general implementation of dynamic programming search with online memory allocation and garbage collection allows for treating very long documents with limited memory footprint.

The invention discloses a low-complexity quantum circuit simulation system, and belongs to the field of quantum computing. The low-complexity quantum circuit simulation system overcomes the technicalproblems in the prior art that the quantum circuit simulation storage space is too large and the calculation time is too long. The system comprises an input module, a storage module and an output module, the storage process of the storage module for data is as follows: (1) a mathematical model is established to represent the operation of a quantum state and quantum logic gate; (2) quantum bits aregrouped and the serial numbers of the quantum bits are rearranged; (3) degree reduction is conducted on a total operation matrix U0, and an output state is calculated. According to the system, a matrix element of a quantum state vector is directly operated by using an connotative inherent law of a quantum logic gate, at the same time, the method has expandability, in the future, a supercomputer can be used to simulate a quantum computer with more bits.

Method and apparatus for rapidly determining whether a particular subblock of data is present in a reduced-redundancy storage system. An aspect of the invention achieves this by hashing each subblock in the storage system into a bitfilter that contains a ‘1’ bit for each position to which at least one subblock hashes. This bitfilter provides an extremely fast way to determine whether a subblock is in the storage system. In a further aspect of the invention, index entries for new subblocks may be buffered in a subblock index write buffer so as to convert a large number of random access read and write operations into a single sequential read and a single sequential write operation. The combination of the bitfilter and the write buffer yields a reduced-redundancy storage system that uses significantly less high speed random access memory than is used by systems that store the entire subblock index in memory.

Disclosed are a partial heap garbage collector, and a partial heap garbage collection method, that during collection checks the time remaining (or equivalently the time taken so far, or the work done so far) to complete the partial heap collection. In a preferred embodiment, the partial heap collection is guaranteed to complete within a fixed time interval. For example, this guarantee may be obtained by applying a worst-case execution time (WCET) and sizing a nursery so that complete evacuation of the nursery can always be achieved on time. As an alternative, a technique, referred to as syncopation, may be used to allow generational collection to be used despite variance in program behavior over the short time scales in which a nursery can be collected. Syncopation may be accomplished via allocation control or via collection control.

The invention relates to a multi-scale seismic full-waveform inversion method based on a local adaptive convexification method. The method comprises steps: pre-processing is carried out, a zero value sequence serves as an initial value, and an initial speed model serves as a starting value; direct wave information is intercepted; forward modeling direct waves are obtained; an objective function inverted by a seismic source function is built; a direct wave residual and a direct wave residual back propagation wave field are obtained; an updating gradient and an updating direction of the seismic source function are calculated, and a step length is searched; the high-precision seismic source function obtained through inversion is outputted; attenuation time window processing is carried out; seismic data in the time window are simulated; local convexification processing and separation processing are carried out; a high-frequency component in an observation record is removed; a least square objective function is built, and a wave field residual is obtained; a residual back propagation wave field of the model space is obtained; the model updated gradient is obtained; the model updated direction is calculated, and the step length is searched; a multi-scale seismic full-waveform inversion result is outputted; and a final inversion result is outputted. The method is widely applied in the technical field of seismic exploration.

The invention discloses a device and a method for compression storing data and searching rapidly. At present, processing prefix search to dictionary uses Trie tree (character tree) generally, a character is stored in each node of the character tree which causes great waste of space. The character Trie tree is improved to character series Trie tree that reduces space waste efficiently. The node content in the character series Trie tree is compression processed in a system that has strict memory requirement, and a compression arithmetic is given. The invention can be used in a mobile store device with strict memory space requirement and can embodiment excellent of the method in condition of mass dictionary data. In addition, the device can be used for Suggest function of a search interface of the present internet, existing examination and (KEY, VALUE) type inquire function.

The invention provides a novel INS (inertial navigation system) / GPS (global position system) combined position and orientation method. The novel INS / GPS combined position and orientation method includes adopting a linear Kalman filter to perform filtering estimation to GPS original measurement data, and outputting optimal GPS navigation estimation value; according to the optimal position estimation value, providing initial position information to the INS, according to the optimal speed estimation value, providing initial speed information to the INS and solving INS initial measuring data to acquire INS navigation information; adopting a dynamic error model to establish a 9-order extended Kalman filter, integrating the INS navigation information with an optimal GPS navigation estimation value, performing feedback rectification to all INS navigation information at the same moment, and outputting optimal position data and orientation data after rectification and integration. The novel INS / GPS combined position and orientation method has such advantages as high precision, fast data processing speed and low hardware requirement and is applicable to low-cost INS / GPS combined position and orientation plan.

The invention provides a topic venation digging method and system based on massive searching logs. The method includes: counting the web searching logs to produce statistic data of a first searching word; counting the statistic data of the first searching word into statistic data of a second searching word; extracting key words of a first statistic record; counting total searching number of the keywords to obtain total searching statistic data of the keywords; calculating a searching heat value of the keywords in unit time; determining a comprehensive searching heat value of a topic in unit time; determining key time nodes of the topic. By the method and the system, topic shift caused by topic correlation can be avoided effectively, and a clear and complete topic venation can be dug out.

Embodiments of the invention relate to processing queries that utilize fact and / or dimension tables. In one aspect, a pre-join filtering phase precedes a star join. The necessary conditions for the pre-join filtering are considered for a given SQL query, including an estimated size of the hash table exceeding a threshold and presence of a local predicate either on the fact table or one or more dimension tables that is not a large dimension table. Once the necessary conditions are satisfied, the execution of the query exploits the pre-join filtering to build a pre-join output filter from columns of a reduced fact table that joins with each large dimension table. Thereafter, all the dimension tables and the fact table are joined in a star join while exploiting each pre-join filter. Accordingly, the order of when joins occur is changed in order to reduce the size of the fact table and to work from the fact table to reduce the size of large dimension tables.

A method for facilitating trick plays includes receiving a request to change a display mode of a certain media stream; fetching a media stream associated data structure associated with the certain media stream; and processing the media stream associated data structure and in response providing a media stream that complies with the requested display mode. The media stream associated data structure may include pointers to independently decodable frames of the certain media stream, and the provision of the changed mode media stream includes transmitting at least some of said independently decodable frames, optionally with duplicating frames, having their timing information adjusted.

The invention relates to a method and device of generating virtual surround sound. The method comprises environment sound and audio frequency object treatment flows. The environment sound treatment flow comprises first order B format signal input, spinning by a spin matrix, binaural transcoding based on a HRTF (Head-Related Transfer Function), and binaural output. The audio frequency object treatment flow comprises encoding one or more than one audio frequency object into a higher-order ambisonic signal, performing binaural transcoding based on BRIR (Binaural Room Impulse Response), performing binaural output, and respectively outputting environment sound and an audio frequency object after corresponding mixture according to left and right sound channels. The device comprises a head tracking means, an audio frequency object input module, an audio frequency object B format coding module, an audio frequency object virtual loudspeaker array signal generation module and an audio frequency object BRIR module. The method and device can effectively generate virtual surround sound with high quality, and are mainly used for cooperating with a virtual reality HMD (head-mounted display) to perform audio frequency stereo headphone playing.

The invention discloses a vehicle model identification method based on pooling multi-scale depth convolution characteristics, comprising the following steps: extracting the depth convolution characteristic of each vehicle model image in a vehicle model database according to different scales, wherein the first scale is not processed; carrying out PCA dimension reduction of the depth convolution characteristics of the remaining scales; carrying out coding using a local characteristic aggregation descriptor; carrying out PCA dimension reduction again to get the characteristic representation of the current scale; cascade-pooling the characteristics of all the scales to get the final characteristic representation of the current image; training a linear support vector machine using the characteristic representation of the vehicle model images to get a vehicle model identification system; and for a vehicle to be identified, acquiring the characteristic representation of the vehicle, and importing the vehicle into the identification system to identify the model of the vehicle. The traditional depth convolution characteristic lacks of geometrical invariability, which limits vehicle model image classification and identification in a variable scenario. The problem is well solved by using pooling multi-scale depth convolution characteristics of images. The vehicle model identification method of the invention is of high practicability and robustness.

Provided are a single target tracking method and an achievement device thereof. The single target tracking method includes that in a video V (V={F0, F1, FN}) formed by N frames of gray level images, a target O0 is selected in an F0 frame, and the image O0 is subjected to graying and width and height normalization processing to obtain an initialized parameter of the image; classifier initializing and updating which include training set constructing, feature extracting and model updating are carried out; and target tracking is carried out on the Ft+1 frame of image through an ft+1 model. The single target tracking method displays target surface appearances in a binary feature based compressed sensing and dimension descending method, can effectively display target deformation, improves shielding resistance and light resistance, and accordingly can achieve robust target tracking; and meanwhile has the advantages of small memory consumption and calculated amount, and achieves real-time tracking speed.

The invention relates to a CT image reconstruction method based on GPU hardware acceleration, which belongs to the field of X-ray CT technology. The software portion of the invention generally comprises a GPU-based CT data preprocessing module, a GPU-based CT data filter module, a GPU-based CT orthogonal protection module, and a GPU-based CT image reconstruction and back projection module. The method can achieve acceleration of CT image reconstruction algorithm by using GPU hardware, and the reconstructed portion and the data processing portion are achieved on GPU. The invention provides a segmentation processing method used for processing larger data, which aims to solve the prior problems of insufficient display memory of GPU and low data transmission speed from memory to the display memory. Different from prior methods, the segmentation processing method only needs a portion of projection data desired for each segment to be reconstructed, thus reducing data transmission and improving the entire reconstruction speed.

The invention discloses a data collection and control system for a target tracking segmented power supply street lamp. The data collection and control system comprises an LCD (liquid crystal display) interface, a serial printer interface, a keyboard interface, a microprocessor, an external storage module, an internal storage module, a clock synchronization module, a GPRS (general packet radio service) module, a wireless ratio frequency communication module or an electric carrier communication module, wherein the LCD interface and the serial printer interface are connected with the microprocessor through a serial port expansion module, the wireless radio communication module or the electric carrier communication module is connected with the target tracking segmented power supply street lamp system, the GPRS module is connected with a remote street lamp management center, and the system adopts the real-time management data structure and the roll polling algorithm for realizing the state roll polling. The system can automatically collect the real-time data of the street lamp system, and can realize the automatic statistics analysis, high-speed large-volume storage and street lamp control, the real-time work state can be mastered in real time, and the operation fault can be found. The street lamps are organized and managed in different segments, the data collection and control system adapts to the duplex requirements of traffic management, energy saving and environment protection, and the management efficiency of an illumination system is improved.

The invention relates to the technical field of network communication, and provides a method for acquiring addresses of network video programs. The method comprises the following steps that: a browser is used to open a plurality of webpages having audio and video programs and a player is ready to play the programs; the player requests the webpages from a remote video server, the player acquires the network address of the request video source according to the parameters transferred by the webpages, and the data source is remotely read and played; and the interactive information between the player and the remote video server are acquired and then analyzed so as to acquire the addresses of the network audio and video programs. Because the mode of combining the browser control and the network address monitoring is utilized to find the network audio and video addresses, and the browser is used for judging whether the webpages include the audio and video player and can control the playing of the player, the problems that the audio and video webpage scripts are complex and that the FLV is difficult to acquire the addresses from the webpages are solved.

The invention discloses a finite state machine actuating device and method, and a method for establishing and using a finite state machine. The finite state machine actuating method comprises the following steps of: receiving a trigger event; acquiring the current state of the finite state machine and acquiring a next state to which a current state is to be shifted according to an input event and a state shifting table; and determining a specific state type which corresponds to the current state from one or more specific state types belonging to a basic state type according to the basic state type, performing predetermined processing by using the specific state type and updating the current state by using the acquired next state. By adopting the method, actuating logic of the finite state machine is improved in combination with the principle of an object-oriented state mode and a Flyweight mode, the specific state type of a single example mode is established, and repeated creation and deletion of an object are avoided, so that resource consumption of a system can be lowered on various application developing and modeling occasions and the like, and the actuating speed and efficiency of the application are improved.

The present invention provides a method for converting source code written in a dynamically typed language into target code written in a statically typed language. The method includes the step of generating in a storage device the target code from the source code, the step of generating the target code including storing in a buffer a signature of a function / method used by a caller in the source code. In this step, generation of an interface class is delayed to a time point immediately before execution of a part of the target code corresponding to the function / method call by the caller, the interface class being needed to generate the caller and serving as an interface in the statically typed language. The method also includes the step of generating in the storage device a consolidated interface class for all signatures stored in the buffer, and loading the generated consolidated interface class from the storage device. In this step, all the signatures in the buffer are deleted upon generation of the consolidated interface class.

The invention provides a column-based storage and research method and equipment based on a hard disk and the internal storage. A data sheet structure on data source is created in the internal storage through creating metainformation of the data sheet corresponding to the data source, and a column-based data block is generated on the current data according to the metainformation and is stored into the hard disk; the internal storage can be used more effectively, the following query data performance in the hard disk reaching the similar performance with the query data in the internal storage is realized, and the strong data analysis ability based on high-speed query efficiency can be further supported in following; moreover, when the column is an index column, the following data query efficiency can be improved by establishing a reverse index for each index column and storing the index column to the document at corresponding position in a solid state disk with the adoption of RadixTree structure.

Provided is an image quick defogging optimized method based on a black channel. Edges of an image and non-edge areas of the image are processed through different templates to obtain a transmission picture, and a precise atmosphere light value is obtained by the fact that a sky area or an area with thickest fog is divided. The method greatly reduces consumption, improves the image recovery speed, and improves image recovery effects.

A method, system, program product and signal bearing medium embodiments of the present invention provide for deploying software modules for software application use in a computer system thereby reducing load time as well as memory requirements. Deployment of a plurality of software modules and associated deployment descriptors into a software module depository and creation of a deployment information repository from the associated deployment descriptors occurs. A name service is initialized with information from the deployment information repository and a requested software module identifier is then mapped to a respective enabler. Having mapped the requested software module to an enabler, the respective software module is enabled for the software application use. On-demand deployment in this manner saves start-up time as well as initial and ongoing memory allocation.

The invention discloses a cache device and a cache method for an embedded browser, which can reasonably buffer network data, reduce storage consumption, fully utilize cached resource and accelerate page view speed. A technical scheme of the invention comprises that: the device comprises an expired data prior eliminating module for acquiring corresponding expired time from a server when the browser acquires the network data, and deleting the expired data when the cache space is insufficient, and a low-frequency utilization prior eliminating module for adding statistics of operating frequency to data in a cache of the embedded browser and deleting data with the lowest operating frequency.

The invention discloses a network intrusion detection method. The network intrusion detection method includes: searching network data to construct a test network data set; performing feature extraction on the test network data set by utilizing a kernel principal component analysis method; constructing a training data set, putting the training data set into a support vector machine classifier for training; obtaining feature datasets, obtaining an optimal feature subset from the feature data set by using a genetic algorithm; utilizing a firefly swarm optimization algorithm to obtain the overalllocal optimal feature subset and the optimal support vector machine parameters from the optimal feature subset, processing the training data set according to the overall local optimal feature subset,and inputting the training data set into a support vector machine classifier for classification modeling to obtain a network intrusion detection model. According to the method, the simplicity and convenience of the algorithm are improved, abnormal data can be more effectively found from samples, the detection accuracy of network intrusion is effectively improved, the missing report rate and the false report rate are reduced, and the overall performance of network intrusion detection is improved.

The present invention relates to a method of compensating the pressure drop in an ordinary ventilator tube. The method comprises the measurement of a first air flow at a first pressure on the apparatus' side of the ventilator tube and with the respiratory mask taken off. Moreover, a parameter is calculated in a function which provides the pressure drop in the ventilator tube in dependence on the air flow. The parameter for the air flow and pressure is calculated. Later, the air flow is calculated with the respiratory mask being put on. By means of the function a correction pressure is calculated from the air flow. Finally, a target pressure is set which results as a sum of the selected mask pressure and the correction pressure. The invention moreover relates to ventilator for performing the aforementioned method and to a memory medium for storing a corresponding program.

A distributed data storage method comprises the following steps: acquiring data resources and data identification corresponding to the data resources; calculating the hash value of the data identification; calculating the page numbers of disk pages, corresponding to the data resources, in a memory unit according to the hash value and the block numbers of disk blocks in the disk pages; saving the data resources in the disk blocks of the corresponding disk pages according to the page numbers of the disk pages and the block numbers of the disk blocks. Besides, the invention provides a distributed data storage system. The distributed data storage method and system can reduce depletion of the internal memory and improve the storage efficiency.

The invention discloses a lung segmenting device of a 3D (three-dimensional) U-Net network based on mixed coarse segmentation features. The lung segmenting device comprises a lung 3D binary image generating module, a lung segmenting module and an optimizing module, wherein the lung 3D binary image generating module is used for sequentially performing Gaussian filtering, binaryzation, marking of communication areas and screening, size matching and trimming on an original lung CT (computed tomography) slice so as to generate a lung 3D binary image; the lung segmenting module is used for calculating an original lung CT image and the lung 3D binary image by the trained 3D U-Net network, and outputting a segmenting probability map; the optimizing module is used for calculating the segmenting probability map by a conditional random field, and outputting the final lung segmenting result. The lung segmenting device has the advantage that while the higher segmenting accuracy is produced, the lower calculation consumption and memory consumption can be guaranteed.

The invention discloses a method for classifying images by performing pairwise-constraint-based online dictionary reweighting. The method comprises the following steps of: performing bottom-layer feature extraction on all images of a training set to construct an initial visual dictionary; performing feature transform on extracted bottom-layer features by adopting sparse coding to obtain coded features; performing maximum aggregation on the coded features to obtain a feature expressed by a vector for the classification of a classifier; and performing online dictionary reweighting on the feature expressed by the vector by utilizing pairwise constraints, and transmitting the feature to the classifier for training and classification. By the pairwise constraints, a relationship between pairwise images can be effectively coded; by an online learning algorithm provided on the basis of a conservative-radical training strategy, training time is greatly shortened, and incremental updating can be realized; and the method is particularly applied to massive datasets.