221results about How to "Save memory space" patented technology

Systems, devices and methods are provided to compress data, and in particular to code and decode data. One aspect of the present subject matter is a data structure. The data structure includes a field representing a decoding structure to decode canonical Huffman encoded data, and a field representing a symbol table. The decoding structure includes a field representing an accelerator table to provide a 2^N-deep direct-index lookup to provide high-frequency symbols for high-frequency data and to provide bracketing indices for low-frequency data. The decoding structure also includes a field for a binary search table to provide a low-frequency symbol index using a binary search bounded by the bracketing indices provided by the accelerator table. The symbol table is adapted to provide a symbol associated with the low-frequency index.

A system for server-side communication support in a distributed computing environment is provided that includes a local reference (202) to a server object (110) on a server system (104). The local reference (202) provides a location of the server object (110) and a type of the server object (110). The local reference (202) resides within a server-side object request broker (114). A type object (204) is accessible by the local reference (202) using the type of the server object (110) from the local reference (202). The type object (204) represents a class of the server object (110). One or more specialized function objects (206) are linked to the type object (204) and correspond in number to one or more methods (190) of the server object (110). The one or more specialized function objects (206) provide wrapper functionality for each method (190) of the server object (110). Each specialized function object (206) provides access to a corresponding one of the one or more methods (190) of the server object (110).

The invention discloses a three-dimensional vector real-time dynamic stacking technique based on multiple LOD (Level of Detail) transparent textures. The three-dimensional vector real-time dynamic stacking technique comprises the following steps of: firstly, drawing various vector layers on the transparent textures of corresponding plotting scales respectively in an internal memory through data pre-processing, automatically carrying out block cutting on the vector layers, then saving the vector layers in a disk as an LOD pyramid data storage structure, and establishing a mapping relation with a same-level terrain and an image data block; and secondly, in a real-time terrain rendering and roaming stage, loading the transparent textures of vectors of the corresponding layers through dynamic selection, stacking the vector layers into the same level of image data block according to an established Alpha passage, and realizing the real-time stacking of the vector layers in the three-dimensional terrain based on an internal memory multi-passage texture blending technology. According to seamless visualization of high-resolution vector map data and a digital elevation model, the effect of the high-resolution vector map data and the digital elevation model in spatial information expression and analysis can be enhanced. According to the technique, the real time stacking of multilayer vector data in the rendering and roaming stage of a large-scale three-dimensional terrain scene is realized, and the dynamic interactive screening of the vector layers is supported.

In a decoder having an improved LLR (log-likelihood-ratio) update method is provided. The method comprising the steps of: providing a parity check matrix; and using merely a set of parameters on a row of the parity check matrix instead of data of the whole non-zero elements of the parity check matrix free from at least one shifting action after each row updating; thereby saving memory space and process time.

A method of updating firmware using the object push profile in the bluetooth object exchange protocol is proposed. This method first makes use of a bluetooth transmitter to connect a bluetooth electronic device whose firmware is to be updated and a host storing a new program code. Next, the bluetooth transmitter checks parameters in an EEPROM of the bluetooth electronic device to determine whether it is necessary to update the firmware. Through the object push profile (OPP) in the object exchange (OBEX) protocol of the bluetooth technology, this new program code is reprogrammed to a flash memory in the bluetooth electronic device. After the end of the reprogramming action, the bluetooth connection between the bluetooth electronic device and the host is cut off to finish firmware update of the bluetooth electronic device.

The invention provides a computing unloading scheduling method based on depth reinforcement learning, which provides a method for computing unloading to make unloading decision for Internet of Thingsequipment, including making decisions on various aspects needing to be unloaded according to the basic model of computing unloading. Based on different optimization objectives, different optimizationobjectives can be achieved by changing the value function. TheDeep-SARSA algorithm is similar to DQN algorithm, which combines reinforcement learning and depth learning. It can effectively change theunloading state and unloading action into training samples of depth learning when cooperating with experience pool. The invention can effectively carry out machine learning on an unloading state modelof an unlimited dimension, reducing the complexity of learning, this method uses neural network as the linear approximator of Q value, which can effectively improve the training speed and reduce thesample required for training. This method can effectively make the best decision through deep reinforcement learning under the given model and optimization objective.

The invention provides a stream application function interface distributing method realized on the basis of HTML5. The method comprises the specific steps that multiple function interfaces of stream application realized on the basis of HTML5 are provided; requests for calling any function interface are responded, and a function module corresponding to the function interface operates; a corresponding response result is returned through a callback function or a returned value corresponding to the function module. The invention further provides a stream application function interface distributing method realized on the basis of HTML5 through the method. The method not only can provide a simpler application development mode for developers, but also can provide new application experience for users, the application can be used only by clicking without being installed, and interaction and experience are close to those of local application.

When a sub-section of a predefined path temporarily cannot be traversed by an autonomous mobile unit, an extended path is found for subsequently traversing the temporarily blocked sub-section. A first path point of the predefined path is determined using at least one predefinable distance criterion, which takes into consideration the distance of the temporarily blocked sub-section from the first path point. An extension sub-section is determined which begins at the first path point, terminates at a second path point of the predefined path and encompasses at least the temporarily blocked sub-section. The extended path is programmed with the extension sub-section being integrated into the predefined path at the first path point.

The invention discloses an electrocardiogram R wave detection method. The method comprises the following steps: filtering an input electrocardiogram signal by a band pass filter, and performing phase compensation; performing differential processing; performing data reduction on a differential signal by adopting linear variation; performing Shannon energy conversion by using a formula d(n)*d(n)*log(d(n)*d(n)); filtering by using an average filter, and performing phase compensation; detecting a maximum point and a minimum point, eliminating a fault R point, and correcting a falsely eliminated point to obtain a position of approximate R wave; and searching the position of true R in a range of periphery of the position of the approximate R wave +/- 25 points. By using the method, the problems of time delay and asystole detection unavailability in the conventional R wave detection method can be solved, and the method can be used for analyzing real-time electrocardiogram to reduce time delay and reduce required memory space, so that the accuracy rate is improved.

Disclosed is a speech recognition method which is based on a dynamic extension of the word models in combination with an evaluation of the pronunciation variations.

The present invention provides a rapid distortion correction method based on a Bayer image. The method comprises the steps of firstly, inputting the Bayer image acquired by a fish-eye lens and a CMOS sensor in an FPGA module, adopting an equidistant internal reference model to map a fish-eye image to a three dimensional spherical surface, utilizing a longitude and latitude mode to describe the three dimensional spherical surface, and obtaining a longitude and latitude coordinate transformation relation of the fish-eye image plane coordinate points and the corresponding mapping points of the three dimensional spherical surface; secondly, adopting a vertical fish-eye projection method to obtain a calculation formula of projecting the mapping points on the three dimensional spherical surface to the corresponding projection points in a two dimensional plane, and projecting the three dimensional spherical surface to the two dimensional plane again; and finally, adopting a transverse scanning interpolation method to transform the two dimensional Bayer image interpolation into a one-dimensional image interpolation, utilizing the multi-channel correlation of the Bayer image to divide an interpolation section into a plurality of segment sections, solving a cubic polynomial curve of the to-be-fitted variables in each section separately to obtain the pixel values of the to-be-interpolated points, and carrying out the interpolation on a corrected two dimensional plane image, thereby improving the Bayer image distortion correction precision.

The invention relates to a method for dynamic data synchronization. When the dynamic data of a host computer is changed, the method includes the steps as follows: the host computer builds the synchronous request data of the record according to the record corresponding to the changed dynamic data, and transmits the synchronous request data to the record level synchronous process; the operation information is recorded according to the synchronous request data or the synchronous request data and the buffer zone of the global data, and the synchronous information is built through the synchronous rule and transmitted to a standby unit; the standby unit performs the synchronous operation for the required synchronous record in a synchronous manner according to the synchronous information transmitted by the host computer; after the standby unit achieves the synchronous operation of the record, the synchronized record is deposited and the response information which is synchronously achieved istransmitted to the host computer; after the host computer receives the response information, the changed dynamic data is synchronously achieved, and the synchronous operation is finished; the invention also discloses a system for dynamic data synchronization, and the method and the system of the invention can synchronize the dynamic data between the memory databases of the host computer and the standby unit in real time.

The invention provides a parallel chain main chain transaction state synchronization method and device and a storage medium, and the method comprises the steps: synchronizing a plurality of parallel chain transactions from corresponding main chain nodes, and sequentially storing transaction Hash of each cross-chain transaction in each synchronized parallel chain transaction to generate a cross-chain transaction Hash sequence; packaging and executing each parallel chain transaction and storing an execution state; generating an execution state sequence of each cross-chain transaction according to the cross-chain transaction hash sequence and the execution state of each cross-chain transaction, and generating bitmap information; generating a verification hash according to each cross-chain transaction; generating a consensus transaction comprising the bitmap information and the check hash, and sending the consensus transaction to the corresponding main chain node so as to verify the checkhash according to each cross-chain transaction; and if the verification is successful, performing parallel chain consensus, and obtaining an execution state of each cross-chain transaction according to the bitmap information after the parallel chain consensus is successful so as to process the main chain service logic corresponding to each cross-chain transaction. According to the application, thememory space occupied by the cross-chain synchronous transaction state is greatly reduced.

In an image processing apparatus of the present invention, interpolation operations are performed with a LUT to obtain CMY image data corresponding to input RGB image data. The image processing apparatus includes a gamut selecting section for selecting a gamut corresponding to an inputted image, from an RGB color space, based on input RGB image data; a LUT writing section for changing CMY image data in the LUT to CMY image data corresponding to grid points contained in the selected gamut; and a position information changing section for writing position information in a memory, wherein the position information indicates RGB color space positions of the grid points contained in the selected gamut. Thus, the image processing apparatus of the present invention can save memory space.

The invention discloses an automatic on-line vision detecting and sorting system and a sorting method. The sorting method comprises the steps of: arranging a sensing module for sensing a product on a production line, shooting a picture only when the product is sensed, and finally removing defective products by using a control module through the follow-up processes of picture processing, defective product detection and the like, thereby automatically detecting and automatically sorting the defective products. The continuous picture shooing does not need, so that a large quantity of memory spaces are saved; and by using the automatic on-line vision detecting and sorting system and the sorting method, the defective products can be removed in real time, so that the real-time performance and the stability of sorting the defective products are improved, the production efficiency is greatly improved and the probability that the defective products enter markets is reduced.

The invention discloses a method for label distribution, comprising the following steps: a label is allocated for each operator edge router in a network, wherein labels allocated for any two operator edge routers are different; and through an inner gateway protocol, the corresponding relation of the label of each operator edge router and a loopback address is notified to each operator edge router and an operator core router. Meanwhile, the invention discloses a label processing method, a label distribution device and a label processing device. The method can realize label distribution without adopting dynamic label distribution protocols such as LDP.

The invention relates to a method for analyzing an XML file by indexing. The technical characteristics are as follows: the method comprises the steps: traversing to illustrate a DTD file with an XML file structure and extracting a subtree tag name under a root node in the DTD file; then creating a Hash table, traversing the XML file needing to be analyzed according to the extracted subtree tag name, inquiring and recording relative positions of starting of all subtree tag names in the XML file, constructing a new item according to data items, and adding the new item into the Hash table to form a subtree index table; and creating a key element index table, and then utilizing an unvalidated IXP analytic model or validated IXP analytic model to carry out analysis. The method has the benefits that: for the large XML file, the analytic speed of an IXP method is far faster than that of a DOM method and an SAX method. By providing a general interface, the mode can be widely applied in analysis of various XML files, and provides a new method for analyzing an XML text.

The present invention provides an application loading method and loading apparatus. The application loading method comprises the steps of loading a basic function module for maintaining operation of an application when the application is detected to start; and loading an additional function module corresponding to any one configuration function when the any one configuration function of the application is detected to be triggered during an operation process of the application so as to execute the any one configuration function. According to the technical scheme, the application can be loaded as required, memory space is saved, the problem of stuck caused by that all functions of the application are loaded for one time in the related technology is solved, and usage experience of users is improved.

The invention discloses a method and a device for realizing on-screen display (OSD) of characters. The method comprises the following steps of: downloading dot matrix information corresponding to the common tag characters and the characters in an American standard code for information interchange (ASCII) character bank to a digital signal processor (DSP) memory; storing the dot matrix information corresponding to characters in a wide-byte character bank in a memory of a host; receiving an OSD display command and judging the types of to-be-displayed OSD characters; if the to-be-displayed OSD characters are the common tag characters and/or belong to the ASCII characters in the ASCII character bank, acquiring the dot matrix information corresponding to the to-be-displayed OSD characters from the memory of the DSP; if the to-be-displayed OSD characters are the characters in the wide-byte character bank, downloading the dot matrix information corresponding to the character from the memory of the host to the memory of the DSP; and uploading the dot matrix information corresponding to the to-be-displayed OSD characters to a video image to display. By using the method and the device, the required dot matrix character bank does not need to be downloaded to the memory of the DSP in advance.

The invention relates to the technical field of computer graphics, in particular to a skeletal animation implementation method. The skeletal animation implementation method is based on the secondary extraction of skeletal animation of virtual skeletons, is used for unifying the skeletal animation of different skeletons to realize the purpose of reuse, and reuses the skeletal animation to the most extent by partitioning virtual points according to kinematics logic. Most of the reuse adopts the storage mode of virtual skeletons; and by constructing a set of standard virtual skeletons, the skeletal animation data link and reuse of different skeletons are realized. The reuse ratio of skeletal animation is increased, and the memory space required by the skeletal animation is reduced.

The invention discloses a method and a device for calculating visit hotness. The method includes: acquiring URL (uniform resource locator) of a current statistical object from a received visit hotness statistical request; judging queue type of the current statistical object according to the URL; and storing the current statistical object to a corresponding queue according to the queue type; wherein the queue type comprises a filter queue and a LFU (least frequently used) queue, the filter queue is sorted according to an LRU (least recently used) algorithm and used for storing statistical objects with visited times lower than a preset threshold, the LFU queue is used for storing statistical objects with visited times higher than the preset threshold, and the filter queue and the LFU queue are fixed in length. The fixed-length filter queue is combined with the fixed-length LFU queue, the filter queue is sorted according to the LRU algorithm and mainly used for filtering Internet objects which are not visited frequently, compared with traditional LFU algorithms, internal memory space is saved, and the LFU algorithms are optimized.

The invention provides a pattern matching method based on decision tree pruning. The simplified AC algorithm of the decision tree pruning method is combined in the pattern matching method, and the pattern matching method comprises the steps of generation of an automatic machine, simplification of the automatic machine, failure pointer calculation and suffix table storage and matching. The conventional automatic machine type pattern matching algorithm is divided into two steps: matching possibility judgment and matching confirmation. The possibility of matching of text strings and pattern strings in a pattern set is judged by the simplified automatic machine and then matching confirmation is performed. Compared with the conventional automatic machine type pattern matching algorithm, memory consumption of the simplified method is reduced for 35%-40% under the premise of guaranteeing speed. Besides, the scale of the automatic machine can be effectively reduced by the decision tree pruning method, and nodes useless for classification judgment are deleted so that memory consumption of the conventional automatic machine type pattern matching can be effectively reduced.

A download server that transmits, when it receives a request of one or more application programs from a client device, to the client device connected to the download server via a wired or wireless network the application programs includes a dependency analyzer, a partitioned shared library generator and a controller. The dependency analyzer that analyzes a dependency of each of the application programs on existing shared libraries to detect object codes, which are actually used by corresponding application program, in the existing shared libraries; the partitioned shared library generator that generates, based on the analyzing result of the dependency analyzer, partitioned shared libraries having only the actually-used object codes; and the controller that transmits, by using a connection module, partitioned shared libraries selected from the generated partitioned shared libraries, and the application programs linked thereto to the client device via the network.

A method for executing SQL script file (SF) in distributed system includes using the first linear program to divide SQL SF to be multiple script block and placing each script block in sequence into preset buffer region (PBR), pausing the first linear program if PBR is full and executing the first linear program continuously till PBR is again to have space, using the second linear program to fetch one script block in buffer region and to release relevant storage space as well as to transmit content of said script block to server end, fetching next script block in buffer region after execution feedback information of server end is received repeating above steps till SQL SF is all executed.

The embodiment of the invention provides a method, a device, and an apparatus for message forwarding of firewall equipment, and a storage medium. The method comprises the following steps: obtaining ato-be-forwarded message which comprises quintuple data; calculating a hash value of the quintuple data; querying whether a simplified session table item matched with the to-be-forwarded message existsin a simplified session array hash table previously built in the current CPU or not according to the hash value; obtaining a corresponding session table item template according to a template sessionaddress in the matched simplified session table item if the simplified session table item matched with the to-be-forwarded message exists in the simplified session array hash table; and forwarding theto-be-forwarded message according to the session table item template. According to the method, the device, and the apparatus for message forwarding of firewall equipment, and the storage medium, a large amount of memory space can be saved during the whole forwarding process; the problem that cache access failure probability is increased when a session table item is large due to large memory spaceoccupation can be avoided; and the stability of the overall forwarding performance of the firewall device can be improved.

The invention discloses a magnetic suspension bearing control method and a magnetic suspension bearing control device. The method comprises the following steps: S10, acquiring a feedback value Xfdb of a bearing rotor deviation reference position and performing position adjustment in DSP to acquire control current Iref; S11, receiving the input control current Iref input by an FPGA chip, performing current adjustment in FPGA, and outputting a control signal; S12, receiving the control signal output by the FPGA by a power amplifier and processing the control signal to control the current value of a bearing coil, so that the position of the bearing rotor is controlled and the rotor can stably suspend in the reference position. The current adjustment is arranged in the FPGA and the control efficiency of the bearing system can be greatly improved; in addition, a current adjustor is conducted in the FPGA and the internal memory space of the DSP is saved, so that the utilization rate of the whole control system is greatly increased.

The invention relates to a moving object detection method resisting illumination variation. An existing moving object detection method is sensitive to the illumination variation frequently, and therefore when illumination variation occurs in a detection environment, accurate detection for a moving object is difficult. The moving object detection method resisting the illumination variation integrates an asymmetrical inter-frame difference method and a self-adaptation mixed gauss background difference method in inter-frame difference. The moving object detection method resisting the illumination variation includes a first step of preprocessing an input moving image, a second step of respectively conducting asymmetrical inter-frame difference moving object detection and self-adaptation mixed gauss background difference moving object detection, and a third step of comparing moving object results independently obtained through the two detection methods, calculating the moving object results, extracting a comprehensive moving object, filling cavities inside the moving object and filtering out some small interference through mathematics morphological processing. Through the moving object detection method resisting the illumination variation, the moving object can be accurately detected under the normal condition, and meanwhile the accuracy rate can reach over 90% under the condition that the illumination variation of the detecting environment occurs. Therefore, the moving object detection method resisting the illumination variation can meet needs of a monitoring system.

The invention discloses a method for preventing missing test or repeating test in mobile phone production, which is characterized by comprising the following steps of: firstly, production serial number writing process: writing a production serial number including a test mark position on a mobile phone when the mobile phone is flowed to a production serial number writing position after a soft is downloaded and installed in a mobile phone board; and secondly, test process: reading the value of the test mark position in the production serial number of the mobile phone via a test tool, judging whether the mobile phone enters into the calibration process or the final test process or the IMEI (International Mobile Equipment Identity) writing process or returns to the production serial number writing process according to the value. The method can be used for preventing the unqualified mobile phone from being flowed out of factories, and also avoiding repeated work and time waste, thereby improving the production efficiency.

The invention discloses a remote terminal controlling method, apparatus and system. The method comprises the following steps: acquiring a control instruction used for realizing remote control over a remotely controlled terminal; obtaining a preset webpage-and-multimedia interaction modes according to the control instructions; building remote connection to the remotely controlled terminal through a preset browser and the webpage-and-multimedia interaction mode; and conducting remote control over the remotely controlled terminal based on the remote connection. According to the invention, both of the controlled terminal and a master control terminal do not need to be equipped with related software and remote display and remote assistance can be performed for the remotely controlled terminal, so that memory space of the controlled terminal and the master control terminal can be saved. Further, as the remote connection is built based on the browser and webpage-and-multimedia interaction mode, the remote control is not subject to the influence of environments of both the remotely controlled terminal and the master control terminal. Therefore, terminal interaction effect is improved.

A method and apparatus for programming a cellular telephone with activation parameters. A cellular telephone to be programmed is connected to a computer via a communications interface. A user input device is provided to allow a user to provide programming information, including a cellular telephone type, to the computer. A visual representation of the cellular telephone type is displayed on a display monitor to allow the user to interactively verity the inputted programming information. Upon such user verification, the computer determines activation parameters and transmits the activation parameters to the cellular telephone. In accordance with another aspect of the invention, certain activation parameters are dynamically computed based on the user provided programming information In one embodiment, the user enters a cellular telephone number to be assigned to the cellular telephone. The computer looks up a system identification associated with the cellular telephone number in a memory unit. Certain activation parameters are dynamically computed based on the cellular telephone number and system identification.