CN102393815A - Parallel processing method of remote sensing operation - Google Patents
Parallel processing method of remote sensing operation Download PDFInfo
- Publication number
- CN102393815A CN102393815A CN2011101955942A CN201110195594A CN102393815A CN 102393815 A CN102393815 A CN 102393815A CN 2011101955942 A CN2011101955942 A CN 2011101955942A CN 201110195594 A CN201110195594 A CN 201110195594A CN 102393815 A CN102393815 A CN 102393815A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- module
- sensing computing
- parallel processing
- sensing operation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Multi Processors (AREA)
Abstract
The invention discloses a parallel processing method of a remote sensing operation, which comprises the following steps of: 1) establishing remote sensing operation base modules; 2) according to remote sensing computing, selecting required remote sensing operation base modules to form a remote sensing operation flow module; and 3) converting the remote sensing operation base modules and the remote sensing operation flow module into an executable file, and realizing parallel processing. According to the invention, the remote sensing operation base modules are firstly established; when carrying out the remote sensing operation, the corresponding remote sensing operation base modules are directly called, and no re-compilation is needed, so that the reusable property of the operation is greatly increased, and the efficiency of scientific research work is improved; and the flow of the operation brings the parallel processing model, through experiments, the operation time can be greatly shortened, and the resources of a computer can be reasonably utilized.
Description
Technical field
The present invention relates to remote sensing algorithm process technical field, refer to a kind of method for parallel processing of remote sensing computing especially.
Background technology
Along with the application in treatment of remote field more and more widely, the remote sensing Processing Algorithm also is more and more.Many treatment of remote algorithms all can comprise public module, nearly all can write these public modules again when still writing algorithm at every turn, can reduce research work efficient greatly like this.
In addition; Remote sensing field language such as IDL and the MatLab etc. that arrive commonly used it is advantageous that the computing of matrix, and promptly there is incomparable advantage the Flame Image Process aspect; But multiple programming is not its speciality; Parallel processing modes such as multithreading, multinuclear computing seldom relate in these language of MatLab, even not support multi-thread mechanism in the version in early days, even if support multithreading; The just inner multithreading support that realizes of interior some built-in function of language does not have the interface of not writing multithread programs to the user yet.Therefore, IDL and MatLab are not suitable for carrying out multiple programming, can only carry out the serial processing of remote sensing computing, and operation time is longer.
Summary of the invention
To the problem that prior art exists, the present invention provides a kind of and can improve research work efficient and shorten the remote sensing remote sensing computing method for parallel processing of operation time.
For realizing above-mentioned purpose, remote sensing computing method for parallel processing of the present invention is specially: 1) set up remote sensing computing basic module; 2) calculate according to remote sensing, choose required remote sensing computing basic module and form remote sensing computing process module; 3) convert remote sensing computing basic module and remote sensing computing process module into executable file, realize parallel processing.
Further, said remote sensing computing basic module comprises load module, output module and processing module, wherein,
Load module is used in reference to needed each original input data of remote sensing computing;
Processing module is used for carrying out the remote sensing computing according to original input data;
Output module is used in reference to the store path to the remote sensing operation result.
Further, said load module, output module and processing module all have the Xml descriptor of standard.
Further, said Xml descriptor comprises attribute information, algorithm information, model information and input information.The general property of attribute information describing module is like information such as module title, function, notes; The details of algorithm in the algorithm information describing module comprise the algorithm invokes mode, input number and to the description of each parameter, like information such as parameter title, parameter type, parameter values; Model information is to the professional parameter that relates in the affiliated algorithm model, and this type parameter does not need when driven algorithm, to be selected to fill in by the user, but set in advance; Input information is that the flow model is whole to outer field input and output parameter, and the relation between the input and output of flow process internal module does not show in this type of.These Xml descriptors are the explanation document to algorithm and flow process, also are simultaneously the parallel necessary data supports of producing of driven algorithm.
Further; Said remote sensing computing process module also comprises the Xml descriptor; This Xml descriptor comprises all load modules, output module and the processing module of forming remote sensing computing process module, and the mutual relationship between all load modules, output module and the processing module.
Further, the interface and the C# that carry through IDL or Matlab carry out shuffling, convert remote sensing computing basic module and remote sensing computing process module into executable file.
The present invention at first sets up the basic module of remote sensing computing, directly calls corresponding remote sensing computing basic module when carrying out the remote sensing computing and gets final product, and need not write again, makes the reusability of algorithm increase greatly, for research work has improved efficient; Secondly, the dirigibility of algorithm production procedure also significantly improves, and can carry out rational algorithm flow customization according to existing data; Once more, the procedure of algorithm has also brought the processing mode of parallelization, through overtesting, can effectively shorten operation time, rationally utilizes the resource of computing machine.
Description of drawings
Fig. 1 is the process flow diagram of method for parallel processing of the present invention.
Embodiment
As shown in Figure 1, the method for parallel processing of remote sensing computing of the present invention may further comprise the steps: set up remote sensing computing basic module; The encapsulation of Flame Image Process each time is called a remote sensing computing basic module.The remote sensing computing basic module of setting up should carry out detailed description and be enough to drive the operation of this time calculating the information of Flame Image Process.
According to the general treatment scheme of remote sensing algorithm, remote sensing computing basic module is divided three classes, and is respectively load module, output module and processing module.Load module only is responsible for referring to needed each original input data of remote sensing computing; Output module is the store path of only responsible remote sensing computing then; Processing module is the real node that calculates of participating in handling, and it is through the algorithm of encapsulation, with obtaining output data after its input data processing.
After setting up remote sensing computing basic module, choose required remote sensing computing basic module and form remote sensing computing process module; For a remote sensing computing process module that constitutes by many remote sensing computing basic modules, need a kind of description standard equally.It needs to describe out which load module is arranged, output module, processing module and the priority operational relation between them in this workflow.
Convert remote sensing computing basic module and remote sensing computing process module into executable file, realize parallel processing.The multi-core parallel concurrent of flow process handle can accelerated process progress, effectively utilize the computational resource of computing machine.Multi-core parallel concurrent for algorithm is handled, and at first need carry out the encapsulation of process level to algorithm.This is because remote sensing algorithm majority is not an executable program, and multinuclear calculating is comparatively convenient to the setting of executable program.The remote sensing popular software is IDL and MatLab, and we utilize translation interface and C# to carry out shuffling these language, and IDL or MatLab are integrated into executable file.After encapsulation is accomplished, use the executable file of the Process class starting algorithm among the C#, be beneficial to carry out parallel processing.
The shadow product of moving on to production remote sensing MODIS normalizing snow melting lid index product (NDSI) is an example; Relate to the output module of the processing module of the load module of the output module of the processing module of the load module of a TOAL, a TOAL, a TOAL, a NDSI, a NDSI, a NDSI, two projection conversion load modules, a projection modular converter and a projection conversion output module altogether; The path, processing module raw data place of the load module storage TOAL of TOAL, path, a projection conversion load module storage projection modular converter raw data place.It is following to form flow process; The load module of TOAL points to the processing module of TOAL; The processing module result of calculation of TOAL is by the load module of the output module sensing NDSI of TOAL, and the processing module result of calculation of NDSI is by an input of the output module sensing projection modular converter of NDSI, and another projection conversion load module of storage projection modular converter raw data points to another input of projection modular converter; The result of projection conversion points to projection conversion output module, shows final calculation result.When calculating this flow process, system finds two lines that can walk abreast through analyzing, and after calculating is accomplished respectively, finally calculates the projection conversion, accomplishes and calculates.
Claims (6)
1. a remote sensing computing method for parallel processing is specially: 1) set up remote sensing computing basic module; 2) calculate according to remote sensing, choose required remote sensing computing basic module and form remote sensing computing process module; 3) convert remote sensing computing basic module and remote sensing computing process module into executable file, realize parallel processing.
2. remote sensing computing method for parallel processing as claimed in claim 1 is characterized in that, said remote sensing computing basic module comprises load module, output module and processing module, wherein,
Load module is used in reference to needed each original input data of remote sensing computing;
Processing module is used for carrying out the remote sensing computing according to original input data;
Output module is used in reference to the store path to the remote sensing operation result.
3. remote sensing computing method for parallel processing as claimed in claim 2 is characterized in that said load module, output module and processing module all have the Xml descriptor of standard.
4. remote sensing computing method for parallel processing as claimed in claim 3 is characterized in that said Xml descriptor comprises attribute information, algorithm information, model information and input information.
5. remote sensing computing method for parallel processing as claimed in claim 4; It is characterized in that; Said remote sensing computing process module also comprises the Xml descriptor; This Xml descriptor comprises all load modules, output module and the processing module of forming remote sensing computing process module, and the mutual relationship between all load modules, output module and the processing module.
6. remote sensing computing method for parallel processing as claimed in claim 5 is characterized in that, the interface and the C# that carry through IDL or Matlab carry out shuffling, converts remote sensing computing basic module and remote sensing computing process module into executable file.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101955942A CN102393815A (en) | 2011-07-13 | 2011-07-13 | Parallel processing method of remote sensing operation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101955942A CN102393815A (en) | 2011-07-13 | 2011-07-13 | Parallel processing method of remote sensing operation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102393815A true CN102393815A (en) | 2012-03-28 |
Family
ID=45861144
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011101955942A Pending CN102393815A (en) | 2011-07-13 | 2011-07-13 | Parallel processing method of remote sensing operation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102393815A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4527271A (en) * | 1982-08-17 | 1985-07-02 | The Foxboro Company | Process control system with improved fault isolation |
| CN1959717A (en) * | 2006-10-09 | 2007-05-09 | 北京道达天际软件技术有限公司 | System and method for preprocessing mass remote sensing data collection driven by order form |
| CN101315424A (en) * | 2008-07-29 | 2008-12-03 | 中国科学院对地观测与数字地球科学中心 | Multi-satellite remote sensing data integrated parallel ground pretreatment system |
-
2011
- 2011-07-13 CN CN2011101955942A patent/CN102393815A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4527271A (en) * | 1982-08-17 | 1985-07-02 | The Foxboro Company | Process control system with improved fault isolation |
| CN1959717A (en) * | 2006-10-09 | 2007-05-09 | 北京道达天际软件技术有限公司 | System and method for preprocessing mass remote sensing data collection driven by order form |
| CN101315424A (en) * | 2008-07-29 | 2008-12-03 | 中国科学院对地观测与数字地球科学中心 | Multi-satellite remote sensing data integrated parallel ground pretreatment system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhao et al. | Bridging the gap between deep learning and sparse matrix format selection | |
| DE102020110688A1 (en) | HARDWARE-AGNOSTIC COMPILER FOR DEEP NEURAL NETWORKS | |
| Sehgal et al. | Guidelines and benchmarks for deployment of deep learning models on smartphones as real-time apps | |
| EP4145351A1 (en) | Neural network construction method and system | |
| CN104267938B (en) | A kind of method and device of the quick development deployment of the application of streaming computing | |
| CN105378689B (en) | Method and apparatus for unified Extensible Firmware Interface (UEFI) driver and agreement | |
| CN105243167A (en) | Data processing method and device | |
| CN101604241A (en) | A kind of code generating method of embedded system | |
| WO2020143236A1 (en) | Method, device, and equipment for accelerating convolutional neural network, and storage medium | |
| CN107508698B (en) | Software defined service reorganization method based on content perception and weighted graph in fog calculation | |
| Dou et al. | Scientific workflow design 2.0: Demonstrating streaming data collections in Kepler | |
| CN113626038A (en) | Code conversion method, device, equipment and storage medium | |
| Qian et al. | R-cnn object detection inference with deep learning accelerator | |
| CN104537713B (en) | A kind of novel three-dimensional reconfiguration system | |
| WO2022247110A1 (en) | Task processing method and apparatus, and electronic device and storage medium | |
| Yao et al. | VenusAI: An artificial intelligence platform for scientific discovery on supercomputers | |
| Ma et al. | Deep generative model driven protein folding simulation | |
| CN102253837B (en) | Based on the Design of Software Framework method of object tree | |
| Ramos-Pollán et al. | BIGS: A framework for large-scale image processing and analysis over distributed and heterogeneous computing resources | |
| CN102393815A (en) | Parallel processing method of remote sensing operation | |
| Peluso et al. | Inference on the edge: Performance analysis of an image classification task using off-the-shelf cpus and open-source convnets | |
| Frajberg et al. | Accelerating deep learning inference on mobile systems | |
| CN109359146A (en) | A kind of automating ETL data processing tools and its application method | |
| Xia et al. | PAI-FCNN: FPGA based inference system for complex CNN models | |
| Moreland et al. | Flexible analysis software for emerging architectures |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20120328 |