CN109376206A - The path optimal selection intelligent method of remote sensing image parallel processing based on data segmentation - Google Patents
The path optimal selection intelligent method of remote sensing image parallel processing based on data segmentation Download PDFInfo
- Publication number
- CN109376206A CN109376206A CN201811070090.6A CN201811070090A CN109376206A CN 109376206 A CN109376206 A CN 109376206A CN 201811070090 A CN201811070090 A CN 201811070090A CN 109376206 A CN109376206 A CN 109376206A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- sensing image
- computer
- parallel
- parallel processing
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 90
- 238000012545 processing Methods 0.000 title claims abstract description 85
- 230000011218 segmentation Effects 0.000 title claims abstract description 40
- 238000003860 storage Methods 0.000 claims abstract description 20
- 230000008569 process Effects 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000004590 computer program Methods 0.000 claims description 10
- 238000009472 formulation Methods 0.000 claims description 5
- 239000000203 mixture Substances 0.000 claims description 5
- 230000003993 interaction Effects 0.000 claims description 4
- 238000007620 mathematical function Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims description 2
- 238000005192 partition Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims 3
- 230000006870 function Effects 0.000 description 16
- 238000005520 cutting process Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000007906 compression Methods 0.000 description 4
- 230000006835 compression Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 2
- 239000010931 gold Substances 0.000 description 2
- 229910052737 gold Inorganic materials 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/38—Concurrent instruction execution, e.g. pipeline or look ahead
- G06F9/3885—Concurrent instruction execution, e.g. pipeline or look ahead using a plurality of independent parallel functional units
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/20—Processor architectures; Processor configuration, e.g. pipelining
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to remote sensing and geographical information system field, the path optimal selection intelligent method of specially a kind of remote sensing image parallel processing based on data segmentation.The present invention first obtains the remote sensing image of input, calculates the calculating cost under the unit information amount of concurrent processor automatically further according to training sampleThe relative entropy r and element number n of every step processing remote sensing image, then calculate the calculating cost ratio psi under the unit information amount of remote sensing image concurrent processorp/φ′pWith the relative entropy and element number ratio (r of every step processing remote sensing imagej×ni)/(rj+1×ni+1), finally according to this ratio in judgement and obtain the path optimal path of remote sensing image parallel processing.It handles the dynamic of path caused by multichannel accessibility possessed by model, optimal selection when the present invention solves the problems, such as concurrent technique to be applied to remote sensing image distributed storage and process field, principle is simple and practical, and time-consuming is most short when meeting parallel processing, most efficiently requires.
Description
Technical field
The invention belongs to remote sensing and geographical information system fields, and in particular to a kind of remote sensing shadow based on data segmentation
As the path optimal selection intelligent method of parallel processing.
Background technique
In raster data structure, space is regularly divided into grid (generally square).The position of geographical entity
It is that the grid line that is occupied with them, row number define with state.The size of each grid represents the spatial decomposition power of definition.
Grid cell acquirement is smaller, and the decomposing force of data is higher, but in contrast, and the data volume of data file is bigger.
Raster data can be regarded as 2 dimension matrixes, the quaternary tree Groupe de la Pyramide mode or small echo gold word of raster data
Tower organizational form in fact forms a kind of parallel/distributed storage organization.That is, data are carried out on a three-dimensional space
Distribution.The coordinate system of this three-dimensional space is the X of two-dimensional space, Y-resolution grade respectively.Z table is fastened in this three-dimensional coordinate
What is shown is the color value or height value of DOM and dem data.In this data back system, not in same stage resolution ratio
Data with region have irrelevance.The data of each level in range there is upper and lower level to cover relationship, due to difference point
The data of resolution grade have certain information multiplicity, and the high data information of stage resolution ratio has included the low data letter of stage resolution ratio
Breath.Therefore, during pyramid generates, low resolution grade data are generated with high-resolution grade data.And in same data
In layer, because data are uncorrelated, data cannot be generated mutually.Resolution ratio is higher, and incoherent data are more, i.e. domain decomposability
Better.Otherwise resolution ratio is low, and incoherent data are just few, and domain decomposability is with regard to poor.Due to the data between same stage resolution ratio
There is no correlation, therefore, carries out domain and decompose the cost that will not bring any communication.But for the different resolution of the same area
Data carry out domain and decompose the increasing for then bringing the traffic.Therefore, the data based on Groupe de la Pyramide have preferable domain decomposability.
It include different treatment processes in remote sensing image parallel process.Such as in Fig. 2 in cutting procedure along with
Data compression.However, the example in Fig. 1 is the process of an information content reduction.The target detection of Fig. 3 is then that an information content increases
The process added.The oval target to detect in figure, and vector polygon and vector point are automatically generated, so as to cause information
Amount increases.It is not difficult to find out that the parallel process of either quaternary tree index or the parallel process of target detection
To be abstracted into model shown in Fig. 4.
Cutting operation is known as longitudinal direction parallel more than the parallel process of non-cutting operation, and cutting operation is grasped less than non-segmentation
The parallel process of work is known as laterally parallel.Fig. 5 gives the parallel and laterally parallel schematic diagram in more general longitudinal direction.It should
Figure indicates the process of the different data states (segmentation data state and other data states) obtained from source data state by chain.From figure
Know that either a or b is an incomplete binary tree.It is apparent from, each bifurcated on incomplete binary tree indicates one
Kind carries out a possibility that parallel.If each mapping process is considered as a parallel task, parallel dimension is the mapping of d
Journey can form the parallel subtask of d item.Multichannel accessibility causes laterally the choosing with longitudinal parallel two kinds of parallel routes parallel
It selects.It is apparent from, if remote sensing image processing function does not change, parallel route does not change.Implement in specific parallel processing
In the process, how to allow computer according to processing the optimal parallel route of function dynamic select will be the key that distributed storage strategy
Problem.
Summary of the invention
The purpose of the present invention is overcome the deficiencies of the prior art and provide it is a kind of based on data segmentation remote sensing image locate parallel
The path optimal selection intelligent method of reason.
The path optimal selection intelligent method of remote sensing image parallel processing provided by the invention based on data segmentation, specifically
Steps are as follows:
1) remote sensing image of pending parallel processing is inputted;
2) judge whether by training? if if true, skipping to step 13);Otherwise continue step 3);
3) quantity and title of Remote Sensing Image Segmentation program He other concurrent processors is manually entered;
4) the calculating cost under the unit information amount of remote sensing image concurrent processor is manually enteredEvery step processing is distant
Feel the relative entropy r and relative elemental number n of image;Wherein, the calculating cost under unit information amountRefer to certain parallel processing
Program handles the time that one or a set of remote sensing image information pixel is paid, and unit is the second;Relative entropy r refers to the distant of input
Feeling image amount of information is 1, by the ratio of the information content of other grade of remote sensing image product and the information content of the remote sensing image of former input;
Relative elemental number n refers to every quantity for generating level-one remote sensing image product;
Does is 5) judgement manually entered reliable? if if true, skipping to step 10);Otherwise continue step 6);
6) training sample is taken from the remote sensing image of input;
7) the calculating cost under the unit information amount of all segmentation procedures and other concurrent processors is calculated
8) the relative entropy r of every step processing remote sensing image is calculated;
9) the element number n of every step processing remote sensing image is calculated;
10) the ratio Φ of the calculating cost under the unit information amount of " concurrent processor to " is calculatedP/Φ’P;
11) relative entropy of every step processing remote sensing image and the ratio (r of relative elemental number are calculatedj×ni)/(rj+1×
ni+1);
12) ratio is stored in database or file;
13) database reads the calculating cost ratio Φ under the unit information amount of input remote sensing image concurrent processorP/
Φ’PWith the relative entropy of every step processing remote sensing image and the ratio (r of relative elemental numberj×ni)/(rj+1×ni+1);
14) judge ΦP/Φ’PWhether (r is greater thanj×ni)/(rj+1×ni+1)? if if true, continuing step 15);Otherwise
Skip to step 16);
15) parallel route for carrying out other parallel work-flows again is first divided in selection;
16) selection first carries out the parallel route that other parallel work-flows are split again;
17) it is handled according to all image files of the parallel route of formulation to input;
18) the path optimal selection intelligent method of the remote sensing image parallel processing based on data segmentation terminates.
Beneficial effects of the present invention
The remote sensing image and specified remote sensing image processing routine that the present invention can be such that computer inputs automatically according to user
Parallel processing time shortest optimal path is accurately obtained, solves and concurrent technique is applied to remote sensing image distributed storage
With path dynamic, optimal selection problem caused by multichannel accessibility possessed by its processing model when process field, principle letter
Single practical, time-consuming is most short when meeting parallel processing, most efficiently requires.
Detailed description of the invention
Fig. 1 is flow chart of the present invention.
Fig. 2 is that quaternary tree index generates schematic diagram (four layers).
Fig. 3 is Raster Images target detection schematic diagram.
Fig. 4 is the schematic diagram of the general parallel process of Raster Images.
Fig. 5 is longitudinal parallel and laterally parallel general schematic diagram.
Specific embodiment
The remote sensing image and specified remote sensing image processing routine that the present invention can be such that computer inputs automatically according to user
Parallel processing time shortest optimal path is accurately obtained, solves and concurrent technique is applied to remote sensing image distributed storage
With path dynamic, optimal selection problem caused by multichannel accessibility possessed by its processing model when process field, principle letter
Single practical, time-consuming is most short when meeting parallel processing, most efficiently requires.
Specific embodiments of the present invention are as follows:
1) remote sensing image of pending parallel processing is inputted;
2) judge whether by training? if if true, skipping to step 13);Otherwise continue step 3);
3) quantity and title of Remote Sensing Image Segmentation program He other concurrent processors is manually entered;
4) the calculating cost under the unit information amount of remote sensing image concurrent processor is manually enteredEvery step handles remote sensing
The relative entropy r and relative elemental number n of image;
Does is 5) judgement manually entered reliable? if if true, skipping to step 10);Otherwise continue step 6);
6) training sample is taken from the remote sensing image of input;
7) the calculating cost under the unit information amount of all segmentation procedures and other concurrent processors is calculated
8) the relative entropy r of every step processing remote sensing image is calculated
9) the element number n of every step processing remote sensing image is calculated
10) the ratio Φ of the calculating cost under the unit information amount of " concurrent processor to " is calculatedP/Φ’P;
11) relative entropy of every step processing remote sensing image and the ratio (r of relative elemental number are calculatedj×ni)/(rj+1×
ni+1);
12) ratio is stored in database or file;
13) database reads the calculating cost ratio Φ under the unit information amount of input remote sensing image concurrent processorP/
Φ’PWith the relative entropy of every step processing remote sensing image and the ratio (r of relative elemental numberj×ni)/(rj+1×ni+1);
14) judge ΦP/Φ’PWhether (r is greater thanj×ni)/(rj+1×ni+1)? if if true, continuing step 15);Otherwise
Skip to step 16);
15) parallel route for carrying out other parallel work-flows again is first divided in selection;
16) selection first carries out the parallel route that other parallel work-flows are split again;
17) it is handled according to all image files of the parallel route of formulation to input;
18) the path optimal selection intelligent method of the remote sensing image parallel processing based on data segmentation terminates.
The method of the remote sensing image of the pending parallel processing of input described in step 1): when program processing starts, manually
Input and inform that the remote sensing image file storage location of the pending parallel processing of computer program, computer program call I/O interface
Open file.
Judge whether described in step 2) by training? if if true, skipping to step 13);Otherwise continue step 3)
Method: computer program defines a global variable TFlag, wherein TFlag is initially set to false, is defaulted as without instruction
Practice, reads whether trained value of statistical indicant carries out assignment to it from stored file and data library, computer condition judges language
Sentence carry out logic judgment, if if true, invocation step 13) computer function;Otherwise continue invocation step 3) computer letter
Number.
The quantity and title of Remote Sensing Image Segmentation program He other concurrent processors are manually entered described in step 3)
Method: by human-computer interaction interface, the function data and title of parallel processing is manually entered, and store and arrive storage file or number
According in library.
The calculating cost under the unit information amount of remote sensing image concurrent processor is manually entered described in step 4)The method of the relative entropy r and relative elemental number n of every step processing remote sensing image: according to the experience and theory of the mankind
Value, by human-computer interaction interface, is manually entered the calculating cost under the unit information amount of remote sensing image concurrent processor
The relative entropy r and relative elemental number n of every step processing remote sensing image, and store into storage file or database.
Does is judgement described in step 5) manually entered reliably? if if true, skipping to step 10);Otherwise continue step 6)
Method: computer program defines a global variable FFlag, wherein and FFlag is initially set to false, and it is unreliable to be defaulted as, from
Read whether reliable value of statistical indicant carries out assignment to it in stored file and data library, computer condition judges that sentence carries out logic
Judgement, if if true, invocation step 10) computer function;Otherwise continue invocation step 6) computer function, executing
The computer function of the step can compare the computer result pair of step 7), step 8) and step 9) automatically after this step
FFlag is modified.
The method for taking a training sample from the remote sensing image of input described in step 6): calling computer IO function,
Each pixel of remote sensing image is read line by line.
Calculating under the unit information amount of all segmentation procedures of calculating described in step 7) and other concurrent processors
CostMethod: when starting to process one or one group of pixel during computer single treatment, when computer records system
Between be tBegin, be disposed during computer single treatment one or one group of pixel when, it is t that computer, which records system time,Eventually,
Calculating cost under unit information amountIf the calculating under the unit information amount of the computer program of parallel partition
Cost isThen remember that the calculating cost under the unit information amount of other paired parallel remote sensing image concurrent processors is
Φ’P。
The method of the relative entropy r of the every step processing remote sensing image of calculating described in step 8): computer calls IO to connect
Mouthful, statistics concurrent processor generates the size of the remote sensing image product All Files of every level-one, and unit is M (million), if raw
J grades of products are produced, then every grade of information content is rj。
The element number n: computer of the every step processing remote sensing image of calculating described in step 9) calls I/O interface, and statistics is simultaneously
Row processing routine generates the quantity of all remote sensing image files of remote sensing image product of every level-one, and unit is a, if i grades of production
Product, then every grade of element number is ni。
The ratio Φ of calculating cost under the unit information amount of calculating described in step 10) " concurrent processor to "P/
Φ’PMethod: computer obtains calculating cost or the list that is calculated of step 7) under the unit information amount of step 4) input
Calculating cost under the information content of position, calls " removing " mathematical function of computer floating type, according to formula ΦP/Φ’P, obtain ratio
Variable.
The relative entropy of the every step processing remote sensing image of calculating described in step 11) and the ratio of relative elemental number
(rj×ni)/(rj+1×ni+1) method: computer obtain step 4) input relative entropy and relative elemental number, or step
The rapid relative entropy 7) being calculated and relative elemental number call " multiplication and division " mathematical function of computer floating type, according to
Formula (rj×ni)/(rj+1×ni+1), obtain ratio variable.
The method that ratio is stored in database or file described in step 12): calling I/O interface, by step 10) and step
The rapid variable storage 11) being calculated in deposit storage file or database, and whether trained global variable
TFlag is set to ture, is stored in storage file or database.
Database described in step 13) reads the calculating under the unit information amount of input remote sensing image concurrent processor
Cost ratio ΦP/Φ’PWith the relative entropy of every step processing remote sensing image and the ratio (r of relative elemental numberj×ni)/
(rj+1×ni+1) method: call I/O interface, the unit information amount of concurrent processor read from storage file or database
Under calculating cost ratio ΦP/Φ’PWith the relative entropy of every step processing remote sensing image and the ratio (r of relative elemental numberj
×ni)/(rj+1×ni+1).
Judgement Φ described in step 14)P/Φ’PWhether (r is greater thanj×ni)/(rj+1×ni+1)? if if true, continuing
Step 15);Otherwise it skips to the method for step 16): calling computer condition discriminant function, judge ΦP/Φ’PWhether (r is greater thanj×
ni)/(rj+1×ni+1), if if true, invocation step 15) computer function;Otherwise continue invocation step 16) computer
Function.
The method for the parallel route for carrying out other parallel work-flows again is first divided in selection described in step 15): definition is global
Variable MFlag carries out the indexed variable of the parallel route of other parallel work-flows as first dividing again, and MFlag is set to true, and
It is stored in file and database.
The method for selecting first to carry out the parallel route that other parallel work-flows are split again described in step 16): definition
Global variable MFlag carries out the indexed variable of the parallel route of other parallel work-flows as first dividing again, and MFlag is set to
False, and be stored in file and database.
The method handled described in step 17) according to all image files of the parallel route of formulation to input:
The condition discriminant function for calling computer, judges the value of MFlag, and if it is true, computer is arranged in automatically carries out parallel remote sensing
When image processing, parallel calling segmentation function carries out the function of other operations to the Landsat images of segmentation again;Otherwise, computer from
Dynamic to be arranged in when carrying out the processing of parallel remote sensing image, parallel calling first carries out the function of other operations again to remote sensing to remote sensing image
The function that image is split.
The path optimal selection intelligent method knot of remote sensing image parallel processing based on data segmentation described in step 18)
The method of beam: the algorithm terminates after computer and arrangement.
Embodiment:
The path optimal selection intelligence side for the remote sensing image parallel processing divided based on data is realized using C# language programming
Method, using 10 calculate nodes, to certain inshore image data (format: TIF;Data volume 197.8M;Size 7129 × 7151;
Pixel depth 8bit) it is handled, 8,10,12 layers of quaternary tree pyramid are constructed, total time is recorded.It is read after being disposed
MFlag value, discovery MFlag value are vacation, that is, know that computer automatically using the paralleling tactic divided again is first compressed, is then forced to set
Setting MFlag value is very, to force computer using the paralleling tactic for first dividing recompression, equally constructs 8,10,12 layers of quaternary tree gold
Word tower records total time.Request compression and cutting operation respectively again, then the calculate node in cloud uses ENVI/IDL algorithm respectively
Compression (compressing a image data by the pyramidal compression method of quaternary tree) and segmentation are executed (by a image data
64 parts of segmentation) processing, difference recording compressed and sliced time, and calculate the calculating generation under identical environment under unit information content
Valence.The present embodiment experiment operation 100 times, take its mean value as experimental result.The test knot obtained using different parallel routes
Fruit is as shown in table 1;It can be seen that computer has automatically selected the higher path of efficiency.The unit of measurement compression and cutting operation
Calculating cost under information content is as shown in table 2.Also correctness of the invention is demonstrated.
1. raster data A paralleling tactic test result (unit: minute) of table
Calculating cost measurement result under 2 unit information amount of table
Claims (18)
1. a kind of path optimal selection intelligent method of the remote sensing image parallel processing based on data segmentation, which is characterized in that tool
Steps are as follows for body:
1) remote sensing image of pending parallel processing is inputted;
2) judge whether by training? if if true, skipping to step 13);Otherwise continue step 3);
3) quantity and title of Remote Sensing Image Segmentation program He other concurrent processors is manually entered;
4) the calculating cost under the unit information amount of remote sensing image concurrent processor is manually enteredEvery step handles remote sensing image
Relative entropy r and relative elemental number n;Wherein, the calculating cost under unit information amountRefer at certain concurrent processor
The time that one or a set of remote sensing image information pixel is paid is managed, unit is the second;Relative entropy r refers to the remote sensing image of input
Information content is 1, by the ratio of the information content of other grade of remote sensing image product and the information content of the remote sensing image of former input;It is relatively first
Plain number n refers to every quantity for generating level-one remote sensing image product;
Does is 5) judgement manually entered reliable? if if true, skipping to step 10);Otherwise continue step 6);
6) training sample is taken from the remote sensing image of input;
7) the calculating cost under the unit information amount of all segmentation procedures and other concurrent processors is calculated
8) the relative entropy r of every step processing remote sensing image is calculated;
9) the element number n of every step processing remote sensing image is calculated;
10) the ratio Φ of the calculating cost under the unit information amount of " concurrent processor to " is calculatedP/Φ’P;
11) relative entropy of every step processing remote sensing image and the ratio (r of relative elemental number are calculatedj×ni)/(rj+1×
ni+1);
12) ratio is stored in database or file;
13) database reads the calculating cost ratio Φ under the unit information amount of input remote sensing image concurrent processorP/Φ’P
With the relative entropy of every step processing remote sensing image and the ratio (r of relative elemental numberj×ni)/(rj+1×ni+1);
14) judge ΦP/Φ’PWhether (r is greater thanj×ni)/(rj+1×ni+1)? if if true, continuing step 15);Otherwise step is skipped to
It is rapid 16);
15) parallel route for carrying out other parallel work-flows again is first divided in selection;
16) selection first carries out the parallel route that other parallel work-flows are split again;
17) it is handled according to all image files of the parallel route of formulation to input;
18) the path optimal selection intelligent method of the remote sensing image parallel processing based on data segmentation terminates.
2. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 1 based on data segmentation
Method, which is characterized in that the method for the remote sensing image of the pending parallel processing of input described in step 1) are as follows: program processing starts
When, it is manually entered and informs the remote sensing image file storage location of the pending parallel processing of computer program, computer program tune
File is opened with I/O interface.
3. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 1 based on data segmentation
Method, which is characterized in that judge whether described in step 2) by training? if if true, skipping to step 13);Otherwise continue to walk
Rapid method 3) are as follows: computer program defines a global variable TFlag, wherein TFlag is initially set to false, is defaulted as not
By training, read whether trained value of statistical indicant carries out assignment, computer condition to it from stored file and data library
Judge sentence carry out logic judgment, if if true, invocation step 13) computer function;Otherwise continue invocation step 3) meter
Calculation machine function.
4. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 1 based on data segmentation
Method, which is characterized in that the quantity of Remote Sensing Image Segmentation program He other concurrent processors is manually entered described in step 3)
With the method for title are as follows: by human-computer interaction interface, the function data and title of parallel processing, and storage to storage is manually entered
In file or database.
5. the path optimal selection intelligence of the remote sensing image parallel processing based on data segmentation described in one of -4 according to claim 1
Energy method, which is characterized in that be manually entered described in step 4) under the unit information amount of remote sensing image concurrent processor
Calculate costThe method of the relative entropy r and relative elemental number n of every step processing remote sensing image are as follows: according to the warp of the mankind
It tests and by human-computer interaction interface the calculating under the unit information amount of remote sensing image concurrent processor is manually entered in theoretical value
CostThe relative entropy r and relative elemental number n of every step processing remote sensing image, and store and arrive storage file or database
In;Wherein, the calculating cost under unit information amountRefer to that certain concurrent processor handles one or a set of remote sensing image information picture
The time that member is paid, unit is the second;Relative entropy r refers to that the remote sensing image information amount of input is 1, by other grade of remote sensing image
The ratio of the information content of product and the information content of the remote sensing image of former input;Relative elemental number n refers to every generation level-one remote sensing shadow
As the quantity of product.
6. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 5 based on data segmentation
Method, which is characterized in that judgement described in step 5) is manually entered reliably? if if true, skipping to step 10);Otherwise continue to walk
Rapid method 6) are as follows: computer program defines a global variable FFlag, wherein FFlag is initially set to false, is defaulted as not
Reliably, read whether reliable value of statistical indicant carries out assignment to it from stored file and data library, computer condition judges sentence
Carry out logic judgment, if if true, invocation step 10) computer function;Otherwise continue invocation step 6) computer letter
Number, the computer function of the step can compare the computer of step 7), step 8) and step 9) automatically after executing this step
As a result FFlag is modified.
7. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 6 based on data segmentation
Method, which is characterized in that the method for taking a training sample from the remote sensing image of input described in step 6) are as follows: call computer
IO function reads each pixel of remote sensing image line by line.
8. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 7 based on data segmentation
Method, which is characterized in that under the unit information amount of all segmentation procedures of calculating described in step 7) and other concurrent processors
Calculating costMethod are as follows: when starting to process one or one group of pixel during computer single treatment, computer note
The recording system time is tBegin, be disposed during computer single treatment one or one group of pixel when, computer records system
Time is tEventually, calculating cost under unit information amountIf the unit information amount of the computer program of parallel partition
Under calculating cost beThen remember the meter under the unit information amount of other paired parallel remote sensing image concurrent processors
Calculation cost is Φ 'P。
9. the path optimal selection intelligence of the remote sensing image parallel processing based on data segmentation according to one of claim 6-8
Energy method, which is characterized in that the method for the relative entropy r of the every step processing remote sensing image of calculating described in step 8): calculate
Machine calls I/O interface, and statistics concurrent processor generates the size of the remote sensing image product All Files of every level-one, unit M
(million), if j grades of products of production, every grade of information content is rj。
10. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 9 based on data segmentation
Method, which is characterized in that the element number n: computer of the every step processing remote sensing image of calculating described in step 9) calls I/O interface,
Statistics concurrent processor generates the quantity of all remote sensing image files of remote sensing image product of every level-one, and unit is a, if
I grades of products are produced, then every grade of element number is ni。
11. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 10 based on data segmentation
Method, which is characterized in that the ratio of the calculating cost under the unit information amount of calculating described in step 10) " concurrent processor to "
Value ΦP/Φ’PMethod: computer obtain step 4) input unit information amount under calculating cost or step 7) be calculated
Unit information amount under calculating cost, call computer floating type " removing " mathematical function, according to formula ΦP/Φ’P, obtain
Ratio variable.
12. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 11 based on data segmentation
Method, which is characterized in that the relative entropy of the processing remote sensing image of the every step of calculating described in step 11) and relative elemental number
Ratio (rj×ni)/(rj+1×ni+1) method: computer obtain step 4) input relative entropy and relative elemental number,
Or relative entropy and relative elemental number that step 7) is calculated, " multiplication and division " mathematical function of computer floating type is called,
According to formula (rj×ni)/(rj+1×ni+1), obtain ratio variable.
13. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 12 based on data segmentation
Method, which is characterized in that the method that ratio is stored in database or file described in step 12): I/O interface is called, by step
10) and the variable storage that is calculated of step 11) is in deposit storage file or database, and whether the trained overall situation
Variable TFlag is set to ture, is stored in storage file or database.
14. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 13 based on data segmentation
Method, which is characterized in that database described in step 13) is read under the unit information amount of input remote sensing image concurrent processor
Calculating cost ratio ΦP/Φ’PWith the relative entropy of every step processing remote sensing image and the ratio (r of relative elemental numberj×
ni)/(rj+1×ni+1) method: call I/O interface, the unit letter of concurrent processor read from storage file or database
Calculating cost ratio Φ under breath amountP/Φ’PWith the relative entropy of every step processing remote sensing image and the ratio of relative elemental number
(rj×ni)/(rj+1×ni+1)。
15. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 14 based on data segmentation
Method, which is characterized in that judgement Φ described in step 14)P/Φ’PWhether (r is greater thanj×ni)/(rj+1×ni+1)? if it is true
Then continue step 15);Otherwise it skips to the method for step 16): calling computer condition discriminant function, judge ΦP/Φ’PIt is whether big
In (rj×ni)/(rj+1×ni+1), if if true, invocation step 15) computer function;Otherwise continue invocation step 16)
Computer function.
16. the path for the remote sensing image parallel processing divided described in one of 0-15 based on data according to claim 1 is most preferably
Select intelligent method, which is characterized in that the parallel route for carrying out other parallel work-flows again is first divided in selection described in step 15)
Method: definition global variable MFlag, will as the indexed variable for carrying out the parallel route of other parallel work-flows again is first divided
MFlag is set to true, and is stored in file and database.
17. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 16 based on data segmentation
Method, which is characterized in that selection described in step 16) first carries out the side for the parallel route that other parallel work-flows are split again
Method: global variable MFlag is defined as the indexed variable for carrying out the parallel route of other parallel work-flows again is first divided, by MFlag
It is set to false, and is stored in file and database.
18. the path optimal selection intelligence side of the remote sensing image parallel processing according to claim 17 based on data segmentation
Method, which is characterized in that handled described in step 17) according to all image files of the parallel route of formulation to input
Method are as follows: the condition discriminant function for calling computer judges the value of MFlag, and if it is true, computer is arranged in progress simultaneously automatically
When the processing of row remote sensing image, parallel calling segmentation function carries out the function of other operations to the Landsat images of segmentation again;Otherwise, it counts
When calculation machine is arranged in progress parallel remote sensing image processing automatically, parallel calling first carries out the function of other operations again to remote sensing image
The function that remote sensing image is split.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811070090.6A CN109376206B (en) | 2018-09-13 | 2018-09-13 | Intelligent path optimal selection method for remote sensing image parallel processing based on data segmentation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811070090.6A CN109376206B (en) | 2018-09-13 | 2018-09-13 | Intelligent path optimal selection method for remote sensing image parallel processing based on data segmentation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109376206A true CN109376206A (en) | 2019-02-22 |
CN109376206B CN109376206B (en) | 2021-06-22 |
Family
ID=65405198
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811070090.6A Active CN109376206B (en) | 2018-09-13 | 2018-09-13 | Intelligent path optimal selection method for remote sensing image parallel processing based on data segmentation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109376206B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101126639A (en) * | 2007-09-18 | 2008-02-20 | 武汉大学 | Quick low altitude remote sensing image automatic matching and airborne triangulation method |
CN101604018A (en) * | 2009-07-24 | 2009-12-16 | 中国测绘科学研究院 | High-definition remote sensing image data disposal route and system thereof |
US9275690B2 (en) * | 2012-05-30 | 2016-03-01 | Tahoe Rf Semiconductor, Inc. | Power management in an electronic system through reducing energy usage of a battery and/or controlling an output power of an amplifier thereof |
CN107391258A (en) * | 2017-07-20 | 2017-11-24 | 武汉大学 | A kind of portable remote sensing image real time processing system of software and hardware one |
CN107610054A (en) * | 2017-07-13 | 2018-01-19 | 中国农业大学 | A kind of preprocess method of remote sensing image data |
-
2018
- 2018-09-13 CN CN201811070090.6A patent/CN109376206B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101126639A (en) * | 2007-09-18 | 2008-02-20 | 武汉大学 | Quick low altitude remote sensing image automatic matching and airborne triangulation method |
CN101604018A (en) * | 2009-07-24 | 2009-12-16 | 中国测绘科学研究院 | High-definition remote sensing image data disposal route and system thereof |
US9275690B2 (en) * | 2012-05-30 | 2016-03-01 | Tahoe Rf Semiconductor, Inc. | Power management in an electronic system through reducing energy usage of a battery and/or controlling an output power of an amplifier thereof |
CN107610054A (en) * | 2017-07-13 | 2018-01-19 | 中国农业大学 | A kind of preprocess method of remote sensing image data |
CN107391258A (en) * | 2017-07-20 | 2017-11-24 | 武汉大学 | A kind of portable remote sensing image real time processing system of software and hardware one |
Non-Patent Citations (1)
Title |
---|
张博: "高分辨率遥感影像多尺度分类方法研究", 《信息科技辑》 * |
Also Published As
Publication number | Publication date |
---|---|
CN109376206B (en) | 2021-06-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US12001807B2 (en) | Applied artificial intelligence technology for narrative generation using an invocable analysis service | |
US20180225391A1 (en) | System and method for automatic data modelling | |
CN102622335B (en) | Automated table transformations from examples | |
CN106022245A (en) | Multi-source remote sensing satellite data parallel processing system and method based on algorithm classification | |
Azam et al. | Discrete cosine transform (DCT) based face recognition in hexagonal images | |
CN114048676A (en) | Seismic fault and horizon interpretation method and device based on multitask deep learning | |
CN110069573A (en) | Product data integration method, apparatus, computer equipment and storage medium | |
CN109961129A (en) | A kind of Ocean stationary targets search scheme generation method based on improvement population | |
CN114387608A (en) | Table structure identification method combining convolution and graph neural network | |
Golubovic et al. | A scalable system for executing and scoring K-means clustering techniques and its impact on applications in agriculture | |
Shekhar et al. | Hierarchical approximations for data reduction and learning at multiple scales | |
CN109376206A (en) | The path optimal selection intelligent method of remote sensing image parallel processing based on data segmentation | |
Stewart et al. | World spatiotemporal analytics and mapping project (wstamp): Discovering, exploring, and mapping spatiotemporal patterns across the world’s largest open soruce data sets | |
Alhaj Ali et al. | Distributed data mining systems: techniques, approaches and algorithms | |
Garcia | A generic approach to spatial individual-based modelling and simulation of plant communities | |
CN109523031A (en) | A kind of big data intelligence machine learning system for depth analysis | |
Dugan et al. | Phase 4: Dcl system using deep learning approaches for land-based or ship-based real-time recognition and localization of marine mammals-distributed processing and big data applications | |
Gallaher et al. | Data rods: High speed, time-series analysis of massive cryospheric data sets using pure object databases | |
CN109460700B (en) | Crop classification-oriented remote sensing data processing method and device | |
AU2020202075A1 (en) | Processing and storage of location data | |
Waseem et al. | Issues and Challenges of KDD Model for Distributed Data Mining Techniques and Architecture | |
Wang | Cellular matrix for parallel k-means and local search to Euclidean grid matching | |
Malott et al. | Distributed Computation of Persistent Homology from Partitioned Big Data | |
CN110276381A (en) | A kind of Data Dimensionality Reduction Algorithm decomposed based on collimation geodesic curve with low-dimensional insertion | |
Ponnaganti et al. | Clustering Multiple Views of Data through Many-objective Evolutionary Approach |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |