CN103106323B - Obtain method and the device of the structure descriptor of figure - Google Patents
Obtain method and the device of the structure descriptor of figure Download PDFInfo
- Publication number
- CN103106323B CN103106323B CN201110379668.8A CN201110379668A CN103106323B CN 103106323 B CN103106323 B CN 103106323B CN 201110379668 A CN201110379668 A CN 201110379668A CN 103106323 B CN103106323 B CN 103106323B
- Authority
- CN
- China
- Prior art keywords
- information
- node
- initial
- time series
- descriptor
- 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.)
- Expired - Fee Related
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a kind of method of structure descriptor obtaining figure, comprising: the structure extracting figure in given image; Being information stream cardon by the Structure Conversion of described figure, described information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of described figure, each node carries out information flow according to the information flow rule in described information stream cardon; Determine the initial quantity of information in described information stream cardon of each node in the structure of described figure; Initial quantity of information according to described each node, according to described information flow rule, calculates the quantity of information of described each node at default time point; Time series according to the information distribution formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure. Present invention achieves the performance analysis of the structure to figure, obtain the Dynamic profiling information of the structure of figure, add the quantity of information of the structure descriptor of figure, it is possible to more effectively describe the structure of figure.
Description
Technical field
Relate generally to technical field of image processing of the present invention, the method for especially a kind of structure descriptor obtaining figure and device.
Background technology
In image processing process, often need to extract the structure of figure in image, such as, from the electronic circuit electronic applications, circuit structure is extracted, as shown in Figure 1, again such as, from the various network architecture (such as communication network, internet, the network of communication lines), extract network data structure, chemical field or biological field extract molecular structure or DNA, RNA structure etc.
After the structure extracting figure, the descriptor of the structure of figure can be obtained further, for comparing fast, sort and searching figure. Wherein, the descriptor of the structure of figure connects node by a group node and one group limit and the information structure being attached on node and limit.
In prior art, after the structure extracting figure from image, by adopting the numeric representation method of the structure of the figure based on graph theory, represent the structure descriptor of figure based on the method (graphentropy) etc. of the spectral analysis method (graphspectrummethod) of figure and the entropy based on figure. But, the static value in above-mentioned method for expressing being all the static analysis based on the structure to figure and obtain, it is too simple that it analyzes process, the static value obtained can not the structural information of abundant phenogram, the structure of figure can not effectively be described.
Summary of the invention
In view of this, embodiments provide method and the device of a kind of structure descriptor obtaining figure, it is possible to the better structural information of phenogram, the structure of more effective description figure.
An aspect according to embodiments of the present invention, it is provided that a kind of method of structure descriptor obtaining figure, comprising:
Given image extracts the structure of figure;
Being information stream cardon by the Structure Conversion of described figure, described information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of described figure, each node carries out information flow according to the information flow rule in described information stream cardon;
Determine the initial quantity of information in described information stream cardon of each node in the structure of described figure;
Initial quantity of information according to described each node, according to described information flow rule, calculates the quantity of information of described each node at default time point;
Time series according to the information distribution formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure.
Another aspect according to embodiments of the present invention, it is provided that a kind of graph structure control methods, comprising:
By the method according to the above-mentioned embodiment of the present invention, obtain the structure descriptor of to be contrasted at least two figure;
If the structure descriptor of described at least two figure is identical, then the structure of at least two figure to be contrasted described in is identical.
Another aspect according to embodiments of the present invention, it is provided that the device of a kind of structure descriptor obtaining figure, comprising:
Structure extraction unit, is configured to the structure of extraction figure in given image;
Figure conversion unit, being configured to the Structure Conversion of described figure is information stream cardon, described information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of described figure, each node carries out information flow according to the information flow rule in described information stream cardon;
Initial value determining unit, the initial quantity of information being configured to determine in the structure of described figure each node in described information stream cardon;
Information computing unit, is configured to the initial quantity of information according to described each node, according to described information flow rule, calculates the quantity of information of described each node at default time point;
Information determination unit, is configured to the time series according to the information distribution formed by described each node in the quantity of information of all default time points, it is determined that the structure descriptor of described figure.
In addition, according to a further aspect in the invention, a kind of storage media is additionally provided. Described storage media comprises the program code that machine can be read, and when performing described program code on information processing device, described program code makes the execution of described information processing device according to the method for the structure descriptor of the above-mentioned acquisition figure of the present invention.
, in accordance with a further aspect of the present invention, in addition a kind of program product is additionally provided. Described program product comprises the instruction that machine can perform, and when performing described instruction on information processing device, described instruction makes the execution of described information processing device according to the method for the structure descriptor of the above-mentioned acquisition figure of the present invention.
Aforesaid method according to embodiments of the present invention, after the structure extracting figure, by being information stream cardon by the thaumatropy of figure, obtain the quantity of information distribution of each node and time correlation, achieve the performance analysis of the structure to figure, and then obtain the Dynamic profiling information of the structure of figure, thus add the quantity of information of the structure descriptor of figure, fully characterize the structure of figure, it is possible to more effectively describe the structure of figure.
Specification sheets part below provides other aspects of the embodiment of the present invention, wherein, the preferred embodiment being used for fully the open embodiment of the present invention is described in detail, and does not apply to limit to it.
Accompanying drawing explanation
Below in conjunction with specific embodiment, and with reference to accompanying drawing, the above-mentioned of the embodiment of the present invention and other object and advantage are further described. In the accompanying drawings, the identical or corresponding Reference numeral of employing is represented by identical or correspondence technology feature or parts.
Fig. 1 is the schematic diagram illustrating and extracting circuit structure in prior art from electronic circuit;
Fig. 2 is the method flow diagram of the structure descriptor illustrating the acquisition figure as embodiment of the present invention offer;
Fig. 3 illustrates that what provide as the embodiment of the present invention is the method flow diagram of Markov figure by the Structure Conversion of figure;
Fig. 4 a is the structural representation illustrating and providing a kind of figure as the embodiment of the present invention;
Fig. 4 b is the schematic diagram of the Markov figure after illustrating the thaumatropy providing Fig. 4 a as the embodiment of the present invention;
Fig. 5 is the structural representation illustrating and providing a kind of figure as the embodiment of the present invention;
Fig. 6 is the structural representation illustrating and providing a kind of figure as the embodiment of the present invention;
Fig. 7 illustrates to provide, as the embodiment of the present invention, the method flow diagram determining the initial quantity of information of each node;
Fig. 8 illustrates to calculate the method flow diagram of each node in the quantity of information of default time point as what the embodiment of the present invention provided;
Fig. 9 is the method flow diagram of the structure descriptor illustrating the determination figure as embodiment of the present invention offer;
Figure 10 is the method flow diagram illustrating another the structure descriptor determining figure provided as the embodiment of the present invention;
Figure 11 is the method flow diagram illustrating the comparison that the structure descriptor based on figure provided as the embodiment of the present invention carries out graph structure;
Figure 12 is the apparatus structure schematic diagram illustrating a kind of structure descriptor obtaining figure provided as the embodiment of the present invention;
Figure 13 is the structural representation illustrating a kind of information computing unit provided as the embodiment of the present invention;
Figure 14 is the structural representation illustrating a kind of information determination unit provided as the embodiment of the present invention;
Figure 15 is the structural representation illustrating a kind of initial value determining unit provided as the embodiment of the present invention;
Figure 16 is the structural representation illustrating another kind of information determination unit provided as the embodiment of the present invention;
Figure 17 is the block diagram of the example arrangement illustrating the Personal Computer as the information processing device adopted in embodiments of the invention.
Embodiment
Embodiments of the invention are described with reference to the accompanying drawings.
See Fig. 2, the method for the structure descriptor of the acquisition figure that the embodiment of the present invention provides comprises:
Step 201: the structure extracting figure in given image.
In the present embodiment, first the structure obtaining figure from image is needed, such as shown in Fig. 1, the structure extracting circuit in electronic circuit image is (such as, Goto, S., " Anefficientalgorithmforthetwo-dimensionalplacementproble minelectricalcircuitlayout ", IEEETransactionsonCircuitsandSystems, vol.28, pp.12-18, 1981), or, (such as communication network from the various network architecture, internet, the network of communication lines) extract network data structure (such as, Schenker, A., " Graph-TheoreticTechniquesforWebCotentMining ", WorldScientific, Singapore, 2005), chemical field or biological field extract molecular structure or DNA, RNA structure is (such as, Faulon, J.L., " Isomorphism, automorphismpartitioning, andcanonicallabelingcanbesolvedinpolynomial-timeformocul argraphs ", JournalofChemicalInformationandComputerSciences, vol.38, pp.432-444, 1998), natural language processing field extracts syntax structure (such as, Biemann, C., " Chinesewhispers:anefficientgraphclusteringalgorithmandit sapplicationtonaturallanguageprocessingproblems ", ProcessingoftheFirstWorkshoponGraphBasedMethodsforNatura lLanguageProcessing, pp.73-80, 2006), and other such as word, fingerprint, image is (such as, Riesen, K., " IAMgraphdatabaserepositoryforgraphbasedpatternrecognitio nandmachinelearning ", Structure, Syntactic, andStatisticalPatternRecognition, pp.287-297, 2008) etc.Each leaching process can adopt prior art, and its detailed process repeats no more.
The Structure Conversion of figure is information stream cardon by step 202.
After the structure obtaining figure, according to the node in the structure of figure and the line between node, it is information stream cardon by the Structure Conversion of this figure, based on this information stream cardon, the structure of this figure is carried out performance analysis so that follow-up, obtain dynamic descriptor.
Wherein, this information stream cardon is in order to characterize by the information flow space of the structure definition of figure, and in the structure of figure, each node carries out information flow according to the information flow rule in information stream cardon.
Step 203, it is determined that the initial quantity of information of each node in information stream cardon in the structure of figure.
Before the quantity of information of each node carries out flowing or propagating according to information flow rule, need the initial quantity of information first determining each node, the determination of this initial quantity of information can be unrelated with the structure of figure, such as directly determine according to the rule (such as univesral distribution) etc. set in advance, can also determine based on the structure of figure, such as, if the structure of figure is non-regular graph structure, then the initial quantity of information of each node is distributed according to the principle of univesral distribution, also namely the initial quantity of information of each node is identical, equals the mean value of gross information content at each node; If the structure of figure is canonical graph structure, when determining the initial quantity of information of each node, it is possible to select one of them node as initial node, it is then determined that the initial quantity of information of this initial node is 1, the initial quantity of information of other node is 0, and the selection of this initial node can be arbitrary. Further, in order to obtain the complete descriptor of the structure of this figure, the selection of this initial node needs repeating query between each node. Above two kinds of modes, specifically please refer to the description of subsequent embodiment.
Step 204, according to the initial quantity of information of each node, according to information flow rule, calculates the quantity of information of each node at default time point.
After determining the initial quantity of information of each node, in the information stream cardon that the structure of figure is corresponding, the quantity of information of each node is according to information flow rule start information flowing or propagates, and not in the same time or default time point, can obtain the quantity of information of each node at current time. Wherein, the time point preset has multiple, can set as required, such as, if the initial moment of setting node is first default time point, also namely the initial quantity of information of node is the node quantity of information under the first default time point, then the quantity of the time point preset can be identical with the quantity of node, naturally it is also possible to different. Can set as required as the interval time between time point, concrete obtain the quantity of information of node under each default time point time, it is possible to and interval between time point is unrelated.
Step 205, according to the time series of the information distribution formed in the quantity of information of all default time points by each node, it is determined that the structure descriptor of figure.
After obtaining the quantity of information that each node presets time point in difference respectively, each node all may correspond to the time series of a quantity of information distribution, according to the time series that the quantity of information of all nodes distributes, can determine the structure descriptor of figure. Specifically, can directly using the time series that obtains as the descriptor of the structure of this figure, it is also possible to this time series is carried out certain process, such as sequence etc., then with the time series after sorting, or a certain sequence selected after sequence is as the descriptor of the structure of figure.
The embodiment of the present invention is after the structure extracting figure, by being information stream cardon by the thaumatropy of figure, obtain the quantity of information distribution of each node and time correlation, achieve the performance analysis of the structure to figure, and then obtain the Dynamic profiling information of the structure of figure, thus add the quantity of information of the structure descriptor of figure, fully characterize the structure of figure, it is possible to more effectively describe the structure of figure.
After given image extracts the structure of figure, it is information stream cardon by the thaumatropy of figure, in one exemplary embodiment of the present invention, this information stream cardon is described for Markov figure, it it is then the process of Markov figure by the Structure Conversion of this figure, as shown in Figure 3, it is possible to comprise the following steps:
Step 301, increases the oriented limit pointing to oneself in the structure of figure on each node.
The structure of figure is described for Fig. 4 a, 4b, comprises a, b, c, d tetra-nodes in the structure of this figure, increases the oriented limit pointing to oneself on each node, and such as Fig. 4 b interior joint a points to the limit of oneself.
Step 302, determines the oriented limit between corresponding node according to the line between the structure interior joint of figure.
As shown in fig. 4 a, line is there is between node a, b and between a, d, and two line all without sensing, also undirected limit it is, then two lines corresponding two oriented limits respectively, a as corresponding in line between ab points to the oriented limit of b and the oriented limit of b sensing a, and the oriented limit of the corresponding a sensing d of the line between ad and d point to the oriented limit of a, and the line between other node is similar.
In another embodiment, if the line between two nodes is oriented limit, then this line only corresponding oriented limit, this oriented limit is identical with the sensing of line between node.
Wherein, the order of step 301 and step 302 can adjust.
Step 303, based on the weight on each oriented limit of original tally distributed computation acquisition of line between node and node.
After obtaining oriented limits all in the structure of figure, the structure of figure exists the original tally distribution of line between node and node, such as, in Fig. 4 a, the original tally of node a is 1, and the original tally of node d is 4, ..., the original tally of line ab is 5, and the original tally of line ad is 7 ..., then distribute according to this original tally, the weight obtaining each oriented limit can be calculated.
When calculating weight, first to the weight normalization method on each node, such as, mark total value on node a is 1 (mark of node a)+5 (mark of limit ab)+6 (mark of limit ac)+7 (mark of limit ad)=19, similar, mark total value on node b is 5+2+8=15, and the mark total value of node c is 18, and the mark total value of node d is 28.
Then, calculate the weights W on each oriented limit. wherein, the weights W on oriented limit is that the mark total value on the node that flows out of the mark value of the line corresponding to this oriented limit and this oriented limit carries out calculating, such as, node a is flowed out on the oriented limit of a to b, then the weights W ab on the oriented limit of a to b is (mark value of ab line)/(the mark total value of node a)=5/19, and node b is flowed out on the oriented limit of b to a, then the weights W ba on the oriented limit of b to a is (mark value of ab line)/(the mark total value of node b)=5/15, similar, Wac=6/19, Wca=6/18, Wcd=9/18, Wdc=9/28, Wbd=8/15, Wdb=8/28. in addition, for the oriented limit of sensing self, similar calculating can obtain, and Waa, Wbb, Wcc, Wdd are followed successively by 1/19,2/15,3/18,4/28.
After above three steps, the Markov figure after the thaumatropy of Fig. 4 a can be obtained, as shown in Figure 4 b.
Structure for other figure adopts above-mentioned steps can obtain its Markov figure too, as shown in Figure 5. In addition, for the structure not having the figure of the mark value of line between clear and definite flag node and node, as shown in Figure 6, then the original tally that can be defaulted as its node is 1, and between node, the original tally of wire is 1, and the Markov figure of its correspondence is as shown in Figure 6.
In the thus obtained structure comprising figure, each node points to the oriented limit of oneself, oriented limit corresponding to line between node, and the Markov figure of the weight on each oriented limit obtained based on the original tally distributed computation of line between node and node, can be used as the information stream cardon of the structure of figure. In this Markov figure, the quantity of information that the rule of information flow is each node carries out information flow according to the weight on the oriented limit between node and oriented limit.
After being information stream cardon by the thaumatropy of figure, in the structure of figure, the quantity of information of each node flows according to the flow rule in information stream cardon, before carrying out information flow, need the initial quantity of information first determining each node, in one exemplary embodiment of the present invention, as shown in Figure 7, it is determined that the method for the initial quantity of information of each node can comprise the following steps:
Step 701, the structure of judgement figure is non-canonical graph structure or canonical graph structure.
In the present embodiment, the initial quantity of information of decision node is carried out according to the structure of figure. The quantity on limit flowed out when each node in the structure of figure is identical, and when each node weight of flowing out limit is also identical, this figure is canonical figure, otherwise is non-regular graph.
Step 702, if non-regular graph structure, using the gross information content of information stream cardon at the mean value of each node as the initial quantity of information of each node.
If non-regular graph structure, then can think figure structure in the initial quantity of information of each node identical, be the mean value of gross information content at all nodes, such as, be non-regular graph shown in Fig. 4 a, Fig. 6.
Step 703, if canonical graph structure, then selects initial node in each node.
In canonical figure, the selection of this initial node is arbitrary, and in order to obtain the complete descriptor of the structure of this figure, the selection of this initial node needs repeating query between each node.
Step 704, is defined as 1 by the initial quantity of information of initial node in information stream cardon, and the initial quantity of information of other node in information stream cardon is defined as 0.
It is canonical figure as shown in Figure 5.
After determining the initial quantity of information of node, further according to the initial quantity of information of each node, according to information flow rule, calculate the quantity of information of each node at default time point, in one exemplary embodiment of the present invention, calculate the process of each node in the quantity of information of default time point, as shown in Figure 8, it is possible to comprise the following steps:
Step 801, the quantity of information calculating each node current preset time point respectively flows to the product of the weight on the oriented limit of a certain node with it.
First, assume that node j is Pjt in the quantity of information of current preset time point t, it is Wji that this node j flows to the weight on the oriented limit of node i, then the product of the weight that node j flows to the oriented limit of node i in the quantity of information of t with it is PjtWji, wherein, node i, j is the node in the structure of figure, i, j are only for identification nodes, node i, j can be same node (the oriented limit that now node j flows to node i also i.e. this node point to the oriented limit of oneself), it is also possible to be different nodes.
Namely such as, the initial quantity of information of node j is Pj0, and also the quantity of information of node j under first default time point (t0) is Pj0. It is Wji that this node j flows to the weight on the oriented limit of node i.
Then the product of the weight that node j flows to the oriented limit of node i in the quantity of information in t0 moment with it is PjoWji.
Step 802, using product corresponding for each node of obtaining and as the quantity of information of this certain node at next default time point.
Assume total n node in the structure of this figure, then node i is at the quantity of information Pit+1 of next default time point t+1
Such as, the quantity of information Pi1 of node i under the 2nd default time point (t1) is
According to the method described above, as long as determining the initial quantity of information of each node, the quantity of information of each node under n default time point (t0～tn-1) can be obtained.Wherein, the initial quantity of information of each node can set in advance, it is also possible to is that the structure based on figure is determined, such as canonical figure or non-regular graph etc.
At each node of acquisition after the quantity of information of default time point, the time series of each node information distribution can be obtained, and then the structure descriptor of figure can be determined accordingly. In one exemplary embodiment of the present invention, if the structure of this figure is non-regular graph structure, then this determines the process of the structure descriptor of figure, as shown in Figure 9, it is possible to comprise the following steps:
Step 901, obtains the time series of the information distribution that each node is formed in the quantity of information of all default time points respectively.
In the present embodiment, in order to obtain the information of the structure being enough to description figure, the quantity of the default time point of regulation at least equals the quantity of the structure interior joint of figure, also namely presets in time point t0～tn-1, and n just equals the quantity of node.
Then according to aforementioned method, the quantity of information of node i under n default time point (t0～tn-1) is: Pi0, Pi1 ..., Pin-1. For non-regular graph, owing to the initial quantity of information of each node is identical, it is univesral distribution, so, the time series of the information distribution that all nodes are formed in the quantity of information of all default time points in the structure of figure, an available matrix representation, is:
Being described for Fig. 4 a, 4b, the structure of this figure is non-regular graph, the initial quantity of information univesral distribution between node a, b, c, d, then Pa0=Pb0=Pc0=Pd0=1/4 (first presets the quantity of information under time point t0).
Based on aforementioned formula: node i is in the quantity of information in t+1 momentKnown:
Weight according to each oriented limit as shown in fig 4b, under the 2nd default time point (t1), the quantity of information of each node is distributed as:
Pa1=Waa*Pa0+Wba*Pb0+Wca*Pc0+Wda*Pd0=0.242325
Pb1=Wbb*Pb0+Wab*Pa0+Wdb*Pd0=0.170551
Pc1=Wcc*Pc0+Wac*Pa0+Wdc*Pd0=0.200971
Pd1=Wdd*Pd0+Wad*Pa0+Wbd*Pb0+Wcd*Pc0=0.386153
Under 3rd default time point (t2), the quantity of information of each node is distributed as:
Pa2=Waa*Pa1+Wba*Pb1+Wca*Pc1+Wda*Pd1=0.233133
Pb2=Wbb*Pb1+Wab*Pa1+Wdb*Pd1=0.196839
Pc2=Wcc*Pc1+Wac*Pa1+Wdc*Pd1=0.234139
Pd2=Wdd*Pd1+Wad*Pa1+Wbd*Pb1+Wcd*Pc1=0.335888
Analogize with this, the quantity of information of each node under the 4th default time point (t3) can be obtained and be distributed as:
Pa3, Pb3, Pc3, Pd3 are followed successively by 0.239902,0.183564,0.220608,0.355926
Then in this Fig. 4 a, the time series of the information distribution that all nodes are formed in the quantity of information of all default time points is:
Step 902, sorts according to the first pre-defined rule to the time series of the information distribution of each node.
The information distribution time series of the structure that can uniquely represent this figure to obtain, this step can also be performed, the time series of the information distribution of each node is sorted, wherein, the first pre-defined rule performed by sequence can be the ordering rule of the obtained unique orderings result set in advance, such as according to predeterminated position order successively to the value of correspondence position in sequence by greatly to little or by little to carrying out greatly sequence etc. In the present embodiment, the sequencing put at preset timed intervals to the quantity of information sequence of each node by little to sorting greatly, wherein, due to Pa0=Pb0=Pc0=Pd0, then according to the value of Pa1, Pb1, Pc1, Pd1 by little to sorting greatly, can obtain:
Step 903, using the time series after sequence as the structure descriptor of this figure.
Namely time series after this sequence can be used as the descriptor of the structure of this figure, certainly the concrete manifestation form of this time series, it is possible to be with the form of above-mentioned matrix, it is also possible to is the sequence etc. that numerical value is arranged in order, to this manifestation, is not construed as limiting herein. Owing to the time series after this sequence can the structure of only table diagram, it may also be useful to the time series after this sequence, it is possible to compare the structure of different figure.
In other embodiments, certainly, if not requiring to obtain unique structure descriptor representing this figure, also namely without the need to obtaining the information distribution time series of unique structure representing this figure, then step 902 can be omitted.
In another exemplary embodiment of the present invention, if the structure of this figure is canonical graph structure, then this determines the process of the structure descriptor of figure, as shown in Figure 10, it is possible to comprise the following steps:
Step 1001, respectively using different nodes as initial node, obtains the time series of the information distribution that initial node and other node are formed in the quantity of information of all default time points.
In the present embodiment with non-regular graph the difference is that, the initial quantity of information of non-regular graph interior joint is according to univesral distribution, and each initial quantity of information is identical, and obtainable each node is distributed as a matrix in the quantity of information of default time point. And for canonical figure, due to the initial quantity of information not univesral distribution of each node, each node can as initial node, so, for the matrix that different node distributes as quantity of information corresponding different during initial node, that is, if there being n node, then corresponding n different quantity of information distribution matrix.
Similar with non-regular graph in the present embodiment it is, obtaining the time series of information distribution corresponding to each initial node, the information of the structure being enough to description figure to obtain, the quantity of the default time point adopted at least equals the quantity of the structure interior joint of figure, also namely presetting in time point t0～tn-1, n just equals the quantity of node. The quantity of information of node i under n default time point (t0～tn-1) is: Pi0, Pi1 ..., Pin-1. In the structure of this figure when any node is as initial node, the time series of the information distribution that all nodes are formed in the quantity of information of all default time points, an all available matrix representation, is:
When different node is as initial node, corresponding matrix is different.
In this step, first using different nodes as initial node, obtain the time series of all information distributions. Hereinafter, it is described for the canonical figure shown in Fig. 5.
In Fig. 5, the original tally of each node is 1, and each node flows out the quantity on limit and each weight all identical (being 1/4) flowing out limit.
When selecting node a as initial node, then Pa0=1, Pb0=Pc0=Pd0=Pe0=Pf0=0 (first presets the quantity of information under time point t0).
Based on aforementioned formula: node i is in the quantity of information in t+1 momentKnown:
Weight according to oriented limit each in Fig. 5 be 1/4 it will be seen that under the 2nd default time point (t1) quantity of information of each node be distributed as:
Pa1=Waa*Pa0+Wba*Pb0+Wca*Pc0+Wfa*Pf0=0.25
Pb1=Wbb*Pb0+Wab*Pa0+Wcb*Pc0+Web*Pe0=0.25
Pc1=Waa*Pa0+Wba*Pb0+Wca*Pc0+Wfa*Pf0=0.25
Pd1=Wdd*Pd0+Wcd*Pc0+Wed*Pe0+Wfd*Pf0=0
Pe1=Wee*Pe0+Wbe*Pb0+Wde*Pd0+Wfe*Pf0=0
Pf1=Wff*Pf0+Waf*Pa0+Wef*Pe0+Wdf*Pd0=0.25
Analogize with this, the quantity of information of each node under the 3rd to the 6th default time point (t2～t5) can be obtained and be distributed as:
Pa2, Pb2, Pc2, Pd2, Pe2, Pf2 are followed successively by 0.25,0.1875,0.1875,0.125,0.125,0.125
Pa3, Pb3, Pc3, Pd3, Pe3, Pf3 are followed successively by 0.1875,0.1875,0.1875,0.1406,0.1406,0.1562
Pa4, Pb4, Pc4, Pd4, Pe4, Pf4 are followed successively by 0.1797,0.1758,0.1758,0.1562,0.1562,0.1562
Pa5, Pb5, Pc5, Pd5, Pe5, Pf5 are followed successively by 0.1719,0.1719,0.1719,0.1611,0.1611,0.1621
Then when selecting node a as initial node, all nodes in the time series of the information distribution that the quantity of information of all default time points is formed are:
With reference to above method of calculation, acquisition can be calculated successively: during using node b as initial node (Pb0=1, Pa0=Pc0=Pd0=Pe0=Pf0=0), the time series of the information distribution that all nodes are formed in the quantity of information of all default time points;During using node c as initial node (Pc0=1, Pa0=Pb0=Pd0=Pe0=Pf0=0), the time series of the information distribution that all nodes are formed in the quantity of information of all default time points; During using node d as initial node (Pd0=1, Pa0=Pb0=Pc0=Pe0=Pf0=0), the time series of the information distribution that all nodes are formed in the quantity of information of all default time points; During using node e as initial node (Pe0=1, Pa0=Pb0=Pc0=Pd0=Pf0=0), the time series of the information distribution that all nodes are formed in the quantity of information of all default time points; During using node f as initial node (Pf0=1, Pa0=Pb0=Pc0=Pd0=Pe0=0), the time series of the information distribution that all nodes are formed in the quantity of information of all default time points.
Step 1002, sorts according to the 2nd pre-defined rule to the time series of the information distribution of each node corresponding to identical initial node.
The information distribution time series of the structure that can uniquely represent this figure to obtain, this step can also be performed, the time series of the information distribution of each node corresponding to identical initial node is sorted respectively, wherein, the first pre-defined rule performed by sequence can be the ordering rule of the obtained unique orderings result set in advance, such as according to predeterminated position order successively to the value of correspondence position in sequence by greatly to little or by little to carrying out greatly sequence etc. In the present embodiment, the sequencing put at preset timed intervals to the quantity of information sequence of each node by little to sorting greatly.
It is described for node a as the information distribution time series of each node corresponding to initial node.
To this time series According to the sequencing of default time point to the value of the quantity of information sequence of each node by little to sorting greatly, first, value according to Pa0, Pb0, Pc0, Pd0, Pe0, Pf0 by little to sorting greatly, owing to the value of Pb0, Pc0, Pd0, Pe0, Pf0 is identical, then continue the value according to Pb1, Pc1, Pd1, Pe1, Pf1 by little to sorting greatly, analogizing with this, after finally can sorting, to be the time series of a be initial node:
With managing the time series taking other node as initial node after can obtaining sequence: Pb, Pc, Pd, Pe, Pf.
Step 1003, according to preset selective rule, the time series of the information distribution of each node that a certain initial node after selected and sorted is corresponding is as the structure descriptor of figure.
The ordered series of numbers that can obtain the figure that depends on each node according to above step represents, further, the sequence of a specific figure can be selected from the sequence that n node is correlated with, remove the dependence to node location, this selective rule can be made an appointment as required, such as, after the time series taking each node as initial node being described as sequence form being sorted, fixing selection jth sequence wherein is as the structure descriptor of this figure, simply, this jth sequence can be first sequence, also namely minimum one. In the present embodiment, owing to Pa, Pb, Pc, Pd, Pe and Pf are all identical, so using (PdtPetPftPbtPctPat) as the structure descriptor of this figure.
In other embodiments, certainly, if not requiring to obtain unique structure descriptor representing this figure, also namely without the need to obtaining the information distribution time series of unique structure representing this figure, then step 1002 can be omitted.
The embodiment of the present invention passes through aforesaid method, after the structure extracting figure, by being information stream cardon by the thaumatropy of figure, obtain the quantity of information distribution of each node and time correlation, achieve the performance analysis of the structure to figure, and then obtain the Dynamic profiling information of the structure of figure, thus add the quantity of information of the structure descriptor of figure, fully characterize the structure of figure, it is possible to more effectively describe the structure of figure.
After the structure descriptor obtaining figure based on above method, in another exemplary embodiment of the present invention, it is also possible to carry out the comparison of graph structure based on the structure descriptor of figure, method as shown in figure 11, it is possible to comprising:
Step 1101, obtains the structure descriptor of to be contrasted at least two figure.
When whether the structure needing to compare multiple figure is identical, also the homotype problem of i.e. figure, first obtains the structure descriptor of all figure according to aforementioned method, also namely obtains a time series or matrix.
Step 1102, whether the structure descriptor comparing figure is identical.
If identical, then illustrate that the structure of figure is identical, if different, then illustrate that the structure of figure is different.
The comparison method of the structure of above figure may be used for the structure comparison of two figure or multiple figure.
It is more than the description to the inventive method embodiment, below the device embodiment realizing aforesaid method is introduced.
See Figure 12, obtain the apparatus structure schematic diagram of the structure descriptor of figure for the embodiment of the present invention is a kind of.
This device can comprise:
Structure extraction unit 1201, is configured to the structure of extraction figure in given image.
Figure conversion unit 1202, being configured to the Structure Conversion of figure is information stream cardon, this information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of figure, each node carries out information flow according to the information flow rule in described information stream cardon.
Initial value determining unit 1203, the initial quantity of information being configured to determine in the structure of figure each node in information stream cardon.
Information computing unit 1204, is configured to the initial quantity of information according to each node, according to information flow rule, calculates the quantity of information of each node at default time point.
Information determination unit 1205, is configured to the time series according to the information distribution formed by each node in the quantity of information of all default time points, it is determined that the structure descriptor of figure.
After structure extraction unit 1201 extracts the structure of figure from given image, by figure conversion unit 1202 according to the node in the structure of figure and the line between node, it is information stream cardon by the Structure Conversion of this figure, based on this information stream cardon, the structure of this figure is carried out performance analysis so that follow-up, obtain dynamic descriptor. Before the quantity of information of each node carries out flowing or propagating according to information flow rule, initial value determining unit 1203 first determines the initial quantity of information of each node, and the determination of this initial quantity of information can be unrelated with the structure of figure, it is also possible to the structure based on figure is determined. Such as, if the structure of figure is non-regular graph structure, then the initial quantity of information of each node is distributed according to the principle of univesral distribution, and also namely the initial quantity of information of each node is identical, equals the mean value of gross information content at each node; If the structure of figure is canonical graph structure, when determining the initial quantity of information of each node, it is possible to select one of them node as initial node, it is then determined that the initial quantity of information of this initial node is 1, the initial quantity of information of other node is 0.The selection of this initial node can be arbitrary, and after determining the initial quantity of information of each node, information computing unit 1204 obtains each node not in the same time or the quantity of information of default time point. Wherein, the time point preset has multiple, it is possible to set as required. Such as, if the initial moment of setting node is first default time point, also namely the initial quantity of information of node is the node quantity of information under the first default time point, then the quantity of the time point preset can be identical with the quantity of node, naturally it is also possible to different. The time series that information determination unit 1205 distributes according to the quantity of information of all nodes, can determine the structure descriptor of figure.
The embodiment of the present invention passes through said units, achieve the performance analysis of the structure to figure, and then obtain the Dynamic profiling information of the structure of figure, thus add the quantity of information of the structure descriptor of figure, fully characterize the structure of figure, it is possible to more effectively describe the structure of figure.
Wherein, in one exemplary embodiment of the present invention, figure changes unit, and can be specifically configured to the Structure Conversion of described figure be Markov figure, described Markov figure comprises each node in the structure of described figure and points to the oriented limit of oneself, oriented limit corresponding to line between node, and the weight on each oriented limit obtained based on the original tally distributed computation of line between node and node; The quantity of information that described information flow rule is each node carries out information flow according to the weight on the oriented limit between node and oriented limit.
In another exemplary embodiment of the present invention, this information computing unit can comprise structure as shown in fig. 13 that:
First computation subunit 1301, the quantity of information being configured to calculate respectively each node current preset time point flows to the product of the weight on the oriented limit of a certain node with it.
2nd computation subunit 1302, be configured to using product corresponding for each node of obtaining and as the quantity of information of described a certain node at next default time point.
In another exemplary embodiment of the present invention, initial value determining unit specifically can be configured to when the structure of described figure is non-regular graph structure, using the gross information content of described information stream cardon at the mean value of each node as the initial quantity of information of described each node. Determine the initial quantity of information of each node in this initial value determining unit after, information determination unit specifically can comprise structure as shown in figure 14:
First acquisition subelement 1401, is configured to obtain respectively the time series of the information distribution that each node is formed in the quantity of information of all default time points.
First sequence subelement 1402, the time series being configured to the information distribution to each node sorts according to the first pre-defined rule.
First determines subelement 1403, is configured to the structure descriptor of the time series after using sequence as described figure.
In another exemplary embodiment of the present invention, initial value determining unit specifically can also comprise structure as shown in figure 15:
Chooser unit 1501, is configured to select initial node in each node when the structure of described figure is canonical graph structure
Initially determining subelement 1502, be configured to the initial quantity of information of described initial node in described information stream cardon is defined as 1, the initial quantity of information of other node in described information stream cardon is defined as 0.
When initial value determining unit according to above-mentioned chooser unit 1501, initially determine the initial quantity of information that each node determined by subelement 1502 after, information determination unit specifically can comprise structure as shown in figure 16:
2nd acquisition subelement 1601, is configured to respectively using different nodes as described initial node, obtains the time series of the information distribution that initial node and other node are formed in the quantity of information of all default time points
Second row sequence subelement 1602, the time series of the information distribution being configured to each node that identical initial node is corresponding sorts according to the 2nd pre-defined rule
2nd determines subelement 1603, is configured to according to preset selective rule, and the time series of the information distribution of each node that the initial node of one after selected and sorted is corresponding is as the structure descriptor of described figure.
The performance analysis of the structure to figure is achieved by each unit of said apparatus in the embodiment of the present invention, and then obtain the Dynamic profiling information of the structure of figure, thus add the quantity of information of the structure descriptor of figure, fully characterize the structure of figure, it is possible to more effectively describe the structure of figure.
In addition, it should also be noted that above-mentioned series of processes and device can also be realized by software and/or hardware and/or firmware. When being realized by software and/or firmware, from storage media or network to the computer with specialized hardware structure, such as the general purpose personal computer 1700 shown in Figure 17 installs the program forming this software, and this computer is when being provided with various program, it is possible to perform various function etc.
In fig. 17, central processing unit (CPU) 1701 performs various process according to the program stored in read-only storage (ROM) 1702 or from storing the program that part 1708 is loaded into random access memory (RAM) 1703. In RAM1703, also store the data required when CPU1701 performs various process etc. as required.
CPU1701, ROM1702 and RAM1703 are connected to each other via bus 1704. Input/output interface 1705 is also connected to bus 1704.
Following parts are connected to input/output interface 1705: importation 1706, comprise keyboard, mouse etc.; Export part 1707, comprise indicating meter, such as cathode tube (CRT), liquid-crystal display (LCD) etc., and loud speaker etc.; Store part 1708, comprise hard disk etc.; With communications portion 1709, comprise NIC such as LAN card, modulator-demodulator unit etc. Communications portion 1709 performs communication process via network such as Internet.
As required, driving mechanism 1710 is also connected to input/output interface 1705. Detachable media 1711 such as disk, CD, magneto-optic disk, semiconductor memory etc. are installed on driving mechanism 1710 as required so that the computer program therefrom read is installed to as required and stores in part 1708.
When series of processes above-mentioned by software simulating, from network such as Internet or storage media, such as detachable media 1711 installs the program forming software.
It will be understood by one skilled in the art that this kind of storage media be not limited to shown in Figure 17 wherein have program stored therein and equipment distributes the detachable media 1711 to provide program to user separately. The example of detachable media 1711 comprises disk (comprising floppy disk (registered trademark)), CD (comprising cd-rom (CD-ROM) and digital universal disc (DVD)), magneto-optic disk (comprising mini-disk (MD) (registered trademark)) and semiconductor memory. Or, storage media can be ROM1702, store in part 1708 hard disk that comprises etc., wherein computer program stored, and is distributed to user together with comprising their equipment.
Also it is noted that the step performing above-mentioned series of processes can naturally according to the order illustrated temporally order execution, but not need necessarily to perform according to time sequence. Some step can perform parallel or independently of one another.
The aforesaid method of the present invention and device embodiment can be applied in various image processing field about in the acquisition process of the structure descriptor of figure, from the electronic circuit electronic applications, such as extract the structure descriptor of acquisition figure after circuit structure, again such as, (such as communication network from the various network architecture, internet, the network of communication lines), extract the structure descriptor obtaining figure after network data structure, chemical field or biological field extract the structure descriptor etc. of acquisition figure after molecular structure or DNA, RNA structure.Such as, after the structure descriptor obtaining figure, so that it may figure is carried out various subsequent disposal by means of this kind of descriptor, whether relatively more each figure is identical, etc.
About the enforcement mode comprising above embodiment, following attached note is also disclosed:
Attached note 1. 1 kinds obtains the method for the structure descriptor of figure, comprising:
Given image extracts the structure of figure;
Being information stream cardon by the Structure Conversion of described figure, described information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of described figure, each node carries out information flow according to the information flow rule in described information stream cardon;
Determine the initial quantity of information in described information stream cardon of each node in the structure of described figure;
Initial quantity of information according to described each node, according to described information flow rule, calculates the quantity of information of described each node at default time point;
Time series according to the information distribution formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure.
The method of attached note 2. according to attached note 1, wherein, the described Structure Conversion by described figure is information stream cardon, comprising:
It is Markov figure by the Structure Conversion of described figure, described Markov figure comprises each node in the structure of described figure and points to the oriented limit of oneself, oriented limit corresponding to line between node, and the weight on each oriented limit obtained based on the original tally distributed computation of line between node and node;
The quantity of information that described information flow rule is each node carries out information flow according to the weight on the oriented limit between node and oriented limit.
The method of attached note 3. according to attached note 2, wherein, the described initial quantity of information according to described each node, according to described information flow rule, calculates described each node in the quantity of information of default time point, comprising:
The quantity of information calculating each node current preset time point respectively flows to the product of the weight on the oriented limit of a certain node with it;
Using product corresponding for each node obtained and as the quantity of information of described a certain node at next default time point.
Attached note 4. is according to method described in any one in attached note 1 to 3, and wherein, the initial quantity of information of each node in described information stream cardon in the described structure determining described figure, comprising:
When the structure of described figure is non-regular graph structure, using the gross information content of described information stream cardon at the mean value of each node as the initial quantity of information of described each node.
The method of attached note 5. according to attached note 4, wherein, the time series of the information distribution that described basis is formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure, comprising:
Obtain the time series of the information distribution that each node is formed in the quantity of information of all default time points respectively;
The time series of the information distribution of each node is sorted according to the first pre-defined rule;
Using the time series after sequence as the structure descriptor of described figure.
Attached note 6. is according to method described in any one in attached note 1 to 3, and wherein, the initial quantity of information of each node in described information stream cardon in the described structure determining described figure, comprising:
When the structure of described figure is canonical graph structure, each node is selected initial node;
The initial quantity of information of described initial node in described information stream cardon is defined as 1, and the initial quantity of information of other node in described information stream cardon is defined as 0.
The method of attached note 7. according to attached note 6, wherein, the time series of the information distribution that described basis is formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure, comprising:
Respectively using different nodes as described initial node, obtain the time series of the information distribution that initial node and other node are formed in the quantity of information of all default time points;
The time series of the information distribution of each node corresponding to identical initial node is sorted according to the 2nd pre-defined rule;
According to preset selective rule, the time series of the information distribution of each node that the initial node of one after selected and sorted is corresponding is as the structure descriptor of described figure.
The method of attached note 8. according to attached note 1, wherein, the quantity of described default time point at least equals the quantity of the structure interior joint of described figure.
The 9. 1 kinds of graph structure control methodss of attached note, comprising:
According to method described in any one in above-mentioned attached note 1 to 8, obtain the structure descriptor of at least two figure;
If the structure descriptor of described at least two figure is identical, then the structure of at least two figure described in is identical.
Attached note 10. 1 kinds obtains the device of the structure descriptor of figure, comprising:
Structure extraction unit, is configured to the structure of extraction figure in given image;
Figure conversion unit, being configured to the Structure Conversion of described figure is information stream cardon, described information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of described figure, each node carries out information flow according to the information flow rule in described information stream cardon;
Initial value determining unit, the initial quantity of information being configured to determine in the structure of described figure each node in described information stream cardon;
Information computing unit, is configured to the initial quantity of information according to described each node, according to described information flow rule, calculates the quantity of information of described each node at default time point;
Information determination unit, is configured to the time series according to the information distribution formed by described each node in the quantity of information of all default time points, it is determined that the structure descriptor of described figure.
The device of attached note 11. according to attached note 10, wherein,
Described figure changes unit, specifically being configured to the Structure Conversion of described figure is Markov figure, described Markov figure comprises each node in the structure of described figure and points to the oriented limit of oneself, oriented limit corresponding to line between node, and the weight on each oriented limit obtained based on the original tally distributed computation of line between node and node; The quantity of information that described information flow rule is each node carries out information flow according to the weight on the oriented limit between node and oriented limit.
The device of attached note 12. according to attached note 11, wherein, described information computing unit comprises:
First computation subunit, the quantity of information being configured to calculate respectively each node current preset time point flows to the product of the weight on the oriented limit of a certain node with it;
2nd computation subunit, be configured to using product corresponding for each node of obtaining and as the quantity of information of described a certain node at next default time point.
Attached note 13. according to device described in any one in attached note 10 to 12, wherein,
Described initial value determining unit, is specifically configured to when the structure of described figure is non-regular graph structure, using the gross information content of described information stream cardon at the mean value of each node as the initial quantity of information of described each node.
The device of attached note 14. according to attached note 13, wherein, described information determination unit comprises:
First acquisition subelement, is configured to obtain respectively the time series of the information distribution that each node is formed in the quantity of information of all default time points;
First sequence subelement, the time series being configured to the information distribution to each node sorts according to the first pre-defined rule;
First determines subelement, is configured to the structure descriptor of the time series after using sequence as described figure.
Attached note 15. is according to device described in any one in attached note 10 to 12, and wherein, described initial value determining unit comprises:
Chooser unit, is configured to select initial node in each node when the structure of described figure is canonical graph structure;
Initially determining subelement, be configured to the initial quantity of information of described initial node in described information stream cardon is defined as 1, the initial quantity of information of other node in described information stream cardon is defined as 0.
The device of attached note 16. according to attached note 15, wherein, described information determination unit comprises:
2nd acquisition subelement, is configured to respectively using different nodes as described initial node, obtains the time series of the information distribution that initial node and other node are formed in the quantity of information of all default time points;
Second row sequence subelement, the time series of the information distribution being configured to each node that identical initial node is corresponding sorts according to the 2nd pre-defined rule;
2nd determines subelement, is configured to according to preset selective rule, and the time series of the information distribution of each node that the initial node of one after selected and sorted is corresponding is as the structure descriptor of described figure.
Although specifically having understood the present invention and advantage thereof, it should be appreciated that various change, replacement and conversion can be carried out when not departing from the spirit and scope of the present invention limited by appended claim. And, the term " comprising " of the embodiment of the present invention, " comprising " or its any other variant are intended to contain comprising of nonexcludability, so that comprise the process of a series of key element, method, article or equipment not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise the key element intrinsic for this kind of process, method, article or equipment. When not more restrictions, the key element limited by statement " comprising ... ", and be not precluded within process, method, article or the equipment comprising described key element and also there is other identical element.
Claims (9)
1. obtain a method for the structure descriptor of figure, comprising:
Given image extracts the structure of figure;
Being information stream cardon by the Structure Conversion of described figure, described information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of described figure, each node carries out information flow according to the information flow rule in described information stream cardon;
Determine the initial quantity of information in described information stream cardon of each node in the structure of described figure;
Initial quantity of information according to described each node, according to described information flow rule, calculates the quantity of information of described each node at default time point;
Time series according to the information distribution formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure,
Wherein, the described Structure Conversion by described figure is information stream cardon, comprising:
It is Markov figure by the Structure Conversion of described figure, described Markov figure comprises each node in the structure of described figure and points to the oriented limit of oneself, oriented limit corresponding to line between node, and the weight on each oriented limit obtained based on the original tally distributed computation of line between node and node;
The quantity of information that described information flow rule is each node carries out information flow according to the weight on the oriented limit between node and oriented limit.
2. method according to claim 1, wherein, the described initial quantity of information according to described each node, calculates described each node in the quantity of information of default time point according to described information flow rule, comprising:
The quantity of information calculating each node current preset time point respectively flows to the product of the weight on the oriented limit of a certain node with it;
Using product corresponding for each node obtained and as the quantity of information of described a certain node at next default time point.
3. method according to claim 1 and 2, wherein, the initial quantity of information of each node in described information stream cardon in the described structure determining described figure, comprising:
When the structure of described figure is non-regular graph structure, using the gross information content of described information stream cardon at the mean value of each node as the initial quantity of information of described each node.
4. method according to claim 3, wherein, the time series of the information distribution that described basis is formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure, comprising:
Obtain the time series of the information distribution that each node is formed in the quantity of information of all default time points respectively;
The time series of the information distribution of each node is sorted according to the first pre-defined rule;
Using the time series after sequence as the structure descriptor of described figure.
5. method according to claim 1 and 2, wherein, the initial quantity of information of each node in described information stream cardon in the described structure determining described figure, comprising:
When the structure of described figure is canonical graph structure, each node is selected initial node;
The initial quantity of information of described initial node in described information stream cardon is defined as 1, and the initial quantity of information of other node in described information stream cardon is defined as 0.
6. method according to claim 5, wherein, the time series of the information distribution that described basis is formed in the quantity of information of all default time points by described each node, it is determined that the structure descriptor of described figure, comprising:
Respectively using different nodes as described initial node, obtain the time series of the information distribution that initial node and other node are formed in the quantity of information of all default time points;
The time series of the information distribution of each node corresponding to identical initial node is sorted according to the 2nd pre-defined rule;
According to preset selective rule, the time series of the information distribution of each node that a certain initial node after selected and sorted is corresponding is as the structure descriptor of described figure.
7. method according to claim 1, wherein, the quantity of described default time point at least equals the quantity of the structure interior joint of described figure.
8. a graph structure control methods, comprising:
According to method described in any one in the claims 1 to 7, obtain the structure descriptor of to be contrasted at least two figure;
If the structure descriptor of described at least two figure is identical, then the structure of at least two figure to be contrasted described in is identical.
9. obtain a device for the structure descriptor of figure, comprising:
Structure extraction unit, is configured to the structure of extraction figure in given image;
Figure conversion unit, being configured to the Structure Conversion of described figure is information stream cardon, described information stream cardon is in order to characterize by the information flow space of the structure definition of described figure, and in the structure of described figure, each node carries out information flow according to the information flow rule in described information stream cardon;
Initial value determining unit, the initial quantity of information being configured to determine in the structure of described figure each node in described information stream cardon;
Information computing unit, is configured to the initial quantity of information according to described each node, according to described information flow rule, calculates the quantity of information of described each node at default time point;
Information determination unit, is configured to the time series according to the information distribution formed by described each node in the quantity of information of all default time points, it is determined that the structure descriptor of described figure,
Wherein, described figure changes unit, and to be configured to the Structure Conversion of described figure be Markov figure, described Markov figure comprises each node in the structure of described figure and points to the oriented limit of oneself, oriented limit corresponding to line between node, and the weight on each oriented limit obtained based on the original tally distributed computation of line between node and node; The quantity of information that described information flow rule is each node carries out information flow according to the weight on the oriented limit between node and oriented limit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110379668.8A CN103106323B (en) | 2011-11-10 | 2011-11-10 | Obtain method and the device of the structure descriptor of figure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110379668.8A CN103106323B (en) | 2011-11-10 | 2011-11-10 | Obtain method and the device of the structure descriptor of figure |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103106323A CN103106323A (en) | 2013-05-15 |
| CN103106323B true CN103106323B (en) | 2016-06-15 |
Family
ID=48314177
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110379668.8A Expired - Fee Related CN103106323B (en) | 2011-11-10 | 2011-11-10 | Obtain method and the device of the structure descriptor of figure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103106323B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10970350B2 (en) | 2019-08-15 | 2021-04-06 | Advanced New Technologies Co., Ltd. | Method and apparatus for processing user interaction sequence data |
| CN110598847B (en) * | 2019-08-15 | 2020-08-28 | 阿里巴巴集团控股有限公司 | Method and device for processing interactive sequence data |
| CN113792152B (en) * | 2021-08-23 | 2023-07-04 | 南京信息工程大学 | Fusion method of triangular diagram and knowledge graph |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102185836A (en) * | 2011-04-15 | 2011-09-14 | 哈尔滨工业大学 | Standalone electronic document protection system based on information stream model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07191959A (en) * | 1993-11-12 | 1995-07-28 | Shinu Ko | Dynamic simulation system |
-
2011
- 2011-11-10 CN CN201110379668.8A patent/CN103106323B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102185836A (en) * | 2011-04-15 | 2011-09-14 | 哈尔滨工业大学 | Standalone electronic document protection system based on information stream model |
Non-Patent Citations (2)
| Title |
|---|
| Chinese Whispers - an Efficient Graph Clustering Algorithm and its Application to Natural Language Processing Problems;Chris Biemann;《Proceedings of the HLT-NAACL-06 Workshop on Textgraphs-06》;20060630;第73-80页 * |
| 基于依赖图的信息流图构建方法;曾建等;《计算机应用研究》;20090630;第26卷(第6期);第2154-2157、2164页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103106323A (en) | 2013-05-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102184169B (en) | Method, device and equipment used for determining similarity information among character string information | |
| CN101286150B (en) | Method and device for creating updated parameter, method and device for displaying relevant key words | |
| CN101315644B (en) | Part classification method based on developable clustering | |
| CN103984714A (en) | Ontology semantics-based supply and demand matching method for cloud manufacturing service | |
| CN104615687A (en) | Entity fine granularity classifying method and system for knowledge base updating | |
| CN106570080A (en) | Multilevel semantic matching method for cloud manufacturing resource services | |
| CN109146707A (en) | Power consumer analysis method, device and electronic equipment based on big data analysis | |
| CN103106323B (en) | Obtain method and the device of the structure descriptor of figure | |
| CN105183880A (en) | Hash join method and device | |
| CN100557616C (en) | Protein complex recognizing method based on range estimation | |
| CN114817575B (en) | Large-scale electric power affair map processing method based on extended model | |
| CN105224577A (en) | Multi-label text classification method and system | |
| Wang et al. | An Improved Grey Prediction Model Based on Matrix Representations of the Optimized Initial Value. | |
| CN104951505A (en) | Large-scale data clustering method based on graphic calculation technology | |
| CN102023993B (en) | Cluster page ranking equipment and method based on clustering/classification and time | |
| CN104778205A (en) | Heterogeneous information network-based mobile application ordering and clustering method | |
| CN109210268B (en) | Big data processing method based on ultralow-power electromagnetic valve | |
| CN103970610A (en) | Method for monitoring node flow of water supply network | |
| CN116401602A (en) | Event detection method, device, equipment and computer readable medium | |
| Urıas et al. | Sensitive dependence on initial conditions for cellular automata | |
| CN108874974A (en) | Parallelization Topic Tracking method based on frequent term set | |
| CN110162704B (en) | Multi-scale key user extraction method based on multi-factor genetic algorithm | |
| CN106294064A (en) | A kind of device and method that baseboard management controller attribute is set | |
| Bogdan et al. | On the Empirical Bayes approach to the problem of multiple testing | |
| CN104391950A (en) | Method for using hash B + tree structure to detect complex events in manufacturing Internet of Things massive data streams |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160615 Termination date: 20181110 |