CN107436906A - A kind of information detecting method and device - Google Patents
A kind of information detecting method and device Download PDFInfo
- Publication number
- CN107436906A CN107436906A CN201610366309.1A CN201610366309A CN107436906A CN 107436906 A CN107436906 A CN 107436906A CN 201610366309 A CN201610366309 A CN 201610366309A CN 107436906 A CN107436906 A CN 107436906A
- Authority
- CN
- China
- Prior art keywords
- information
- characteristic information
- primitive character
- characteristic
- matching
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Remote Sensing (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of information detecting method and device, after primitive character information in the characteristic information and original image in getting latest image, acquired characteristic information is matched with primitive character information, obtain matching result, when matching result does not meet preparatory condition, judge that information point changes.That is the present invention judges whether information point changes with the characteristic information in image, and for the identification step of text information, the step of original image and latest image obtain characteristic information simplifies.And when using the primitive character information of original image as standard, without training substantial amounts of primitive character information as training sample.And for any one information point in map, the primitive character information of information point is known and unique, the characteristic information so got can directly be matched with unique primitive character information of same information point, number of matches be reduced, so as to improve detection efficiency.
Description
Technical field
The invention belongs to maps processing technical field, more specifically, more particularly to a kind of information detecting method and device.
Background technology
In map field, POI (Point of Interest, information point) refers to the significant point in geographic significance,
Such as shop, bar, gas station, hospital and station etc., and from map field, map is exactly by POI and road group
Into each POI is the most basic key element that map user needs in all types of usage scenarios, such as from a POI to another POI
Route planning, target POI position coordinates and target POI peripheral positions other element retrieval, it can be seen that POI number
Amount and quality have significant impact to the usage experience of map.
But POI corresponding to POI be it is continually changing, such as POI titles change or increase new POI titles, because
This is in order to ensure POI with being actually consistent, it is necessary to the actual conditions of continual monitoring POI positions.Conventional letter at present
Ceasing detection method is：Obtain the image of POI positions；Pass through OCR (Optical Character Recognition, optics
Character recognition) identify word in image, POI titles are compared one by one in the word and database that then will identify that,
When the word identified from POI titles are different in database when, judge that the POI of POI positions changes, then need
New POI of the word that will identify that as POI positions.
Although whether above- mentioned information detection method can be changed with automatic detection POI, above- mentioned information detection
Method needs to train substantial amounts of training sample to cover numerous font styles, needs to perform word when identifying word by OCR
The steps such as symbol cutting, character recognition and space of a whole page processing, execution is cumbersome, and the word identified needs and POI in database
Title carries out the comparison of semantic level one by one, so as to reduce detection efficiency.
The content of the invention
In view of this, it is an object of the invention to provide a kind of information detecting method and device, for improving detection efficiency.
Technical scheme is as follows：
The present invention provides a kind of information detecting method, and methods described includes：
Obtain the characteristic information in the latest image of information point position and obtain the original image of described information point
In primitive character information；
Acquired characteristic information is matched with the primitive character information, obtains matching result；
When the matching result meets preparatory condition, judge that described information point does not change；
When the matching result does not meet preparatory condition, judge that described information point changes.
The present invention also provides a kind of information detector, and described device includes：
Acquiring unit, characteristic information in latest image for obtaining information point position and obtains information point
Primitive character information in original image；
Matching unit, for acquired characteristic information to be matched with primitive character information, obtain matching result；
First identifying unit, for when the matching result meets preparatory condition, judging that described information point does not become
Change；
Second identifying unit, for when the matching result does not meet preparatory condition, judging that described information point becomes
Change.
Compared with prior art, above-mentioned technical proposal provided by the invention has the following advantages that：
As shown from the above technical solution, the characteristic information and original in the latest image for getting information point position
After primitive character information in beginning image, acquired characteristic information is matched with primitive character information, obtains matching knot
Fruit, when matching result does not meet preparatory condition, judge that information point changes.That is technical scheme provided by the invention
It is to judge whether information point changes with the characteristic information in image, it is former for the identification step of text information
The step of beginning image and latest image obtain characteristic information simplifies.And it is used as standard in the primitive character information using original image
When, without training substantial amounts of primitive character information as training sample.And for any one information point in map,
The primitive character information of information point is known and unique, and the characteristic information so got can be with same information point only
One primitive character information is directly matched, and number of matches is reduced for characters matching mode, so as to improve detection
Efficiency.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is the required accompanying drawing used in technology description to be briefly described, it should be apparent that, drawings in the following description are the present invention
Some embodiments, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis
These accompanying drawings obtain other accompanying drawings.
Fig. 1 is a kind of flow chart of information detecting method provided in an embodiment of the present invention；
Fig. 2 is the schematic diagram for the two images that the embodiment of the present invention uses；
Fig. 3 is the flow chart that characteristic information matches in information detecting method provided in an embodiment of the present invention；
Fig. 4 is another flow chart of information detecting method provided in an embodiment of the present invention；
Fig. 5 is the structural representation of information detector provided in an embodiment of the present invention；
Fig. 6 is the structural representation of matching unit in information detector provided in an embodiment of the present invention.
Embodiment
Inventor to multiple POI examine on the spot and found：POI changes in actual applications often bring whole shops
Change particularly on signboard, therefore taken off from this angle, information detecting method and device provided in an embodiment of the present invention
From the detection mode of existing scene word, but whether characteristic information changes in the front and rear two images by matching same POI
To judge whether POI has changed.
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
Part of the embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
Referring to Fig. 1, it illustrates the first flow chart of information detecting method provided in an embodiment of the present invention, can wrap
Include following steps：
101：Obtain the original in the original image of the characteristic information and acquisition POI in the latest image of POI positions
Beginning characteristic information.Wherein latest image is the image of POI positions under current time, and the feature letter recorded in latest image
Breath is the specific features information of POI under current time.
For any one POI, original image is the unique images that POI is stored in database, wherein original graph
Seem by pretreated image, such as the image to getting zooms in and out and removes noise in image etc., so
The pixel of obtained original image reduces, and then reduces the capacity of original image so that occupancy of the original image in database
Space reduction.After original image is obtained, unique primitive character information is extracted from original image, and will unique original spy
Reference breath is stored in database, is corresponded with original image.
Same consideration is to practical application and detection efficiency, after latest image is got, it is also desirable to latest image
The pretreatment such as noise is zoomed in and out and removed, then extracts characteristic information from pretreated latest image.
In embodiments of the present invention, primitive character information and the characteristic information extracted from latest image are same types
Information.Such as above-mentioned primitive character information and the characteristic information that is extracted from latest image are the features in correspondence image
Point, wherein characteristic point are the station location markers of a regional area with certain feature, are called a little, are to be abstracted as one
Position concept, in order to determine the corresponding relation of same location point in two images, and a good characteristic point possess it is following
Characteristic：
Higher multiplicity：Multiplicity refers to the two images of the same target or scene shot at different conditions
In overlapping region, repetition degree that same characteristic point can detect in overlapping region.Wherein multiplicity is higher, illustrates two width figures
The characteristic point quantity that can be detected simultaneously in the overlapping region of picture is more.Conversely, then explanation is due to the different condition of shooting
And the characteristic point for causing much to detect in piece image does not detect in another piece image.
It is unique：Characteristic point should have certain uniqueness, so just can guarantee that characteristic point is easily distinguished and correctly
Detection.
Locality：Characteristic point should be local, can so allow image to exist under different visual conditions a small amount of
Geometric distortion.
More quantity：The quantity of characteristic point should be enough it is more, accordingly even when can also be detected for small target suitable
The characteristic point of amount.
Accuracy：Characteristic point should have very high accuracy in position, yardstick etc..
Efficiency：The efficiency of feature point detection should be sufficiently high, so just can be suitably used in application in real time.
Features described above point is divided into two classes：Narrow sense characteristic point and Feature Points in Broad Sense, wherein narrow sense characteristic point are defined based on position,
And its position has a conventional attribute meaning in itself, for example angle point in latest image and crosspoint can be used as narrow sense feature
Point；And Feature Points in Broad Sense is defined based on region, and it can be an abstract characteristic area, and its attribute is exactly characteristic area tool
Standby attribute.In embodiments of the present invention, can be extracted arbitrarily from latest image and original image with feature based extraction algorithm
A type of characteristic point.
For example CenSurE (Center Surround Extremas, around center extreme value) feature extraction can be based on and calculated
Method is extracted, and CenSurE feature extraction algorithms can extract preferable characteristic point in image at a high speed, and wherein CenSurE characteristic points are
Calculated on each yardstick of image, extreme value is extracted on all yardsticks and position to obtain CenSurE characteristic points.
Or it is also based on SIFT (Scale-Invariant Feature Transform, scale invariant feature
Conversion) feature extraction algorithm extraction, the characteristic point obtained by SIFT feature extraction algorithm is SIFT feature, is DoG spaces
The Local Extremum composition of (Difference of Gaussian, difference of Gaussian), and by building DoG pyramids SIFT
Characteristic point has Scale invariant shape.
102：Acquired characteristic information is matched with primitive character information, obtains matching result.Wherein match
Purpose is：Judge whether acquired characteristic information and primitive character information are consistent, and accordingly, matching result is then used to refer to institute
The characteristic information of acquisition and the consistent degree of primitive character information, determine whether matching result meets preparatory condition with consistent degree.
In embodiments of the present invention, the feasible pattern of matching is：By acquired each characteristic information and each original spy
Reference breath is matched successively, the characteristic information pair matched, and the characteristic information pair based on matching, obtains matching result.
Such as characteristic information and primitive character information are respectively：The feelings of CenSurE characteristic points and original CenSurE characteristic points
Under condition, each CenSurE characteristic points and each original CenSurE characteristic points can be matched successively, be matched
CenSurE characteristic points pair, thus matching result can be obtained based on the CenSurE characteristic points pair of matching.
103：When matching result meets preparatory condition, judge that POI does not change.
104：When matching result does not meet preparatory condition, judge that POI changes.
Wherein, preparatory condition is the condition whether to be changed for judging POI, and such as above-mentioned preparatory condition can be consistent
Degree between 90% to 100%, i.e., have in the characteristic information obtained in above-mentioned latest image 90% to 100% information with it is original
Characteristic information is consistent.
If there are 90% to 100% information and primitive character information one in the characteristic information obtained in above-mentioned latest image
Cause, then show that most partial informations are consistent with primitive character information in the characteristic information of above-mentioned acquisition, now can be determined that
POI does not change；If information and primitive character information one in the characteristic information obtained in above-mentioned latest image less than 90%
Cause, then show part in the characteristic information of above-mentioned acquisition or only very little partial information is consistent with primitive character information, now
It can be determined that POI changes.
In embodiments of the present invention, matching result can be represented with similarity, that is, calculate the characteristic information pair of matching
Similarity, and using similarity as matching result.The similarity is used for the similar of indicative character information and primitive character information
Degree, therefore in actual application, it is that similarity determines that a conduct is preset that can pass through multiple actual POI image authentications
The threshold value of condition, when similarity is more than the threshold value determined, show most characteristic informations in the characteristic information of said extracted
With primitive character information matches, and then judge POI do not change；When similarity is no more than the threshold value determined, show above-mentioned
Part or only very little Partial Feature information and primitive character information matches, and then judge POI in the characteristic information of extraction
Change.
When characteristic information and primitive character information are corresponding characteristic point and primitive character point, above-mentioned similarity can adopt
With any one following calculation：
The percentage of all characteristic points pair, the boundary rectangle of the characteristic point pair matched account for shared by the number of feature points of matching
The methods of ratio of entire area corresponding to latest image, the simple number of feature points using matching are to characterize similarity.Hair
A person of good sense passes through actual POI image authentications, and characterizing similarity using the number of feature points of matching has obtained preferable result.
As shown in Fig. 2 left figure is POI positions H latest image in wherein Fig. 2, right figure is extracted from database
Whether the position H gone out original image, two images are determined by the matching of the characteristic point in latest image and original image
It is similar, if similarity is more than threshold value, illustrate that matching result meets preparatory condition, and then can determine that POI does not become
Change, otherwise determine that POI changes.
In the case where judging that POI changes, directly primitive character information can be replaced with and be obtained from latest image
The characteristic information taken, to update POI corresponding to POI；Or in the case where judging that POI changes, by artificial
Examine to determine whether to update POI, specifically include the POI to change in the display interface of electronic equipment, this
Sample staff is examined on the spot according to POI to the POI positions provided on display interface, certain when manually examining POI
When changing, primitive character information is replaced with to the characteristic information obtained from latest image, to update POI corresponding to POI
Information, accuracy in detection is improved by this automatic detection and the situation of artificial nucleus' reality combination.
It was found from above-mentioned technical proposal, characteristic information and original in the latest image for getting information point position
After primitive character information in beginning image, acquired characteristic information is matched with primitive character information, obtains matching knot
Fruit, when matching result does not meet preparatory condition, judge that information point changes.That is technical scheme provided by the invention
It is to judge whether information point changes with the characteristic information in image, it is former for the identification step of text information
The step of beginning image and latest image obtain characteristic information simplifies.And it is used as standard in the primitive character information using original image
When, without training substantial amounts of primitive character information as training sample.And for any one information point in map,
The primitive character information of information point is known and unique, and the characteristic information so got can be with same information point only
One primitive character information is directly matched, and number of matches is reduced for characters matching mode, so as to improve detection
Efficiency.
Above- mentioned information detection method is applied in multiple POI and carries out actually detected discovery by inventor：The embodiment of the present invention
The accuracy rate of the information detecting method of offer is up to 99.5%, recalls as 65%.Serviced in 2.6GHZ Intel (R) Xeon (R)
On device, run-time efficiency：1s/ figures, EMS memory occupation is extremely low, thus proves, infomation detection side provided in an embodiment of the present invention
Method realizes review process automatic and that efficiently whether checking POI changes, can substantially reduce people for prior art
Power audits cost, improves detection efficiency and the degree of accuracy, and effectively accelerates the renewal of map datum.
In embodiments of the present invention, above-mentioned acquired each characteristic information and each primitive character information are carried out successively
Match somebody with somebody, the process of the characteristic information pair matched is referred to shown in Fig. 3, may comprise steps of：
301：Feature description is carried out to each characteristic information and each primitive character information, obtained and each characteristic information pair
The first descriptor and the second descriptor corresponding with each primitive character information answered.
It is appreciated that：Characteristic matching is centered on characteristic information, the local feature of neighborhood is matched, therefore
First having to be characterized information when carrying out characteristic matching and establish feature description, this feature description is normally referred to as descriptor, its
In a good descriptor should possess following characteristic：
Multiplicity：For same a part of local feature of a target, its descriptor should be consistent as far as possible.
Compactness：One close descriptor will not include the information of redundancy, so that memory space can be saved, and more can be straight
Connect the speed of effect characteristicses matching process.
Efficiency：The characteristic information of piece image may a lot (such as more than 1000, even more more), it is therefore desirable to a height
The descriptor of effect, it is that the dimension of the characteristic vector of this descriptor should be as low as possible.
In embodiments of the present invention, there are the situation of rotation, preferably SIFT descriptors to adapt to parts of images, because
The directive characteristic vector of SIFT descriptor bands, son can specifically be described by the SIFT of SIFT descriptors to believe feature
Breath is described.Wherein SIFT description are a kind of expressions of characteristic information neighborhood Gaussian image gradient statistical result, by right
Image-region piecemeal around characteristic information, calculation block inside gradient histogram, generates unique vector, and this vector is figure
The one kind in corresponding region is abstracted as in, has uniqueness.Certain embodiment of the present invention can also use other kinds of description
Symbol, will not enumerate in this embodiment of the present invention.
302：Based on each first descriptor and each second descriptor, each characteristic information and each primitive character are believed
Breath carries out cross-matched, the initial characteristicses information pair matched.So-called cross-matched is the characteristic information phase in two images
Mutually alternately match, by taking image A and image B as an example, since image A first characteristic information, by image A first feature
Information is matched with image B first characteristic information, if matched, continues to match image A second feature letter
Second characteristic information of breath and image B；If mismatching, matching image A second feature information and image B first spy
Reference breath to whole characteristic informations in image A until match successively；Believe in last feature for having matched image A
Breath, is matched with the characteristic information in image A one by one according still further to aforesaid way since image B first characteristic information.
By above-mentioned matching process twice, result that image A matches with image B is X, the result that image B matches with image A
For Y, then final image A and the result of image B cross-matcheds are X and Y, i.e., each characteristic information in the two results to for
The initial characteristicses information pair matched somebody with somebody.
Accordingly, above-mentioned first descriptor and the second descriptor are SIFT descriptors, and SIFT descriptors use space X YZ
In the case that three-dimensional vector carries out feature description, the matching process of features described above information can be：
Three-dimensional vector between feature based information and primitive character information, calculate characteristic information and primitive character information it
Between distance, if distance between the two in pre-determined distance, judge two characteristic informations for matching initial characteristicses information
It is right；If distance between the two not in pre-determined distance, judges that two characteristic informations are unmatched characteristic information pair.
303：To the initial characteristicses information of matching to carrying out geometry verification, the characteristic information pair matched.Wherein geometry
The purpose of verification is that the characteristic information of geometrical relationship is not met with removal, can such as use homography matrix (homography) to carry out
Verification.
The so-called characteristic information for not meeting geometrical relationship refers to each feature letter of initial characteristicses information centering latest image
Geometrical relationship between breath is different from the geometrical relationship between corresponding primitive character information in original image.Such as initial characteristicses
Geometrical relationship between four characteristic informations A, B, C and D of information centering latest image is：Characteristic information A, B and C are the same as always
On line, and characteristic information D is not on this straight line；For original image, four corresponding with aforementioned four characteristic information
Primitive character information 1,2,3 and 4 all point-blank, then after being verified by geometry, characteristic information D is removed.
, can be by unmatched characteristic information to from the whole characteristic informations got by above-mentioned matching process
Remove, can so as to improve the degree of accuracy of the characteristic information pair of matching, and then when based on the characteristic information of matching to calculating similarity
Further to improve the degree of accuracy of similarity, so as to improve the degree of accuracy of testing result.
Herein it should be noted is that：Method shown in above-mentioned Fig. 1 can use the characteristic information of any one type
Matched, and in order to improve the degree of accuracy of testing result, method shown in above-mentioned Fig. 1 can use the feature of at least two types
Information carries out cascade matching, as shown in figure 4, it illustrates the characteristic information that corresponding types are extracted by two kinds of feature extraction algorithms
Cascade the process of matching, specifically may comprise steps of：
401：Characteristic information is extracted from latest image by fisrt feature extraction algorithm and extracted by fisrt feature
Algorithm extracts primitive character information from original image.
402：It will be matched, obtained with primitive character information by the characteristic information that fisrt feature extraction algorithm obtains
With result.
403：When the matching result that the characteristic information obtained by fisrt feature extraction algorithm obtains with primitive character information
When meeting preparatory condition, judge that POI does not change.
404：When the matching result that the characteristic information obtained by fisrt feature extraction algorithm obtains with primitive character information
When not meeting preparatory condition, characteristic information is extracted from latest image by second feature extraction algorithm and passes through second feature
Extraction algorithm extracts primitive character information from original image.
405：By by the characteristic information that second feature extraction algorithm is got with being obtained by second feature extraction algorithm
To primitive character information matched, obtain matching result.
406：When the matching result that the characteristic information obtained by second feature extraction algorithm obtains with primitive character information
When meeting preparatory condition, judge that POI does not change.
407：When the matching result that the characteristic information obtained by second feature extraction algorithm obtains with primitive character information
When not meeting preparatory condition, judge that POI changes.
Above-mentioned fisrt feature extraction algorithm and second feature extraction algorithm can use any one extraction algorithm, such as
One feature extraction algorithm can use CenSurE feature extraction algorithms, and second feature extraction algorithm can be carried using SIFT feature
Algorithm is taken, is matched by the cascade of both algorithms to detect whether POI changes, relative to using single algorithm
For matching somebody with somebody, the degree of accuracy and the recall rate of detection are improved.
For foregoing each method embodiment, in order to be briefly described, therefore it is all expressed as to a series of combination of actions, but
It is that those skilled in the art should know, the present invention is not limited by described sequence of movement, because according to the present invention, certain
A little steps can use other orders or carry out simultaneously.Secondly, those skilled in the art should also know, be retouched in specification
The embodiment stated belongs to preferred embodiment, necessary to involved action and the module not necessarily present invention.
Corresponding with above method embodiment, the embodiment of the present invention also provides a kind of information detector, its structural representation
Figure is as shown in figure 5, can include：Acquiring unit 11, matching unit 12, the first identifying unit 13 and the second identifying unit 14.
Acquiring unit 11, the characteristic information and acquisition information point in latest image for obtaining information point position
Original image in primitive character information.Wherein latest image is the image of POI positions under current time, and newest figure
The characteristic information recorded as in is the specific features information of POI under current time.
For any one POI, original image is the unique images that POI is stored in database, wherein original graph
Seem by pretreated image, such as the image to getting zooms in and out and removes noise in image etc., so
The pixel of obtained original image reduces, and then reduces the capacity of original image so that occupancy of the original image in database
Space reduction.After original image is obtained, unique primitive character information is extracted from original image, and will unique original spy
Reference breath is stored in database, is corresponded with original image.
Same consideration is to practical application and detection efficiency, after latest image is got, it is also desirable to latest image
The pretreatment such as noise is zoomed in and out and removed, then extracts characteristic information from pretreated latest image.In the present invention
In embodiment, primitive character information and the characteristic information extracted from latest image are the information of same type.
Such as above-mentioned primitive character information and the characteristic information that is extracted from latest image are the features in correspondence image
Point, the characteristic point are divided into two classes：Narrow sense characteristic point and Feature Points in Broad Sense, wherein narrow sense characteristic point are defined based on position, and its
Position has a conventional attribute meaning in itself, for example angle point in latest image and crosspoint can be used as narrow sense characteristic point；And
Feature Points in Broad Sense is defined based on region, and it can be an abstract characteristic area, and its attribute is exactly that characteristic area possesses
Attribute.In embodiments of the present invention, any one can be extracted from latest image and original image with feature based extraction algorithm
The characteristic point of type.
CenSurE feature extraction algorithms can be such as based on to extract, preferable characteristic point in image can be extracted at a high speed, wherein
CenSurE characteristic points are calculated on each yardstick of image, and extreme value is extracted on all yardsticks and position to obtain
CenSurE characteristic points.
Or SIFT feature extraction algorithm can be based on and extracted, the characteristic point that the method obtains is SIFT feature, is
The Local Extremum composition in DoG spaces, and there is Scale invariant shape by building DoG pyramid SIFT features.
Matching unit 12, for acquired characteristic information to be matched with primitive character information, obtain matching result.
The purpose wherein matched is：Judge whether acquired characteristic information and primitive character information are consistent, and accordingly, matching result is then
It is used to refer to acquired characteristic information and the consistent degree of primitive character information, determines whether matching result meets with consistent degree
Preparatory condition.
In embodiments of the present invention, the feasible matching way of matching unit 12 is：By acquired each characteristic information with
Each primitive character information is matched successively, the characteristic information pair matched, and the characteristic information pair based on matching, is obtained
Matching result.
Such as characteristic information and primitive character information are respectively：The feelings of CenSurE characteristic points and original CenSurE characteristic points
Under condition, each CenSurE characteristic points and each original CenSurE characteristic points can be matched successively, be matched
CenSurE characteristic points pair, thus matching result can be obtained based on the CenSurE characteristic points pair of matching.
First identifying unit 13, for when matching result meets preparatory condition, judging that information point does not change.
Second identifying unit 14, for when matching result does not meet preparatory condition, judging that information point changes.
Wherein, preparatory condition is the condition whether to be changed for judging POI, and such as above-mentioned preparatory condition can be consistent
Degree between 90% to 100%, i.e., have in the characteristic information obtained in above-mentioned latest image 90% to 100% information with it is original
Characteristic information is consistent.
If there are 90% to 100% information and primitive character information one in the characteristic information obtained in above-mentioned latest image
Cause, then show that most partial informations are consistent with primitive character information in the characteristic information of above-mentioned acquisition, now first judges
Unit 13 can be determined that POI does not change；If in the characteristic information obtained in above-mentioned latest image less than 90% information with
Primitive character information is consistent, then shows in the characteristic information of above-mentioned acquisition part or only very little partial information and primitive character
Information is consistent, and now the second identifying unit 14 can be determined that POI changes.
In embodiments of the present invention, matching result can be represented with similarity, and the similarity is believed for indicative character
The similarity degree of breath and primitive character information, therefore in actual application, can pass through multiple actual POI image authentications is
Similarity determines a threshold value as preparatory condition, when similarity is more than the threshold value determined, shows the feature of said extracted
Most characteristic informations and primitive character information matches in information, and then the first identifying unit 13 judges that POI does not change；
When similarity is no more than the threshold value determined, show in the characteristic information of said extracted part or only very little Partial Feature information
With primitive character information matches, and then the second identifying unit 14 judge POI change.
When characteristic information and primitive character information are corresponding characteristic point and primitive character point, above-mentioned similarity can adopt
With any one following calculation：
The percentage of all characteristic points pair, the boundary rectangle of the characteristic point pair matched account for shared by the number of feature points of matching
The methods of ratio of entire area corresponding to latest image, the simple number of feature points using matching are to characterize similarity.Hair
A person of good sense passes through actual POI image authentications, and characterizing similarity using the number of feature points of matching has obtained preferable result.
It was found from above-mentioned technical proposal, characteristic information and original in the latest image for getting information point position
After primitive character information in beginning image, acquired characteristic information is matched with primitive character information, obtains matching knot
Fruit, when matching result does not meet preparatory condition, judge that information point changes.That is technical scheme provided by the invention
It is to judge whether information point changes with the characteristic information in image, it is former for the identification step of text information
The step of beginning image and latest image obtain characteristic information simplifies.And it is used as standard in the primitive character information using original image
When, without training substantial amounts of primitive character information as training sample.And for any one information point in map,
The primitive character information of information point is known and unique, and the characteristic information so got can be with same information point only
One primitive character information is directly matched, and number of matches is reduced for characters matching mode, so as to improve detection
Efficiency.
Above- mentioned information detection means is applied in multiple POI and carries out actually detected discovery by inventor：The embodiment of the present invention
The accuracy rate of the information detecting method of offer is up to 99.5%, recalls as 65%.Serviced in 2.6GHZ Intel (R) Xeon (R)
On device, run-time efficiency：1s/ figures, EMS memory occupation is extremely low, thus proves, infomation detection dress provided in an embodiment of the present invention
Put for prior art, realize review process automatic and that efficiently whether checking POI changes, people can be substantially reduced
Power audits cost, improves detection efficiency and the degree of accuracy, and effectively accelerates the renewal of map datum.
Referring to Fig. 6, it illustrates the structural representation of matching unit in information detector provided in an embodiment of the present invention
Figure, can include：Obtain subelement 121, coupling subelement 122, verification subelement 123 and computation subunit 124.
Obtain subelement 121, for carrying out feature description to each characteristic information and each primitive character information, obtain with
First descriptor corresponding to each characteristic information and second feature corresponding with each primitive character information symbol.It is appreciated that
It is：Characteristic matching is centered on characteristic information, and the local feature of neighborhood is matched, therefore first when carrying out characteristic matching
First to be characterized information and establish feature description, this feature description is normally referred to as descriptor.
In embodiments of the present invention, there are the situation of rotation, preferably SIFT descriptors to adapt to parts of images, because
The directive characteristic vector of SIFT descriptor bands, son can specifically be described by the SIFT of SIFT descriptors to believe feature
Breath is described.Wherein SIFT description are a kind of expressions of characteristic information neighborhood Gaussian image gradient statistical result, by right
Image-region piecemeal around characteristic information, calculation block inside gradient histogram, generates unique vector, and this vector is figure
The one kind in corresponding region is abstracted as in, has uniqueness.Certain embodiment of the present invention can also use other kinds of description
Symbol, will not enumerate in this embodiment of the present invention.
Coupling subelement 122, for based on each first descriptor and each second descriptor, to each characteristic information and
Each primitive character information carries out cross-matched, the initial characteristicses information pair matched.So-called cross-matched is two images
In characteristic information alternate matching, i.e., on the basis of the characteristic information of latest image, by each feature in latest image
Information is matched with each primitive character information in original image；Again using each primitive character information in original image as
Benchmark, each primitive character information in original image is matched with each characteristic information in latest image.
Such as by above-mentioned matching process twice, the result that image A matches with image B is X, and image B matches with image A
As a result it is Y, then final image A and the result of image B cross-matcheds are X and Y, i.e. each characteristic information pair in the two results
For the initial characteristicses information pair of matching.
Subelement 123 is verified, for the initial characteristicses information to matching to carrying out geometry verification, the feature letter matched
Breath pair.The purpose of wherein geometry verification is that the characteristic information of geometrical relationship is not met with removal, can such as use homography matrix
(homography) verified.
The so-called characteristic information for not meeting geometrical relationship refers to each feature letter of initial characteristicses information centering latest image
Geometrical relationship between breath is different from the geometrical relationship between corresponding primitive character information in original image.Such as initial characteristicses
Geometrical relationship between four characteristic informations A, B, C and D of information centering latest image is：Characteristic information A, B and C are the same as always
On line, and characteristic information D is not on this straight line；For original image, four corresponding with aforementioned four characteristic information
Primitive character information 1,2,3 and 4 all point-blank, then after being verified by geometry, characteristic information D is removed, to carry
The degree of accuracy of the characteristic information pair of height matching.
Computation subunit 124, for the characteristic information pair based on matching, obtain matching result.Wherein computation subunit can
To calculate the similarity of the characteristic information pair of matching, using similarity as matching result.
Passing through above-mentioned matching process, can be by unmatched characteristic information to from the whole characteristic informations got
Remove, can so as to improve the degree of accuracy of the characteristic information pair of matching, and then when based on the characteristic information of matching to calculating similarity
Further to improve the degree of accuracy of similarity, so as to improve the degree of accuracy of testing result.
In addition, above-mentioned acquiring unit can use the characteristic information of any one type to be matched, and in order to improve inspection
The degree of accuracy of result is surveyed, above- mentioned information detection means can use cascade matching way, specifically：Acquiring unit 11 passes through first
Feature extraction algorithm extracts characteristic information from latest image and extracted by fisrt feature extraction algorithm from original image
Primitive character information, when the first identifying unit 13 judges that matching result meets preparatory condition, triggering acquiring unit 11 is led to again
Cross second feature extraction algorithm and characteristic information is extracted from latest image and by second feature extraction algorithm from original image
Middle extraction primitive character information, so matching unit 12 can to the characteristic information that is got by second feature extraction algorithm and
Primitive character information is matched, and determines whether POI occurs again to trigger the first identifying unit 13 and the second identifying unit 14
Change.
Above-mentioned fisrt feature extraction algorithm and second feature extraction algorithm can use any one extraction algorithm, such as
One feature extraction algorithm can use CenSurE feature extraction algorithms, and second feature extraction algorithm can be carried using SIFT feature
Algorithm is taken, is matched by the cascade of both algorithms to detect whether POI changes, relative to using single algorithm
For matching somebody with somebody, the degree of accuracy and the recall rate of detection are improved.
It should be noted that each embodiment in this specification is described by the way of progressive, each embodiment weight
Point explanation is all difference with other embodiment, between each embodiment identical similar part mutually referring to.
For device class embodiment, because it is substantially similar to embodiment of the method, so description is fairly simple, related part is joined
See the part explanation of embodiment of the method.
Finally, it is to be noted that, herein, such as first and second or the like relational terms be used merely to by
One entity or operation make a distinction with another entity or operation, and not necessarily require or imply these entities or operation
Between any this actual relation or order be present.Moreover, term " comprising ", "comprising" or its any other variant meaning
Covering including for nonexcludability, so that process, method, article or equipment including a series of elements not only include that
A little key elements, but also the other element including being not expressly set out, or also include for this process, method, article or
The intrinsic key element of equipment.In the absence of more restrictions, the key element limited by sentence "including a ...", is not arranged
Except other identical element in the process including the key element, method, article or equipment being also present.
The foregoing description of the disclosed embodiments, those skilled in the art are enable to realize or using the present invention.To this
A variety of modifications of a little embodiments will be apparent for a person skilled in the art, and generic principles defined herein can
Without departing from the spirit or scope of the present invention, to realize in other embodiments.Therefore, the present invention will not be limited
The embodiments shown herein is formed on, and is to fit to consistent with principles disclosed herein and features of novelty most wide
Scope.
Claims (10)
1. a kind of information detecting method, it is characterised in that methods described includes：
In the original image for obtaining the characteristic information and acquisition described information point in the latest image of information point position
Primitive character information；
Acquired characteristic information is matched with the primitive character information, obtains matching result；
When the matching result meets preparatory condition, judge that described information point does not change；
When the matching result does not meet preparatory condition, judge that described information point changes.
2. according to the method for claim 1, it is characterised in that described by acquired characteristic information and the primitive character
Information is matched, and obtains matching result, including：
Acquired each characteristic information is matched successively with each primitive character information, the feature letter matched
Breath pair；
Characteristic information pair based on matching, obtain the matching result.
3. according to the method for claim 2, it is characterised in that it is described by acquired each characteristic information with it is each described
Primitive character information is matched successively, the characteristic information pair matched, including：
Feature description is carried out to each characteristic information and each primitive character information, obtains believing with each feature
First descriptor corresponding to breath and the second descriptor corresponding with each primitive character information；
Based on each first descriptor and each second descriptor, to each characteristic information and each original
Beginning characteristic information carries out cross-matched, the initial characteristicses information pair matched；
To the initial characteristicses information of the matching to carrying out geometry verification, the characteristic information pair matched.
4. according to the method for claim 2, it is characterised in that the characteristic information pair based on matching, obtain described
With result, including：The similarity of the characteristic information pair of the matching is calculated, and using the similarity as the matching result.
5. according to the method for claim 1, it is characterised in that in the latest image for obtaining information point position
Primitive character information in the original image of characteristic information and acquisition described information point, including：
Characteristic information is extracted from the latest image by fisrt feature extraction algorithm and extracted by the fisrt feature
Algorithm extracts primitive character information from the original image；
When the matching result does not meet preparatory condition, extraction characteristic information and the institute from the latest image are returned
The extraction primitive character information from the original image is stated, to be extracted by second feature extraction algorithm from the latest image
Characteristic information and primitive character information is extracted from the original image by the second feature extraction algorithm.
6. a kind of information detector, it is characterised in that described device includes：
Acquiring unit, characteristic information in latest image for obtaining information point position and obtains the original of information point
Primitive character information in image；
Matching unit, for acquired characteristic information to be matched with primitive character information, obtain matching result；
First identifying unit, for when the matching result meets preparatory condition, judging that described information point does not change；
Second identifying unit, for when the matching result does not meet preparatory condition, judging that described information point changes.
7. device according to claim 6, it is characterised in that the matching unit, for by acquired each feature
Information is matched successively with each primitive character information, the characteristic information pair matched, and the feature based on matching
Information pair, obtain the matching result.
8. device according to claim 7, it is characterised in that the matching unit includes：
Subelement is obtained, for carrying out feature description to each characteristic information and each primitive character information, is obtained
The first descriptor corresponding with each characteristic information and second feature corresponding with each primitive character information accord with；
Coupling subelement, for based on each first descriptor and each second descriptor, to each feature
Information and each primitive character information carry out cross-matched, the initial characteristicses information pair matched；
Subelement is verified, for the initial characteristicses information to the matching to carrying out geometry verification, the characteristic information matched
It is right；
Computation subunit, for the characteristic information pair based on matching, obtain the matching result.
9. device according to claim 7, it is characterised in that the matching result is the characteristic information pair of the matching
Similarity.
10. device according to claim 6, it is characterised in that the acquiring unit, calculated for being extracted by fisrt feature
Method is extracted characteristic information from the latest image and carried by the fisrt feature extraction algorithm from the original image
Take primitive character information；
Second identifying unit, for when the matching result does not meet preparatory condition, triggering the acquiring unit and passing through
Second feature extraction algorithm extracts characteristic information and by the second feature extraction algorithm from institute from the latest image
State extraction primitive character information in original image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610366309.1A CN107436906A (en) | 2016-05-27 | 2016-05-27 | A kind of information detecting method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610366309.1A CN107436906A (en) | 2016-05-27 | 2016-05-27 | A kind of information detecting method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107436906A true CN107436906A (en) | 2017-12-05 |
Family
ID=60454580
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610366309.1A Pending CN107436906A (en) | 2016-05-27 | 2016-05-27 | A kind of information detecting method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107436906A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109726255A (en) * | 2018-12-18 | 2019-05-07 | 斑马网络技术有限公司 | Automatic update method, device, system and the storage medium of POI |
| CN109977191A (en) * | 2019-04-01 | 2019-07-05 | 国家基础地理信息中心 | Problem map detection method, device, electronic equipment and medium |
| CN110084254A (en) * | 2018-01-23 | 2019-08-02 | 北京国双科技有限公司 | Method and device is determined based on the similar image of social networks |
| CN110246164A (en) * | 2019-05-30 | 2019-09-17 | 中国科学院长春光学精密机械与物理研究所 | Visible images and SAR image registration method and system |
| CN110390279A (en) * | 2019-07-08 | 2019-10-29 | 丰图科技(深圳)有限公司 | Coordinate recognition method, device, equipment and computer readable storage medium |
| CN110609879A (en) * | 2018-06-14 | 2019-12-24 | 百度在线网络技术（北京）有限公司 | Interest point duplicate determination method and device, computer equipment and storage medium |
| CN112132142A (en) * | 2020-09-27 | 2020-12-25 | 平安医疗健康管理股份有限公司 | Text region determination method, text region determination device, computer equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101275854A (en) * | 2007-03-26 | 2008-10-01 | 日电（中国）有限公司 | Method and equipment for updating map data |
| CN101794316A (en) * | 2010-03-30 | 2010-08-04 | 高翔 | Real-scene status consulting system and coordinate offset method based on GPS location and direction identification |
| CN102163288A (en) * | 2011-04-06 | 2011-08-24 | 北京中星微电子有限公司 | Eyeglass detection method and device |
| CN103914546A (en) * | 2014-04-09 | 2014-07-09 | 百度在线网络技术（北京）有限公司 | Data updating method and device thereof |
| CN104751475A (en) * | 2015-04-16 | 2015-07-01 | 中国科学院软件研究所 | Feature point optimization matching method for static image object recognition |
| CN105373610A (en) * | 2015-11-17 | 2016-03-02 | 广东欧珀移动通信有限公司 | Indoor map updating method and server |
-
2016
- 2016-05-27 CN CN201610366309.1A patent/CN107436906A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101275854A (en) * | 2007-03-26 | 2008-10-01 | 日电（中国）有限公司 | Method and equipment for updating map data |
| CN101794316A (en) * | 2010-03-30 | 2010-08-04 | 高翔 | Real-scene status consulting system and coordinate offset method based on GPS location and direction identification |
| CN102163288A (en) * | 2011-04-06 | 2011-08-24 | 北京中星微电子有限公司 | Eyeglass detection method and device |
| CN103914546A (en) * | 2014-04-09 | 2014-07-09 | 百度在线网络技术（北京）有限公司 | Data updating method and device thereof |
| CN104751475A (en) * | 2015-04-16 | 2015-07-01 | 中国科学院软件研究所 | Feature point optimization matching method for static image object recognition |
| CN105373610A (en) * | 2015-11-17 | 2016-03-02 | 广东欧珀移动通信有限公司 | Indoor map updating method and server |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110084254A (en) * | 2018-01-23 | 2019-08-02 | 北京国双科技有限公司 | Method and device is determined based on the similar image of social networks |
| CN110609879A (en) * | 2018-06-14 | 2019-12-24 | 百度在线网络技术（北京）有限公司 | Interest point duplicate determination method and device, computer equipment and storage medium |
| CN109726255A (en) * | 2018-12-18 | 2019-05-07 | 斑马网络技术有限公司 | Automatic update method, device, system and the storage medium of POI |
| CN109977191A (en) * | 2019-04-01 | 2019-07-05 | 国家基础地理信息中心 | Problem map detection method, device, electronic equipment and medium |
| CN109977191B (en) * | 2019-04-01 | 2021-04-30 | 国家基础地理信息中心 | Problem map detection method, device, electronic equipment and medium |
| CN110246164A (en) * | 2019-05-30 | 2019-09-17 | 中国科学院长春光学精密机械与物理研究所 | Visible images and SAR image registration method and system |
| CN110390279A (en) * | 2019-07-08 | 2019-10-29 | 丰图科技(深圳)有限公司 | Coordinate recognition method, device, equipment and computer readable storage medium |
| CN112132142A (en) * | 2020-09-27 | 2020-12-25 | 平安医疗健康管理股份有限公司 | Text region determination method, text region determination device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107436906A (en) | A kind of information detecting method and device | |
| CN110427905B (en) | Pedestrian tracking method, device and terminal | |
| CN107729818B (en) | Multi-feature fusion vehicle re-identification method based on deep learning | |
| CN109657631B (en) | Human body posture recognition method and device | |
| JP4970195B2 (en) | Person tracking system, person tracking apparatus, and person tracking program | |
| Piccioli et al. | Robust method for road sign detection and recognition | |
| CN111695486B (en) | High-precision direction signboard target extraction method based on point cloud | |
| EP1199648A1 (en) | Image shape descriptor extraction and searching | |
| CN110008909B (en) | Real-name system business real-time auditing system based on AI | |
| CN110969166A (en) | Small target identification method and system in inspection scene | |
| CN108319952B (en) | Vehicle feature extraction method and device | |
| TWI504858B (en) | A vehicle specification measuring and processing device, a vehicle specification measuring method, and a recording medium | |
| CN112257660B (en) | Method, system, equipment and computer readable storage medium for removing invalid passenger flow | |
| CN109919060A (en) | A kind of identity card content identifying system and method based on characteristic matching | |
| CN111435421B (en) | Traffic-target-oriented vehicle re-identification method and device | |
| CN102930251B (en) | Bidimensional collectibles data acquisition and the apparatus and method of examination | |
| CN104636730A (en) | Method and device for face verification | |
| CN108960115A (en) | Multi-direction Method for text detection based on angle point | |
| CN112541434B (en) | Face recognition method based on central point tracking model | |
| CN111400533A (en) | Image screening method and device, electronic equipment and storage medium | |
| CN109740609A (en) | A kind of gauge detection method and device | |
| CN108986137A (en) | Human body tracing method, device and equipment | |
| CN103488966A (en) | Intelligent mobile phone capable of identifying real-name ticket information | |
| CN114548214A (en) | Target tracking method, computer equipment and storage device | |
| CN114494373A (en) | High-precision rail alignment method and system based on target detection and image registration |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200507 Address after: 310052 room 508, floor 5, building 4, No. 699, Wangshang Road, Changhe street, Binjiang District, Hangzhou City, Zhejiang Province Applicant after: Alibaba (China) Co.,Ltd. Address before: 100080 Beijing City, Haidian District Suzhou Street No. 3 floor 16 room 2 Applicant before: AUTONAVI INFORMATION TECHNOLOGY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171205 |
|
| RJ01 | Rejection of invention patent application after publication |