CN105205494B - Similar pictures recognition methods and device - Google Patents
Similar pictures recognition methods and device Download PDFInfo
- Publication number
- CN105205494B CN105205494B CN201510549204.5A CN201510549204A CN105205494B CN 105205494 B CN105205494 B CN 105205494B CN 201510549204 A CN201510549204 A CN 201510549204A CN 105205494 B CN105205494 B CN 105205494B
- Authority
- CN
- China
- Prior art keywords
- feature point
- pair
- picture
- point pair
- feature
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/757—Matching configurations of points or features
Abstract
Present disclose provides a kind of similar pictures recognition methods and devices, belong to picture recognition field.Method includes: to carry out feature point extraction to two picture to be identified, obtains multiple characteristic points of two pictures;According to multiple characteristic points of two pictures, multiple fisrt feature points pair of two pictures are obtained;Determine multiple second feature points pair;When the quantity that the quantity ratio between multiple second feature points pair and multiple fisrt feature point pair is greater than the first preset threshold and multiple second feature point pair is less than amount threshold, according to the line angle of multiple second feature points pair, to multiple second feature points to screening, multiple third feature points pair are obtained;When the quantity ratio between multiple third feature points pair and multiple second feature point pair is greater than the second preset threshold, determine that two pictures are similar pictures.The disclosure avoids the identification higher situation of error rate in identification process of the less characteristic point to scene, improves the speed and accuracy of similar pictures identification.
Description
Technical field
This disclosure relates to picture recognition field more particularly to a kind of similar pictures recognition methods and device.
Background technique
With the development of information technology, similar pictures search is a big hot spot of current search area research, similar pictures
The premise of search is the identification of similar pictures.Identification is carried out to similar pictures based on characteristics algorithm and is increasingly becoming similar pictures
The mainstream research direction of recognition methods, for example, SIFT (Scale Invariant Feature Transform, not using scale
Become feature) or SURF (Speeded Up Robust Features accelerates robust feature) etc..To be based on SIFT algorithm to similar
For picture is identified, this method comprises: extracting the SIFT point of every picture;Using the Euclidean distance of the SIFT point as
The similarity determination of two pictures is measured.For example, taking some SIFT point Dot11 in picture 1, and finds out in picture 2 and be somebody's turn to do
The nearest SIFT point Dot21 of the SIFT point Euclidean distance and secondary close SIFT point Dot22 of Euclidean distance, in the two SIFT points,
If nearest distance with it is secondary it is close at a distance between ratio be less than preset threshold, it is determined that in picture 1 in Dot11 and picture 2
This pair of SIFT point of Dot21 be initial SIFT point pair, further, in order to filter out correct matched initial SIFT point pair,
Again by nearest distance with it is secondary it is close at a distance between ratio be compared with default ratio, when ratio is less than default ratio,
Then think the initial SIFT point to being correctly to match;When ratio between correct matched SIFT point pair and all initial SIFT points pair
When greater than preset reduced value, it is determined that this two picture is similar pictures.Similar pictures are known based on characteristics algorithm
During other, computation complexity will be improved, and similar pictures identify that real-time is poor.
In order to reduce the complexity to similar pictures identification, and identification real-time is improved, currently, people use RANSAC
(Random Sample Consensus, random sampling are consistent) algorithm is to the characteristic point obtained based on characteristics algorithm to sieving
Choosing.The screening process includes: from the correct matched characteristic point by being obtained based on characteristics algorithm in set, random selection one
A sample is based on the sample initialization model, obtains RANSAC sample to establish sample initialization model.Then, being based on should
RANSAC sample screens this feature point to set, obtains and the matched characteristic point pair of sample initialization model.But
Characteristic point to it is less when, as characteristic point to it is only several to or more than ten clock synchronizations, using RANSAC algorithm to this feature point into
During row screening, the characteristic point pair of erroneous matching is likely to be to the sample of random selection in set from this feature point, and
The probability of the mistake is larger, therefore, increases by RANSAC algorithm to error rate of this feature point to screening, leads to similar pictures
Recognition accuracy reduces.
Summary of the invention
To overcome the problems in correlation technique, the disclosure provides a kind of similar pictures recognition methods and device.
According to the first aspect of the embodiments of the present disclosure, a kind of similar pictures recognition methods is provided, comprising:
Feature point extraction is carried out to two picture to be identified, obtains multiple characteristic points of two picture;
According to multiple characteristic points of two picture, multiple fisrt feature points pair of two picture are obtained;
From multiple fisrt feature point centering, multiple second feature points pair are determined, which matches to be correct
Characteristic point pair;
It is preset when the quantity ratio between multiple second feature point pair and multiple fisrt feature point pair is greater than first
Threshold value and the quantity of multiple second feature point pair are less than amount threshold, according to the line angle of multiple second feature point pair,
To multiple second feature point to screening, multiple third feature points pair are obtained;
When the quantity ratio between multiple third feature point pair and multiple second feature point pair is greater than the second default threshold
When value, determine that two picture is similar pictures.
In the first possible implementation of first aspect, feature point extraction is carried out to two picture to be identified, is obtained
Multiple characteristic points to two picture include:
Using scale invariant feature SIFT or accelerate robust feature SURF, feature is carried out to two similar pictures to be identified
Point extracts, and obtains multiple characteristic points of two picture.
In second of possible implementation of first aspect, from multiple fisrt feature point centering, multiple second are determined
Characteristic point pair, the second feature point is to for correct matched characteristic point pair, this method further include:
It is preset when the quantity ratio between multiple second feature point pair and multiple fisrt feature point pair is greater than first
Threshold value and the quantity of multiple second feature point pair are not less than amount threshold, using stochastical sampling consistency RANSAC algorithm pair
Multiple second feature point is to screening.
It is right according to the line angle of multiple second feature point pair in the third possible implementation of first aspect
Multiple second feature point to carry out screening include:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line determines that the angular interval that each first angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line determines that the angular interval that each second angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number.
It is right according to the line angle of multiple second feature point pair in the 4th kind of possible implementation of first aspect
Multiple second feature point to carry out screening include:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line retains the second feature point pair that the first angle is in predetermined angle range;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line retains the second feature point pair that the second angle is in predetermined angle range.
According to the second aspect of an embodiment of the present disclosure, a kind of similar pictures identification device is provided, comprising:
Feature point extraction module obtains two picture for carrying out feature point extraction to two picture to be identified
Multiple characteristic points;
Fisrt feature point obtains two picture for multiple characteristic points according to two picture to acquisition module
Multiple fisrt feature points pair;
Second feature point is to determining module, for determining multiple second feature points pair from multiple fisrt feature point centering,
The second feature point is to for correct matched characteristic point pair;
Third feature point to screening module, for when multiple second feature point pair and multiple fisrt feature point pair it
Between quantity ratio be greater than the quantity of the first preset threshold and multiple second feature point pair and be less than amount threshold, it is more according to this
The line angle of a second feature point pair obtains multiple third feature points pair to multiple second feature point to screening;
Similar pictures determining module, for when the number between multiple third feature point pair and multiple second feature point pair
When measuring ratio greater than the second preset threshold, determine that two picture is similar pictures.
In the first possible implementation of second aspect, this feature point extraction module is used to use scale invariant feature
SIFT accelerates robust feature SURF, carries out feature point extraction to two similar pictures to be identified, obtains two picture
Multiple characteristic points.
May be in implementation at second of second aspect, which is used for when multiple the determining module
Quantity ratio between two characteristic points pair and multiple fisrt feature point pair is greater than the first preset threshold and multiple second spy
Sign point pair quantity be not less than amount threshold, using stochastical sampling consistency RANSAC algorithm to multiple second feature point into
Row screening.
In the third possible implementation of second aspect, which is used for screening module:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line determines that the angular interval that each first angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line determines that the angular interval that each second angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number.
In the 4th kind of possible implementation of second aspect, which is used for screening module:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line retains the second feature point pair that the first angle is in predetermined angle range;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line retains the second feature point pair that the second angle is in predetermined angle range.
According to the third aspect of an embodiment of the present disclosure, a kind of similar pictures identification device is provided, comprising:
Processor;
The instruction that can be performed for storage processor;
Wherein, which is configured as:
Feature point extraction is carried out to two picture to be identified, obtains multiple characteristic points of two picture;
According to multiple characteristic points of two picture, multiple fisrt feature points pair of two picture are obtained;
From multiple fisrt feature point centering, multiple second feature points pair are determined, which matches to be correct
Characteristic point pair;
It is preset when the quantity ratio between multiple second feature point pair and multiple fisrt feature point pair is greater than first
Threshold value and the quantity of multiple second feature point pair are less than amount threshold, according to the line angle of multiple second feature point pair,
To multiple second feature point to screening, multiple third feature points pair are obtained;
When the quantity ratio between multiple third feature point pair and multiple second feature point pair is greater than the second default threshold
When value, determine that two picture is similar pictures.
The technical scheme provided by this disclosed embodiment can include the following benefits:
The method and device that the embodiment of the present disclosure provides, determined for the first time correct matched second feature point to later,
It is special to multiple second according to the line angle of multiple second feature points pair under the scene of the negligible amounts of second feature point pair
Sign point retains the second feature point pair being located in the overlapping region of similar pictures, screens out and be not located at weight to being screened again
The second feature point pair in folded region, then based on remaining third feature point pair and multiple second feature point pair and multiple second
Quantity ratio between characteristic point pair, carrys out the identification of further progress similar pictures, so that the knowledge in less characteristic point to scene
The identification higher situation of error rate is avoided during not, improves the speed and accuracy of similar pictures identification.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
The disclosure can be limited.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the disclosure
Example, and together with specification for explaining the principles of this disclosure.
Fig. 1 is a kind of flow chart of similar pictures recognition methods shown according to an exemplary embodiment.
Fig. 2 is a kind of flow chart of similar pictures recognition methods shown according to an exemplary embodiment.
Fig. 3 is a kind of similar pictures identification device block diagram shown according to an exemplary embodiment.
Fig. 4 is a kind of block diagram of similar pictures identification device 400 shown according to an exemplary embodiment.
Fig. 5 is a kind of block diagram of similar pictures identification device 500 shown according to an exemplary embodiment.
Specific embodiment
To keep the purposes, technical schemes and advantages of the disclosure clearer, below in conjunction with attached drawing to disclosure embodiment party
Formula is described in further detail.
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all implementations consistent with this disclosure.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects be described in detail in claims, the disclosure.
Fig. 1 is a kind of flow chart of similar pictures recognition methods shown according to an exemplary embodiment, as shown in Figure 1,
Similar pictures recognition methods is for including the following steps in terminal.
In a step 101, feature point extraction is carried out to two picture to be identified, obtains multiple features of two picture
Point.
In a step 102, according to multiple characteristic points of two picture, multiple fisrt feature points of two picture are obtained
It is right.
In step 103, from multiple fisrt feature point centering, multiple second feature points pair are determined, the second feature point
To for correct matched characteristic point pair.
At step 104, when the quantity ratio between multiple second feature point pair and multiple fisrt feature point pair
Greater than the first preset threshold and the quantity of multiple second feature point pair is less than amount threshold, according to multiple second feature point
Pair line angle obtain multiple third feature points pair to multiple second feature point to screening.
In step 105, when the quantity ratio between multiple third feature point pair and multiple second feature point pair is big
When the second preset threshold, determine that two picture is similar pictures.
The method that the embodiment of the present disclosure provides is determining correct matched second feature point to later, second for the first time
Under the scene of the negligible amounts of characteristic point pair, according to the line angle of multiple second feature points pair, to multiple second feature points pair
It is screened again, retains the second feature point pair being located in the overlapping region of similar pictures, screen out and be not located at overlapping region
Second feature point pair, then based on remaining third feature point pair and multiple second feature point pair and multiple second feature point
Quantity ratio between, carrys out the identification of further progress similar pictures, so that in less characteristic point to the identification process of scene
In avoid identification the higher situation of error rate, improve similar pictures identification speed and accuracy.
In the first possible implementation, feature point extraction is carried out to two picture to be identified, obtains this two figures
Multiple characteristic points of piece include:
Using scale invariant feature SIFT or accelerate robust feature SURF, feature is carried out to two similar pictures to be identified
Point extracts, and obtains multiple characteristic points of two picture.
In second of possible implementation, from multiple fisrt feature point centering, multiple second feature points pair are determined, it should
Second feature point is to for correct matched characteristic point pair, this method further include:
It is preset when the quantity ratio between multiple second feature point pair and multiple fisrt feature point pair is greater than first
Threshold value and the quantity of multiple second feature point pair are not less than amount threshold, using stochastical sampling consistency RANSAC algorithm pair
Multiple second feature point is to screening.
In the third possible implementation, according to the line angle of multiple second feature point pair, to multiple second
Characteristic point to carry out screening include:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line determines that the angular interval that each first angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line determines that the angular interval that each second angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number.
In the 4th kind of possible implementation, according to the line angle of multiple second feature point pair, to multiple second
Characteristic point to carry out screening include:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line retains the second feature point pair that the first angle is in predetermined angle range;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line retains the second feature point pair that the second angle is in predetermined angle range.
All the above alternatives can form the alternative embodiment of the disclosure, herein no longer using any combination
It repeats one by one.
Fig. 2 is a kind of flow chart of similar pictures recognition methods shown according to an exemplary embodiment, as shown in Fig. 2,
Similar pictures recognition methods is for including the following steps in terminal or server.
In step 201, feature point extraction is carried out to two picture to be identified, obtains multiple features of two picture
Point.
In the embodiments of the present disclosure, two picture to be identified can be any in progress similar pictures identification process
Two pictures, the identification object of the similar pictures identification process can be in all pictures or certain several photograph album in a photograph album
All pictures be not specifically limited herein it is, of course, also possible to be the certain pictures specified by user.
The feature point extraction process of the embodiment of the present disclosure may include carrying out characteristic point respectively to two picture to be identified
It extracts, obtains multiple characteristic points of two picture.It should be noted that carrying out feature point extraction to two picture to be identified
It include: that SIFT or SURF carry out feature point extraction to two picture to be identified.
In step 202, according to multiple characteristic points of two picture, multiple fisrt feature points of two picture are obtained
It is right.
The process of the acquisition fisrt feature point pair can be carried out based on SIFT or SURF, and the embodiment of the present disclosure does not make this to have
Body limits.
During obtaining multiple fisrt feature points pair, be related to nearest distance with it is secondary it is close at a distance between ratio
The process being compared with preset threshold, at this point, the setting of preset threshold is directly related to finally obtained fisrt feature point
Pair number how much, when preset threshold setting it is larger when, obtained fisrt feature point is relatively fewer to number, when default threshold
Value setting it is smaller when, obtained fisrt feature point number is relatively more.
In step 203, from multiple fisrt feature point centering, multiple second feature points pair are determined, the second feature point
To for correct matched characteristic point pair.
Correct matched fisrt feature point pair in order to obtain is needed to multiple fisrt feature point to screening again,
So that it is determined that multiple second feature points pair.
In the embodiment of the present disclosure, only to be illustrated for determining the process of multiple second feature points pair based on SIFT.Its
In, for screening the default ratio of second feature point pair by terminal or server settings, if to improve second feature point pair
This can be preset the smaller of ratio setting by accuracy, but computation complexity will improve, if to the accurate of second feature point pair
Degree require it is lower, this can also be preset ratio setting it is larger.
In step 204, when the quantity ratio between multiple second feature point pair and multiple fisrt feature point pair
Greater than the first preset threshold and when the quantity of multiple second feature point pair is less than amount threshold, then it is special to calculate multiple second
The first angle between the line and vertical line of sign point pair.
It should be noted that being only illustrated so that two picture levels are put as an example in the embodiment of the present disclosure.And another
In embodiment, which can also vertically be put, then when calculating angle, can calculate multiple second feature point pair
The second angle between line and horizontal line.
Wherein, the quantity ratio between multiple second feature points and multiple fisrt feature point can be used for measuring two figures
The similarity degree of piece illustrates similarity degree height when quantity ratio is larger, when quantity ratio is smaller, illustrates that similarity degree is low, because
The first preset threshold can be first arranged when to multiple fisrt feature point to screening in this, when the quantity ratio is greater than first in advance
If when threshold value, it is believed that two picture is likely to similar pictures, in order to improve the accuracy of identification, it is also necessary to further
Step 205 is executed, and when the quantity ratio is less than or equal to the first preset threshold, it is believed that two pictures are not similar diagrams
Piece terminates identification process.
Wherein, the amount threshold is by terminal or server settings, and the amount threshold is a definite value.It is more obtaining this
During a second feature point pair, the also available quantity to multiple second feature point pair, when second feature point logarithm
Mesh is very few, for example, multiple second feature point is to only several pairs or more than ten clock synchronizations, if using RANSAC algorithm to multiple
Two characteristic points are likely to be the second of erroneous matching to the sample of random selection in set to screening, from the second feature point
Characteristic point pair, and the probability for generating the mistake is larger, therefore, it is necessary to the quantity to multiple second feature point pair to judge,
When the quantity of multiple second feature point pair is less than the amount threshold, the line of multiple second feature point pair can be calculated and hung down
The first angle between straight line, and first angle is screened, correct matched multiple second feature point pair is obtained, when
When the quantity of multiple second feature point pair is not less than the amount threshold, RANSAC algorithm can be used to multiple second feature point
To screening.
In the embodiment of the present disclosure, only with the quantity between multiple second feature point pair and multiple fisrt feature point pair
Ratio is illustrated for being less than amount threshold greater than the first preset threshold, and the quantity of multiple second feature point pair.And
In another embodiment, if quantity ratio between multiple second feature point pair and multiple fisrt feature point pair is greater than the
One preset threshold when the quantity of multiple second feature point pair is not less than amount threshold, can also still adopt in the next steps
With RANSAC algorithm to multiple second feature point to screening, to carry out similar pictures identification.Certainly, when multiple second
Quantity ratio between characteristic point pair and multiple fisrt feature point pair is not more than the first preset threshold, no matter second feature point
Pair number be how many, it is considered that two pictures are not similar pictures, terminate identification process.
In step 205, determine that the angular interval that each first angle is fallen into, statistics fall in second in each section
Characteristic point determines the angular interval of preset number to wiring quantity.
Wherein, which is the section by the angular divisions of straight line at least two equal angular spans, straight line
Angle is [0-180] degree.For example, [0-180] degree is divided into 18 sections, then the angular span in each angular interval is 10
Degree, first angular interval is [0,10], and second angular interval is [10,20], and third angular interval is [20,30], such as
This analogizes, and the 17th angular interval is [160,170], and the 18th angular interval is [170,180].
During statistics falls in second feature point in each section to wiring quantity, for each first angle,
Determine the angular interval that first angle is fallen into, and the wiring quantity in the real-time statistics angular intervals.For example, there is one first
The angle of angle is 15 degree, it is determined that the angle falls into above-mentioned second angular interval [10,20], then in real time at second angle
It spends and adds 1 in the statistics number in section.
Wherein, the angular span of the angular interval is by terminal or server settings, and the angular interval of the preset number is by this
Angular span and the second feature point for falling into each angular interval counted determine wiring quantity.When falling in each section
When the interior second feature point has counted completion to wiring quantity, by traversing the angular interval of statistics completion, find multiple
Continuous angular interval keeps second feature point in the continuous angular interval most to wiring quantity, by multiple continuous angle
Degree section is determined as the angular interval of preset number.For example, after being divided and counted based on above-mentioned angular interval division mode,
All angular intervals are traversed, find the second feature point in 2 continuous angular intervals [80,90] and [90,100] to line
Quantity is most, then the angular interval of the preset number is [80,90] and [90,100].
In step 206, when the wiring quantity included in the angular interval of the preset number is most, retains line and fall
Enter the second feature point pair in the angular interval of the preset number, to obtain multiple third feature points pair.
In general, if similar between two pictures, it may be considered that there are overlapping regions for two picture, it is correct matched
Characteristic point corresponds to the integrated distribution in the overlapping region between picture, however, position of the overlapping region in picture may also
There are certain displacements, therefore, can determine overlapping region according to the first angle of multiple second feature points pair, process can wrap
It includes: when the wiring quantity included in the angular region of the preset number is most, illustrating in the angular region of the preset number
Included second feature point logarithm is most, thus can determine the overlapping region of two picture to be identified, so, it can be by line
The second feature point fallen into the angular interval of the preset number to correct matched second feature point pair is regarded as, retain this
Two characteristic points pair, to obtain multiple third feature points pair.
It should be noted that not falling within the point of the second feature in the angular interval of the preset number to next for line
It says, then can determine the second feature point to being not in the overlapping region of two picture to be identified, it is possible to think
The second feature point is not to being correct matched second feature point pair.
Above-mentioned steps 205 and 206 are according to the line angle of multiple second feature point pair, to multiple second feature point
To screening, the process of multiple third feature points pair is obtained.In the embodiments of the present disclosure, only according to the angle of preset number
Section to second feature point to being illustrated for screening, and in another embodiment, can also be according to predetermined angle model
It encloses to second feature point to screening, obtains multiple third feature points pair.Wherein, the predetermined angle range is by terminal or service
Device setting, by taking two picture levels are put as an example, which can be 90 degree or so of any angle range, example
Such as, [80,100] etc., due to for similar pictures, it is also believed that there is no any position between the overlapping region of two pictures
It moves, therefore, if characteristic point is to correct matching, angle of this feature point pair between horizontal line also should be with its placing direction one
It causes, is between predetermined angle range, then the second feature point pair that can be located at the first included angle within the scope of predetermined angle
Regard correct matched second feature point pair as, and it is special to screen out the first included angle is not located within the scope of predetermined angle second
Sign point pair.
Similarly, when two picture to be identified is vertically to put, then the line of multiple second feature point pair is calculated
The second angle between horizontal line retains the second angle and is in the second feature point pair of predetermined angle range, obtains multiple the
Three characteristic points pair.
Wherein, during obtaining multiple third feature points pair, the also available number to multiple third feature point pair
Amount.
In step 207, when the quantity ratio between multiple third feature point pair and multiple second feature point pair is big
When the second preset threshold, determine that two picture is similar pictures.
Wherein, the second preset threshold can be configured by terminal or server, if to improve the identification of similar pictures
Accuracy, the second preset threshold can be arranged it is larger, and if recognition accuracy is required it is lower, can also be pre- by second
If threshold value setting is smaller.
It should be noted that when between multiple third feature point pair and multiple second feature point pair quantity ratio not
When greater than the second preset threshold, it is determined that two picture is dissimilar picture.
The method that the embodiment of the present disclosure provides is determining correct matched second feature point to later, second for the first time
Under the scene of the negligible amounts of characteristic point pair, according to the line angle of multiple second feature points pair, to multiple second feature points pair
It is screened again, retains the second feature point pair being located in the overlapping region of similar pictures, screen out and be not located at overlapping region
Second feature point pair, then based on remaining third feature point pair and multiple second feature point pair and multiple second feature point
Quantity ratio between, carrys out the identification of further progress similar pictures, so that in less characteristic point to the identification process of scene
In avoid identification the higher situation of error rate, improve similar pictures identification speed and accuracy.
Fig. 3 is a kind of similar pictures identification device block diagram shown according to an exemplary embodiment.Referring to Fig. 3, the device
Including feature point extraction module 301, fisrt feature point is to module 302 is obtained, and for second feature point to determining module 303, third is special
Sign point is to screening module 304 and similar pictures determining module 305.
Feature point extraction module 301 obtains two picture for carrying out feature point extraction to two picture to be identified
Multiple characteristic points;
Fisrt feature point obtains two picture for multiple characteristic points according to two picture to acquisition module 302
Multiple fisrt feature points pair;
Second feature point is to determining module 303, for determining multiple second feature points from multiple fisrt feature point centering
Right, the second feature point is to for correct matched characteristic point pair;
Third feature point is to screening module 304, for when multiple second feature point pair and multiple fisrt feature point pair
Between quantity ratio be greater than the quantity of the first preset threshold and multiple second feature point pair and be less than amount threshold, according to
The line angle of multiple second feature point pair obtains multiple third feature points to multiple second feature point to screening
It is right;
Similar pictures determining module 305, between multiple third feature point pair and multiple second feature point pair
Quantity ratio be greater than the second preset threshold when, determine two picture be similar pictures.
In one embodiment of the present disclosure, this feature point extraction module 301 be used for using scale invariant feature SIFT or
Accelerate robust feature SURF, feature point extraction is carried out to two similar pictures to be identified, obtains multiple spies of two picture
Sign point.
In one embodiment of the present disclosure, which is used to determining module 303 work as multiple second feature
Point is greater than the first preset threshold and multiple second feature point pair to the quantity ratio between multiple fisrt feature point pair
Quantity be not less than amount threshold, using stochastical sampling consistency RANSAC algorithm to multiple second feature point to sieving
Choosing.
In one embodiment of the present disclosure, which is used for screening module 304:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line determines that the angular interval that each first angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line determines that the angular interval that each second angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number is most to wiring quantity
More, retaining line, this falls into the point pair of the second feature in the angular interval of the preset number.
In one embodiment of the present disclosure, which is used for screening module 304:
When two picture to be identified is put for level, then calculate the line of multiple second feature point pair with it is vertical
The first angle between line retains the second feature point pair that the first angle is in predetermined angle range;Or,
When two picture to be identified is vertically to put, then the line and level of multiple second feature point pair are calculated
The second angle between line retains the second feature point pair that the second angle is in predetermined angle range.
About the device in above-described embodiment, wherein modules execute the concrete mode of operation in related this method
Embodiment in be described in detail, no detailed explanation will be given here.
Fig. 4 is a kind of block diagram of similar pictures identification device 400 shown according to an exemplary embodiment.For example, device
400 can be mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, and medical treatment is set
It is standby, body-building equipment, personal digital assistant etc..
Referring to Fig. 4, device 400 may include following one or more components: processing component 402, memory 404, power supply
Component 406, multimedia component 408, audio component 410, the interface 412 of input/output (I/O), sensor module 414, and
Communication component 416.
The integrated operation of the usual control device 400 of processing component 402, such as with display, telephone call, data communication, phase
Machine operation and record operate associated operation.Processing component 402 may include that one or more processors 420 refer to execute
It enables, to perform all or part of the steps of the methods described above.In addition, processing component 402 may include one or more modules, just
Interaction between processing component 402 and other assemblies.For example, processing component 402 may include multi-media module, it is more to facilitate
Interaction between media component 408 and processing component 402.
Memory 404 is configured as storing various types of data to support the operation in device 400.These data are shown
Example includes the instruction of any application or method for operating on device 400, contact data, and telephone book data disappears
Breath, picture, video etc..Memory 404 can be by any kind of volatibility or non-volatile memory device or their group
It closes and realizes, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM) is erasable to compile
Journey read-only memory (EPROM), programmable read only memory (PROM), read-only memory (ROM), magnetic memory, flash
Device, disk or CD.
Power supply module 406 provides electric power for the various assemblies of device 400.Power supply module 406 may include power management system
System, one or more power supplys and other with for device 400 generate, manage, and distribute the associated component of electric power.
Multimedia component 408 includes the screen of one output interface of offer between the device 400 and user.Some
In embodiment, screen may include liquid crystal display (LCD) and touch panel (TP).If screen includes touch panel, screen
It may be implemented as touch screen, to receive input signal from the user.Touch panel includes one or more touch sensors
To sense the gesture on touch, slide, and touch panel.The touch sensor can not only sense the side of touch or sliding action
Boundary, but also detect duration and pressure relevant to the touch or slide.In some embodiments, multimedia component
408 include a front camera and/or rear camera.When device 400 is in operation mode, such as screening-mode or video screen module
When formula, front camera and/or rear camera can receive external multi-medium data.Each front camera and postposition are taken the photograph
As head can be a fixed optical lens system or have focusing and optical zoom capabilities.
Audio component 410 is configured as output and/or input audio signal.For example, audio component 410 includes a Mike
Wind (MIC), when device 400 is in operation mode, when such as call mode, recording mode, and voice recognition mode, microphone is matched
It is set to reception external audio signal.The received audio signal can be further stored in memory 404 or via communication set
Part 416 is sent.In some embodiments, audio component 410 further includes a loudspeaker, is used for output audio signal.
I/O interface 412 provides interface between processing component 402 and peripheral interface module, and above-mentioned peripheral interface module can
To be keyboard, click wheel, button etc..These buttons may include, but are not limited to: home button, volume button, start button and lock
Determine button.
Sensor module 414 includes one or more sensors, and the state for providing various aspects for device 400 is commented
Estimate.For example, sensor module 414 can detecte the state that opens/closes of device 400, the relative positioning of component, such as the group
Part is the display and keypad of device 400, and sensor module 414 can be with 400 1 components of detection device 400 or device
Position change, the existence or non-existence that user contacts with device 400, the temperature in 400 orientation of device or acceleration/deceleration and device 400
Degree variation.Sensor module 414 may include proximity sensor, be configured to detect without any physical contact attached
The presence of nearly object.Sensor module 414 can also include optical sensor, such as CMOS or ccd image sensor, for being imaged
It is used in.In some embodiments, the sensor module 414 can also include acceleration transducer, gyro sensor,
Magnetic Sensor, pressure sensor or temperature sensor.
Communication component 416 is configured to facilitate the communication of wired or wireless way between device 400 and other equipment.Device
400 can access the wireless network based on communication standard, such as WiFi, 2G or 3G or their combination.In an exemplary implementation
In example, communication component 416 receives broadcast singal or broadcast related information from external broadcasting management system via broadcast channel.
In one exemplary embodiment, which further includes near-field communication (NFC) module, to promote short range communication.Example
Such as, NFC module can be based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra wide band (UWB) technology,
Bluetooth (BT) technology and other technologies are realized.
In the exemplary embodiment, device 400 can be believed by one or more application specific integrated circuit (ASIC), number
Number processor (DSP), digital signal processing appts (DSPD), programmable logic device (PLD), field programmable gate array
(FPGA), controller, microcontroller, microprocessor or other electronic components are realized, for executing above-mentioned similar pictures identification side
Method.
In the exemplary embodiment, a kind of non-transitorycomputer readable storage medium including instruction, example are additionally provided
It such as include the memory 404 of instruction, above-metioned instruction can be executed by the processor 420 of device 400 to complete the above method.For example,
The non-transitorycomputer readable storage medium can be ROM, random access memory (RAM), CD-ROM, tape, floppy disk and
Optical data storage devices etc..
In the exemplary embodiment, a kind of non-transitorycomputer readable storage medium is additionally provided, when the storage medium
In instruction by mobile terminal processor execute when so that mobile terminal is able to carry out similar pictures provided by the above embodiment
Recognition methods.
Fig. 5 is a kind of block diagram of the device 500 of similar pictures identification shown according to an exemplary embodiment.For example, dress
Setting 500 may be provided as a server.Referring to Fig. 5, device 500 includes processing component 522, further comprises one or more
A processor, and the memory resource as representated by memory 532, can be by the finger of the execution of processing component 522 for storing
It enables, such as application program.The application program stored in memory 532 may include it is one or more each correspond to
The module of one group of instruction.In addition, processing component 522 is configured as executing instruction, to execute above-mentioned similar pictures recognition methods.
Device 500 can also include the power management that a power supply module 526 is configured as executive device 500, and one has
Line or radio network interface 550 are configured as device 500 being connected to network and input and output (I/O) interface 558.Dress
Setting 500 can operate based on the operating system for being stored in memory 532, such as Windows ServerTM, Mac OS XTM,
UnixTM,LinuxTM, FreeBSDTMOr it is similar.
Those skilled in the art after considering the specification and implementing the invention disclosed here, will readily occur to its of the disclosure
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the disclosure, these modifications, purposes or
Person's adaptive change follows the general principles of this disclosure and including the undocumented common knowledge in the art of the disclosure
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the disclosure are by following
Claim is pointed out.
It should be understood that the present disclosure is not limited to the precise structures that have been described above and shown in the drawings, and
And various modifications and changes may be made without departing from the scope thereof.The scope of the present disclosure is only limited by the accompanying claims.
Claims (11)
1. a kind of similar pictures recognition methods, which is characterized in that the described method includes:
Feature point extraction is carried out to two picture to be identified, obtains multiple characteristic points of two picture;
According to multiple characteristic points of two picture, multiple fisrt feature points pair of two picture are obtained;
From the multiple fisrt feature point centering, determine that multiple second feature points pair, the second feature point are matched to be correct
Characteristic point pair;
It is preset when the quantity ratio between the multiple second feature point pair and the multiple fisrt feature point pair is greater than first
Threshold value and the quantity of the multiple second feature point pair are less than amount threshold, according to the line of the multiple second feature point pair
Angle obtains multiple third feature points pair to the multiple second feature point to screening;
When the quantity ratio between the multiple third feature point pair and the multiple second feature point pair is greater than the second default threshold
When value, determine that two picture is similar pictures.
2. the method according to claim 1, wherein described propose two picture to be identified progress characteristic point
It takes, the multiple characteristic points for obtaining two picture include:
Using scale invariant feature SIFT or accelerate robust feature SURF, characteristic point is carried out to two similar pictures to be identified and is mentioned
It takes, obtains multiple characteristic points of two picture.
3. determination is more the method according to claim 1, wherein described from the multiple fisrt feature point centering
A second feature point pair, the second feature point to for correct matched characteristic point pair, the method also includes:
It is preset when the quantity ratio between the multiple second feature point pair and the multiple fisrt feature point pair is greater than first
Threshold value and the quantity of the multiple second feature point pair are not less than amount threshold, using stochastical sampling consistency RANSAC algorithm
To the multiple second feature point to screening.
4. the method according to claim 1, wherein the line angle according to the multiple second feature point pair
It spends, includes: to screening is carried out to the multiple second feature point
When two picture to be identified is put for level, then calculate the line of the multiple second feature point pair with it is vertical
The first angle between line determines that the angular interval that each first angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number to wiring quantity
At most, retain described in line and fall into the second feature point pair in the angular interval of the preset number;Or,
When two picture to be identified is vertically to put, then the line and level of the multiple second feature point pair are calculated
The second angle between line determines that the angular interval that each second angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number to wiring quantity
At most, retain described in line and fall into the second feature point pair in the angular interval of the preset number.
5. the method according to claim 1, wherein the line angle according to the multiple second feature point pair
It spends, includes: to screening is carried out to the multiple second feature point
When two picture to be identified is put for level, then calculate the line of the multiple second feature point pair with it is vertical
The first angle between line retains the second feature point pair that the first angle is in predetermined angle range;Or,
When two picture to be identified is vertically to put, then the line and level of the multiple second feature point pair are calculated
The second angle between line retains the second feature point pair that the second angle is in predetermined angle range.
6. a kind of similar pictures identification device, which is characterized in that described device includes:
Feature point extraction module obtains the more of two picture for carrying out feature point extraction to two picture to be identified
A characteristic point;
Fisrt feature point, for multiple characteristic points according to two picture, obtains two picture to module is obtained
Multiple fisrt feature points pair;
Second feature point is to determining module, for determining multiple second feature points pair, institute from the multiple fisrt feature point centering
Second feature point is stated to for correct matched characteristic point pair;
Third feature point to screening module, for when the multiple second feature point pair and the multiple fisrt feature point pair it
Between quantity ratio be greater than the quantity of the first preset threshold and the multiple second feature point pair and be less than amount threshold, according to institute
The line angle for stating multiple second feature points pair obtains multiple third feature to the multiple second feature point to screening
Point pair;
Similar pictures determining module, for when the number between the multiple third feature point pair and the multiple second feature point pair
When measuring ratio greater than the second preset threshold, determine that two picture is similar pictures.
7. device according to claim 6, which is characterized in that the feature point extraction module is also used to using Scale invariant
Feature SIFT accelerates robust feature SURF, carries out feature point extraction to two similar pictures to be identified, obtains described two
Multiple characteristic points of picture.
8. device according to claim 6, which is characterized in that the second feature point is also used to when described determining module
Quantity ratio between multiple second feature points pair and the multiple fisrt feature point pair is greater than the first preset threshold and described
The quantity of multiple second feature points pair is not less than amount threshold, using stochastical sampling consistency RANSAC algorithm to the multiple the
Two characteristic points are to screening.
9. device according to claim 6, which is characterized in that the third feature point is also used to screening module:
When two picture to be identified is put for level, then calculate the line of the multiple second feature point pair with it is vertical
The first angle between line determines that the angular interval that each first angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number to wiring quantity
At most, retain described in line and fall into the second feature point pair in the angular interval of the preset number;Or,
When two picture to be identified is vertically to put, then the line and level of the multiple second feature point pair are calculated
The second angle between line determines that the angular interval that each second angle is fallen into, statistics fall in the second feature in each section
Point determines the angular interval of preset number, wiring quantity included in the angular interval of the preset number to wiring quantity
At most, retain described in line and fall into the second feature point pair in the angular interval of the preset number.
10. device according to claim 6, which is characterized in that the third feature point is also used to screening module:
When two picture to be identified is put for level, then calculate the line of the multiple second feature point pair with it is vertical
The first angle between line retains the second feature point pair that the first angle is in predetermined angle range;Or,
When two picture to be identified is vertically to put, then the line and level of the multiple second feature point pair are calculated
The second angle between line retains the second feature point pair that the second angle is in predetermined angle range.
11. a kind of similar pictures identification device characterized by comprising
Processor;
The instruction that can be performed for storage processor;
Wherein, the processor is configured to:
Feature point extraction is carried out to two picture to be identified, obtains multiple characteristic points of two picture;
According to multiple characteristic points of two picture, multiple fisrt feature points pair of two picture are obtained;
From the multiple fisrt feature point centering, determine that multiple second feature points pair, the second feature point are matched to be correct
Characteristic point pair;
It is preset when the quantity ratio between the multiple second feature point pair and the multiple fisrt feature point pair is greater than first
Threshold value and the quantity of the multiple second feature point pair are less than amount threshold, according to the line of the multiple second feature point pair
Angle obtains multiple third feature points pair to the multiple second feature point to screening;
When the quantity ratio between the multiple third feature point pair and the multiple second feature point pair is greater than the second default threshold
When value, determine that two picture is similar pictures.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510549204.5A CN105205494B (en) | 2015-08-31 | 2015-08-31 | Similar pictures recognition methods and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510549204.5A CN105205494B (en) | 2015-08-31 | 2015-08-31 | Similar pictures recognition methods and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105205494A CN105205494A (en) | 2015-12-30 |
CN105205494B true CN105205494B (en) | 2018-12-11 |
Family
ID=54953166
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510549204.5A Active CN105205494B (en) | 2015-08-31 | 2015-08-31 | Similar pictures recognition methods and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105205494B (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105913069B (en) * | 2016-04-27 | 2019-05-31 | 南京维睛视空信息科技有限公司 | A kind of image-recognizing method |
CN106021542A (en) * | 2016-05-26 | 2016-10-12 | 珠海市魅族科技有限公司 | Image display method and image server, terminal |
CN106682683B (en) * | 2016-11-03 | 2020-09-29 | 知酒(上海)网络科技有限公司 | Wine label picture identification method and device |
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 |
CN110414588A (en) * | 2019-07-23 | 2019-11-05 | 广东小天才科技有限公司 | Picture mask method, device, computer equipment and storage medium |
CN110927767A (en) * | 2019-11-28 | 2020-03-27 | 合肥工业大学 | Following system for special crowds |
CN111143581A (en) * | 2019-12-31 | 2020-05-12 | 汇医达(杭州)科技有限公司 | Data box system for journal posting and manuscript screening |
CN111724442B (en) * | 2020-05-28 | 2022-04-22 | 上海商汤智能科技有限公司 | Image processing method and device, electronic device and storage medium |
CN112507992B (en) * | 2021-02-04 | 2021-05-07 | 腾讯科技(深圳)有限公司 | Method, device, equipment and medium for determining shooting distance between road images |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103473565A (en) * | 2013-08-23 | 2013-12-25 | 华为技术有限公司 | Image matching method and device |
CN103745449A (en) * | 2013-12-24 | 2014-04-23 | 南京理工大学 | Rapid and automatic mosaic technology of aerial video in search and tracking system |
-
2015
- 2015-08-31 CN CN201510549204.5A patent/CN105205494B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103473565A (en) * | 2013-08-23 | 2013-12-25 | 华为技术有限公司 | Image matching method and device |
CN103745449A (en) * | 2013-12-24 | 2014-04-23 | 南京理工大学 | Rapid and automatic mosaic technology of aerial video in search and tracking system |
Non-Patent Citations (2)
Title |
---|
基于局部特征和进化算法的人脸识别;李根;《中国博士学位论文全文数据库》;20140930;全文 * |
基于距离约束的图像拼接技术研究;吴恩生;《中国优秀硕士学位论文全文数据库》;20150630;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN105205494A (en) | 2015-12-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105205494B (en) | Similar pictures recognition methods and device | |
US10115019B2 (en) | Video categorization method and apparatus, and storage medium | |
CN107239535A (en) | Similar pictures search method and device | |
CN105069426B (en) | Similar pictures judgment method and device | |
CN106651955A (en) | Method and device for positioning object in picture | |
CN106572299A (en) | Camera switching-on method and device | |
CN106951884A (en) | Gather method, device and the electronic equipment of fingerprint | |
CN105758319B (en) | The method and apparatus for measuring target object height by mobile terminal | |
CN105426878B (en) | Face cluster method and device | |
CN107862093B (en) | File attribute identification method and device | |
CN104933700B (en) | A kind of method and apparatus carrying out picture material identification | |
CN110933468A (en) | Playing method, playing device, electronic equipment and medium | |
CN106339695A (en) | Face similarity detection method, device and terminal | |
CN108108671A (en) | Description of product information acquisition method and device | |
CN106303198A (en) | Photographing information acquisition methods and device | |
CN108989687A (en) | camera focusing method and device | |
CN108717542A (en) | Identify the method, apparatus and computer readable storage medium of character area | |
CN110490303A (en) | Super-network construction method, application method, device and medium | |
CN104573642A (en) | Face recognition method and device | |
CN106295468B (en) | Face identification method and device | |
CN106997356A (en) | The sorting technique and device of picture | |
CN106339452B (en) | Event response method and device | |
CN112749590B (en) | Object detection method, device, computer equipment and computer readable storage medium | |
CN107688765B (en) | Fingerprint acquisition method and device | |
CN109376674A (en) | Method for detecting human face, device and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |