CN109086744A - Information processing method and device - Google Patents
Information processing method and device Download PDFInfo
- Publication number
- CN109086744A CN109086744A CN201810993318.2A CN201810993318A CN109086744A CN 109086744 A CN109086744 A CN 109086744A CN 201810993318 A CN201810993318 A CN 201810993318A CN 109086744 A CN109086744 A CN 109086744A
- Authority
- CN
- China
- Prior art keywords
- sift
- video image
- cluster centre
- description
- cluster
- 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
- 230000010365 information processing Effects 0.000 title claims abstract description 39
- 238000003672 processing method Methods 0.000 title claims abstract description 21
- 238000000034 method Methods 0.000 claims abstract description 37
- 239000000284 extract Substances 0.000 claims abstract description 18
- 238000000605 extraction Methods 0.000 claims description 6
- 238000012550 audit Methods 0.000 abstract description 20
- 238000010586 diagram Methods 0.000 description 4
- 238000012512 characterization method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000003064 k means clustering Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- 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]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
- H04N21/234—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs
- H04N21/23418—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
Information processing method and device provided by the invention, which comprises obtain target video image;Extract SIFT description of the target video image;Using preset image data base, effective SIFT description is determined from the SIFT of target video image description;When the number of effective SIFT of target video image description meets preset number condition, determination retrieves the target video image.The present invention realizes the automation review process of video content, and according to cluster centre corresponding with the SIFT of reference video image description, and the distance between cluster centre in SIFT description and described image database of target video image relationship, to determine whether to retrieve the target video image, the audit precision and review efficiency of video content are improved, while saving a large amount of human cost.
Description
Technical field
The present invention relates to technical field of information processing, more specifically, being related to information processing method and device.
Background technique
With the rise and the prevalence of various video softwares of net cast platform, all there can be the video data of magnanimity to exist daily
It is propagated on internet, in order to guarantee health, the harmony of video content, needs strictly to audit the video content of magnanimity, with
Prevent the propagation of sensitive information.
Traditional technical solution, the usually mode dependent on manual inspection video content, to determine in video whether deposit
It is all relatively low in sensitive information, either review efficiency, or audit precision, it can not effectively meet examining for massive video data
Core demand.
Therefore, there is an urgent need to a kind of effective technical solutions at present, to improve the review efficiency of video content and examine
Core precision.
Summary of the invention
In view of this, the present invention provides a kind of information processing method and device, to solve the audit of current video content
The efficiency technical problem lower with audit precision.
To achieve the above object, the invention provides the following technical scheme:
A kind of information processing method, which comprises
Obtain target video image;
Extract Scale invariant features transform SIFT description of the target video image;
Using preset image data base, determine that effective SIFT is retouched from the SIFT of target video image description
State son;It wherein, include: cluster centre corresponding with the SIFT of reference video image description in described image database;Institute
Stating effective SIFT and describing son is that the SIFT for meeting pre-determined distance condition at a distance from the cluster centre in described image database is retouched
State son;
When the number of effective SIFT of target video image description meets preset number condition, determination is retrieved
The target video image.
Preferably, the establishment process of described image database includes:
Obtain reference video image;
Extract SIFT description of the reference video image;
SIFT description based on the reference video image carries out clustering processing, obtains cluster result;
According to the cluster result, described image database is established.
Preferably, the acquisition reference video image includes:
Obtain reference video;
The reference video is decoded as at least one video image;
Clustering processing is carried out at least one described video image, obtains the first cluster centre set;First cluster
It is included at least in centralization: the corresponding cluster centre of each video image at least one described video image;
From the corresponding video image of cluster centre each in the first cluster centre set, it is pre- to extract satisfaction
If the video image of extracting rule, as the reference video image.
Preferably, SIFT description based on the reference video image carries out clustering processing, obtains cluster result
Include:
Clustering processing is carried out to SIFT description of the reference video image, obtains the second cluster centre set;It is described
It is included at least in second cluster centre set: in the corresponding cluster of each SIFT description of the reference video image
The heart;
The difference between the first SIFT description cluster centre corresponding with the first SIFT description is calculated, as institute
State the difference of the first SIFT description;First SIFT describes son and is, in SIFT description of the reference video image
Any SIFT description;
Clustering processing is carried out to the difference of SIFT description of the reference video image, obtains third cluster centre collection
It closes;It is included at least in the third cluster centre set: the difference difference of each SIFT description of the reference video image
Corresponding cluster centre;
Wherein, the cluster result includes at least the second cluster centre set and the third cluster centre set.
Preferably, described according to the cluster result, establishing described image database includes:
Each SIFT based on the reference video image describes son, establishes each poly- in the second cluster centre set
Corresponding relationship in class center and the third cluster centre set between each cluster centre;
According to the second cluster centre set, the third cluster centre set and the corresponding relationship, described in foundation
Image data base.
It preferably, include: the second cluster centre set and third cluster centre set in described image database;The benefit
With preset image data base, determine that effective SIFT describes son and includes: from the SIFT of target video image description
From the second cluster centre set, determining nearest cluster centre at a distance from the 2nd SIFT description is made
For the first cluster centre;Wherein, it is any in SIFT description of the target video image that the 2nd SIFT, which describes son,
SIFT description;
From the third cluster centre set, cluster centre corresponding with first cluster centre is determined, make
For the second cluster centre;
When the 2nd SIFT description and the maximum range value of second cluster centre are less than pre-determined distance threshold value,
2nd SIFT is described into son and is determined as effective SIFT description.
Preferably, described when the number of effective SIFT description of the target video image meets preset number condition
When, determine that retrieving the target video image includes:
All SIFT when the number of effective SIFT description of the target video image, with the target video image
Ratio between the number of son is described, when being greater than default fractional threshold, determination retrieves the target video image.
A kind of information processing unit, described device include:
Target image acquiring unit, for obtaining target video image;
Sub- extraction unit is described, the SIFT for extracting the target video image describes son;
Sub- determination unit is effectively described, for utilizing preset image data base, from the SIFT of the target video image
Effective SIFT description is determined in description;It wherein, include: to be retouched with the SIFT of reference video image in described image database
State the corresponding cluster centre of son;Effective SIFT describes son and is, at a distance from the cluster centre in described image database
Meet SIFT description of pre-determined distance condition;
As a result determination unit, the number for effective SIFT description when the target video image meet preset number
When condition, determination retrieves the target video image.
Preferably, described device further include:
The Database unit, for obtaining reference video image;The SIFT for extracting the reference video image is retouched
State son;SIFT description based on the reference video image carries out clustering processing, obtains cluster result;It is tied according to the cluster
Fruit establishes described image database.
It preferably, include: the second cluster centre set and third cluster centre set in described image database;It is described to have
Effect describes sub- determination unit and is specifically used for:
From the second cluster centre set, determining nearest cluster centre at a distance from the 2nd SIFT description is made
For the first cluster centre;Wherein, it is any in SIFT description of the target video image that the 2nd SIFT, which describes son,
SIFT description;
From the third cluster centre set, cluster centre corresponding with first cluster centre is determined, make
For the second cluster centre;
When the 2nd SIFT description and the maximum range value of second cluster centre are less than pre-determined distance threshold value,
2nd SIFT is described into son and is determined as effective SIFT description.
It can be seen from the above technical scheme that information processing method provided by the invention and device, obtain target video
Image;Extract SIFT description of the target video image;Using preset image data base, from the target video image
SIFT description in determine effective SIFT description;When the number symbol of effective SIFT description of the target video image
When closing preset number condition, determination retrieves the target video image, realizes the automation review process of video content, and
And it is retouched according to the SIFT for describing sub corresponding cluster centre and the target video image with the SIFT of reference video image
The distance between the cluster centre in son and described image database relationship is stated, to determine whether to retrieve the target video figure
Picture, improves the audit precision and review efficiency of video content, while saving a large amount of human cost.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The embodiment of invention for those of ordinary skill in the art without creative efforts, can also basis
The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of flow chart of information processing method provided in an embodiment of the present invention;
Fig. 2 is a kind of flow chart of image data base establishment process provided in an embodiment of the present invention;
Fig. 3 is another flow chart of image data base establishment process provided in an embodiment of the present invention;
Fig. 4 is another flow chart of information processing method provided in an embodiment of the present invention;
Fig. 5 is a kind of structural schematic diagram of information processing unit provided in an embodiment of the present invention;
Fig. 6 is another structural schematic diagram of information processing unit provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The audit of traditional video content is to determine in video usually by the way of manually checking video content
No there are sensitive informations, either review efficiency, or audit precision, all relatively low, especially face the video data of magnanimity
When, not only review efficiency not can guarantee with audit precision, but also will increase a large amount of human cost.Therefore, the present invention provides
Examining for video content not only can be improved to realize the automation audit of video content in following information processing method and device
Core efficiency and audit precision, but also human cost can be greatly reduced, mitigate work load.
Referring to Fig. 1, Fig. 1 is a kind of flow chart of information processing method provided in an embodiment of the present invention.
As shown in Figure 1, which comprises
S110: target video image is obtained.
Pending video can be decoded as at least one video image.Target video image refers to, from pending video
In any one video image for extracting.
S120: Scale invariant features transform SIFT description of the target video image is extracted.
SIFT (Scale-invariant feature transform, Scale invariant features transform) is at image
A kind of description in reason field, this description have scale invariability, can detect that key point and a kind of part are special in the picture
Sign description.
S130: utilizing preset image data base, determines effectively from the SIFT of target video image description
SIFT description.
It include: cluster centre corresponding with the SIFT of reference video image description in described image database.It is described
Reference video image refers to the video image as audit benchmark.For example, if desired auditing in video content with the presence or absence of sensitivity
Information is specially then the video image for carrying sensitive information using reference video image.
SIFT description based on the reference video image carries out clustering processing, available and reference video image
The corresponding cluster centre of SIFT description, wherein cluster centre corresponding with the SIFT of reference video image description is used
In the characteristic information for characterizing the reference video image.
Wherein, it is to meet at a distance from the cluster centre in described image database default that effective SIFT, which describes son,
The SIFT of distance condition describes son.
When any SIFT description of target video image meets at a distance from the cluster centre in described image database
When pre-determined distance condition, can determine between target video image and reference video image has the same or similar feature letter
Breath.
S140: it when the number of effective SIFT of target video image description meets preset number condition, determines
Retrieve the target video image.
When the number of effective SIFT of target video image description meets preset number condition, show that target regards
Frequency image and reference video image are the same or similar image, it may be determined that retrieve the target video image.
Correspondingly, when the number of effective SIFT of target video image description does not meet preset number condition,
Determination does not retrieve the target video image.
It is described to retrieve the mesh when reference video image is the video image for carrying sensitive information in one example
Video image characterization is marked, carries sensitive information in the target video image;It is described not retrieve the target video image
It characterizes, does not carry sensitive information in the target video image.
When retrieving any video image of pending video, characterizes in the pending video and carry sensitive letter
Breath, can no longer be handled other video images in the pending video, to improve the audit effect of the pending video
Rate;When all video images of pending video are not retrieved, characterize in the pending video and do not carry sensitivity
Information, to realize that the automation for pending video is audited.
Information processing method provided in this embodiment obtains target video image;Extract the target video image
SIFT description;Using preset image data base, determined effectively from the SIFT of target video image description
SIFT description;When the number of effective SIFT of target video image description meets preset number condition, inspection is determined
Rope realizes the automation review process of video content to the target video image, and according to reference video image
SIFT is described in SIFT description and described image database of sub corresponding cluster centre and the target video image
The distance between cluster centre relationship improve examining for video content to determine whether to retrieve the target video image
Core precision and review efficiency, while saving a large amount of human cost.
In a particular application, the embodiment of the present application also provides the establishment processes of corresponding image data base.
Referring to Fig. 2, Fig. 2 is a kind of flow chart of image data base establishment process provided in an embodiment of the present invention.
As shown in Fig. 2, described image database creation process includes:
S210: reference video image is obtained.
Reference video image is can be and extracts from reference video for the video image as audit benchmark
At least one video image.
S220: SIFT description of the reference video image is extracted.
The SIFT of reference video image describes son, can be used for describing the characteristic information of reference video image.Specifically, work as institute
When to state reference video image be the image for carrying sensitive information, the SIFT of the reference video image describes son, can be used for retouching
State the sensitive information of the reference video image.
S230: SIFT description based on the reference video image carries out clustering processing, obtains cluster result.
Wherein, K-means clustering algorithm can be used, son is described to the SIFT of the reference video image and is carried out at cluster
Reason.
The cluster result includes at least: in each SIFT of reference video image description respectively corresponding cluster
The heart.
S240: according to the cluster result, described image database is established.
When the reference video image is the video image for carrying sensitive information, described image database is exactly to be used for
Retrieval carries the database of the video image of sensitive information, when using Image-Database Retrieval to target video image, then
It determines that the target video image carries sensitive information, realizes the audit for being directed to target video image.
In image data base establishment process provided in this embodiment, reference video image is obtained, the reference video is extracted
The SIFT of image describes son, and SIFT description based on the reference video image carries out clustering processing, obtains cluster result, and
According to the cluster result, described image database is established, to improve the standard of the image feature information in image data base
True property.
Referring to Fig. 3, Fig. 3 is another flow chart of image data base establishment process provided in an embodiment of the present invention.
As shown in figure 3, described image database creation process includes:
S310: reference video image is obtained.
In one example, step S310 is specific can include:
A1, reference video is obtained.
A2, the reference video is decoded as at least one video image.
Wherein, using CPU hard decoder or GPU hard decoder mode, the reference video is decoded as at least one video
Image.
A3, clustering processing is carried out at least one described video image, obtains the first cluster centre set.
It is included at least in the first cluster centre set: each video image difference at least one described video image
Corresponding cluster centre.
In one example, K-means clustering algorithm can be used, clustering processing is carried out at least one described video image,
In, the value of K depends on the number of different video content.For example, when reference video includes N number of different video content, then K
Value is equal to N, N > 1.The different video content can refer to sensitive camera lens different in reference video.In different clusters
The heart corresponds to different video contents.
A4, from the corresponding video image of cluster centre each in the first cluster centre set, extract full
The video image of the default extracting rule of foot, as the reference video image.
It wherein, can the cluster centre be nearest with selected distance for any cluster centre in the first cluster centre set
M video image, as the corresponding reference video image of the cluster centre, and then it is corresponding to obtain each cluster centre
Reference video image.In one example, the value of the M is preferably 4, naturally it is also possible to and it is other values, it can be flexible according to specific requirements
Setting.
Further, the M video image is to be greater than the view of default similarity threshold with the similarity of the cluster centre
Frequency image.In one example, the default similarity threshold is 90%, naturally it is also possible to it is other values, it can be according to specific requirements spirit
Setting living.
When the value of M is 4 and default similarity threshold is 90%, the standard of the video image extracted can be fully ensured that
True property.The video image extracted under this condition can fully comply with the image feature information that the cluster centre is characterized.
Correspondingly, the default extracting rule can include: from the corresponding video image of cluster centre, extract and the cluster
The similarity at center is greater than the video image of default similarity threshold;Alternatively, being extracted from the corresponding video image of cluster centre
It is greater than the video image of the preset number of default similarity threshold with the similarity of the cluster centre.
S320: SIFT description of the reference video image is extracted.
SIFT description for extracting the reference video image, refers to all reference video images that extraction is got
SIFT description.
In one example, using OpenCV (Open Source Computer Vision Library, computer of increasing income
Vision library), extract SIFT description of the reference video image.
S330: clustering processing is carried out to SIFT description of the reference video image, obtains the second cluster centre set.
Include at least in the second cluster centre set: each SIFT description of the reference video image is right respectively
The cluster centre answered.
It is by ginsengs all in pending video when SIFT description to the reference video image carries out clustering processing
All SIFT description for examining video image, which is summarised in, comes together to carry out unified clustering processing, and records each SIFT description
Which incorporated into respectively to cluster centre.
In one example, K-mean clustering algorithm can be used, son is described to the SIFT of the reference video image and is gathered
Class processing, obtains the second cluster centre set, wherein K value is preferably 1024, i.e., includes in the described second cluster centre set
1024 cluster centres.
S340: calculating the difference between the first SIFT description cluster centre corresponding with the first SIFT description,
Difference as the first SIFT description.
The reference video image includes at least one SIFT description.First SIFT describes son and is, the reference
Any SIFT in SIFT description of video image describes son.
S350: clustering processing is carried out to the difference of SIFT description of the reference video image, is obtained in third cluster
Heart set.
When the difference of SIFT description to the reference video image carries out clustering processing, being will be in pending video
The difference of all SIFT description of all reference video images, which is summarised in, comes together to carry out unified clustering processing, and records each
The difference of a SIFT description is incorporated into respectively to which cluster centre.
It is included at least in the third cluster centre set: the difference of each SIFT description of the reference video image
Corresponding cluster centre.
In one example, K-mean clustering algorithm can be used, to the difference of SIFT description of the reference video image
Clustering processing is carried out, obtains third cluster centre set, wherein K value is preferably 256, i.e., in the described third cluster centre set
Including 256 cluster centres.
Wherein, the step S330~S350 can be used for realizing the step S230 in previous embodiment, the cluster result
It includes at least: the second cluster centre set and the third cluster centre set.
S360: each SIFT based on the reference video image describes son, establishes in the second cluster centre set
Corresponding relationship in each cluster centre and the third cluster centre set between each cluster centre.
Wherein, any SIFT description of the reference video image is in the second cluster centre set in corresponding cluster
The heart between corresponding cluster centre, has corresponding relationship in third cluster centre set with the difference of SIFT description.
S370: it according to the second cluster centre set, the third cluster centre set and the corresponding relationship, establishes
Described image database.
In the second cluster centre set in each cluster centre and the third cluster centre set in each cluster
Corresponding relationship between the heart can refer to shown in the following table 1:
1 second cluster centre aggregate information table of table
In table 1, cluster centre number is the number of cluster centre in the second cluster centre set;SIFT description is second
Corresponding SIFT description of cluster centre in cluster centre set;The difference of description is the difference of SIFT description in table 1;Difference
The number of cluster centre belonging to value is the volume of the cluster centre in the corresponding third cluster centre set of difference for describe in table 1 son
Number.
In practical applications, it only needs to store the second cluster centre set and third cluster centre set in image data base
In corresponding relationship between the number of each cluster centre, numerical value and each cluster centre.
Wherein, step S360~S370 can be used for realizing the step S240 in previous embodiment.
Image data base establishment process provided in this embodiment obtains reference video image;Extract the reference video figure
The SIFT of picture describes son;Clustering processing is carried out to SIFT description of the reference video image, obtains the second cluster centre collection
It closes;The difference between the first SIFT description and the corresponding cluster centre of the first SIFT description is calculated, as described the
The difference of one SIFT description;Clustering processing is carried out to the difference of SIFT description of the reference video image, obtains third
Cluster centre set;Each SIFT based on the reference video image describes son, establishes in the second cluster centre set
Corresponding relationship in each cluster centre and the third cluster centre set between each cluster centre;It is poly- according to described second
Class centralization, the third cluster centre set and the corresponding relationship, reference video image can sufficiently be covered by establishing
The image data base of middle characteristic information provides accurate, reliable database and supports for retrieval, the audit of target video image.
Referring to Fig. 4, Fig. 4 is another flow chart of information processing method provided in an embodiment of the present invention.
It in the present embodiment, is included at least in preset image data base: the second cluster centre set and third cluster centre
Set.
As shown in Figure 4, which comprises
S410: target video image is obtained.
Target video image refers to, any one video image extracted from pending video.
S420: SIFT description of the target video image is extracted.
S430: from the second cluster centre set, in determining nearest cluster at a distance from the 2nd SIFT description
The heart, as the first cluster centre.
Wherein, it is that any SIFT in SIFT description of the target video image is retouched that the 2nd SIFT, which describes son,
State son.
For example, there are 1024 cluster centres in the first cluster centre set, the 2nd SIFT description and this 1024 are calculated
The distance between cluster centre, will in 1024 cluster centres with the 2nd SIFT description apart from nearest cluster centre,
It is denoted as the first cluster centre.
S440: it from the third cluster centre set, determines in cluster corresponding with first cluster centre
The heart, as the second cluster centre.
Specifically, in the second cluster centre set each cluster centre with it is each in the third cluster centre set
There is corresponding relationship, step S440 is specifically, according in the second cluster centre set in each cluster between cluster centre
Corresponding relationship in the heart and the third cluster centre set between each cluster centre, from the third cluster centre set
In, cluster centre corresponding with first cluster centre is determined, as the second cluster centre.
For example, there are 256 cluster centres in third cluster centre set, determined from this 256 cluster centres and institute
The corresponding cluster centre of the first cluster centre is stated, the second cluster centre is denoted as.
S450: when the 2nd SIFT description and the maximum range value of second cluster centre are less than pre-determined distance threshold
When value, the 2nd SIFT is described into son and is determined as effective SIFT description.
Cluster centre corresponding with first cluster centre, may have one, it is also possible to have multiple, that is, described
Two cluster centres have at least one.
It calculates in the 2nd SIFT description and at least one second cluster centre between each second cluster centre
Distance, using the maximum value for the distance calculated as maximum range value, when the maximum range value is less than pre-determined distance threshold value
When, the 2nd SIFT is described into son and is determined as effective SIFT description.
From all SIFT of target video image description, all of the target video image can be determined
Effective SIFT describe son.
Correspondingly, the pre-determined distance condition is, it is less than pre-determined distance with the maximum range value of second cluster centre
Threshold value.
In one example, the pre-determined distance threshold value is preferably 0.5, naturally it is also possible to and it is other values, it can be according to specific requirements
Carry out flexible setting.
S460: the institute when the number of effective SIFT description of the target video image, with the target video image
There is the ratio between the number of SIFT description, when being greater than default fractional threshold, determination retrieves the target video image.
The preset number condition is the ratio between the number of all SIFT description of the target video image
Greater than default fractional threshold.
In one example, the default fractional threshold is preferably 50%, naturally it is also possible to and it is other values, it can be according to specific requirements
Carry out flexible setting.
Correspondingly, the number of effective SIFT description when the target video image, with the target video image
Ratio between the number of all SIFT description, when no more than default fractional threshold, determination does not retrieve the target video
Image.
When any video image of pending video is retrieved, determine that the pending video is retrieved;When to
When all video images of audit video are not retrieved, determine that the pending video is not retrieved, thus realization pair
The audit of pending video.
When reference video image is the image for carrying sensitive information, the pending video is retrieved characterization, institute
It states pending video and carries sensitive information;The pending video is not retrieved characterization, and the pending video does not carry
There is sensitive information.
Information processing method provided in this embodiment obtains target video image;Extract the target video image
SIFT description;From the second cluster centre set, determining nearest cluster centre at a distance from the 2nd SIFT description,
As the first cluster centre;From the third cluster centre set, determine corresponding with first cluster centre poly-
Class center, as the second cluster centre;When the 2nd SIFT description is small with the maximum range value of second cluster centre
When pre-determined distance threshold value, the 2nd SIFT is described into son and is determined as effective SIFT description;When the target video
Ratio between the number of effective SIFT description of image, and the number of all SIFT description of the target video image
Value, when being greater than default fractional threshold, determination retrieves the target video image.Information processing method of the invention, have compared with
High retrieval rate and retrieval precision, and then while realizing video audit automation, improve the review efficiency of video with
Precision is audited, artificial burden is alleviated.
The embodiment of the invention also provides information processing unit, the information processing unit is for implementing the embodiment of the present invention
The information processing method of offer, the technology contents of information processing unit described below can be with above-described information processing side
The technology contents of method with correspond to each other reference.
Referring to Fig. 5, Fig. 5 is a kind of structural schematic diagram of information processing unit provided in an embodiment of the present invention.
As shown in figure 5, the information processing unit includes:
Target image acquiring unit 510, for obtaining target video image.
Sub- extraction unit 520 is described, the SIFT for extracting the target video image describes son.
Sub- determination unit 530 is effectively described, for utilizing preset image data base, from the target video image
Effective SIFT description is determined in SIFT description.
It wherein, include: cluster centre corresponding with the SIFT of reference video image description in described image database;
Effective SIFT describes son and is, the SIFT of pre-determined distance condition is met at a distance from the cluster centre in described image database
Description.
As a result determination unit 540, the number for effective SIFT description when the target video image meet default
When number condition, determination retrieves the target video image.
Correspondingly, when the number of effective SIFT of target video image description does not meet preset number condition,
Determination does not retrieve the target video image.
Information processing unit provided in this embodiment obtains target video image;Extract the target video image
SIFT description;Using preset image data base, determined effectively from the SIFT of target video image description
SIFT description;When the number of effective SIFT of target video image description meets preset number condition, inspection is determined
Rope realizes the automation review process of video content to the target video image, and according to reference video image
SIFT is described in SIFT description and described image database of sub corresponding cluster centre and the target video image
The distance between cluster centre relationship improve examining for video content to determine whether to retrieve the target video image
Core precision and review efficiency, while saving a large amount of human cost.
Referring to Fig. 6, Fig. 6 is another structural schematic diagram of information processing unit provided in an embodiment of the present invention.
As shown in fig. 6, the information processing unit in addition to include previous embodiment in target image acquiring unit 510,
It describes sub- extraction unit 520, effectively describe outside sub- determination unit 530 and result determination unit 540, can also include: database
Establish unit 550.
The Database unit 550, for obtaining reference video image;Extract the SIFT of the reference video image
Description;SIFT description based on the reference video image carries out clustering processing, obtains cluster result;According to the cluster
As a result, establishing described image database.
In one example, the acquisition reference video image includes:
Obtain reference video;The reference video is decoded as at least one video image;To at least one described video
Image carries out clustering processing, obtains the first cluster centre set;Included at least in the first cluster centre set: it is described at least
The corresponding cluster centre of each video image in one video image;Each cluster from the first cluster centre set
In the corresponding video image in center, the video image for meeting default extracting rule is extracted, as the reference video figure
Picture.
In one example, SIFT description based on the reference video image carries out clustering processing, obtains cluster knot
Fruit includes:
Clustering processing is carried out to SIFT description of the reference video image, obtains the second cluster centre set;It is described
It is included at least in second cluster centre set: in the corresponding cluster of each SIFT description of the reference video image
The heart;
The difference between the first SIFT description cluster centre corresponding with the first SIFT description is calculated, as institute
State the difference of the first SIFT description;First SIFT describes son and is, in SIFT description of the reference video image
Any SIFT description;
Clustering processing is carried out to the difference of SIFT description of the reference video image, obtains third cluster centre collection
It closes;It is included at least in the third cluster centre set: the difference difference of each SIFT description of the reference video image
Corresponding cluster centre;
Wherein, the cluster result includes at least the second cluster centre set and the third cluster centre set.
In one example, described according to the cluster result, establishing described image database includes:
Each SIFT based on the reference video image describes son, establishes each poly- in the second cluster centre set
Corresponding relationship in class center and the third cluster centre set between each cluster centre;
According to the second cluster centre set, the third cluster centre set and the corresponding relationship, described in foundation
Image data base.
Wherein, described image database can be preset in the information processing unit, as the information processing unit
A part, can also be preset in other devices except the information processing unit, be not specifically limited herein.
It include: the second cluster centre set and third cluster centre set in one example, in described image database;It is described
Sub- determination unit 530 is effectively described to be specifically used for:
From the second cluster centre set, determining nearest cluster centre at a distance from the 2nd SIFT description is made
For the first cluster centre;Wherein, it is any in SIFT description of the target video image that the 2nd SIFT, which describes son,
SIFT description;
From the third cluster centre set, cluster centre corresponding with first cluster centre is determined, make
For the second cluster centre;
When the 2nd SIFT description and the maximum range value of second cluster centre are less than pre-determined distance threshold value,
2nd SIFT is described into son and is determined as effective SIFT description.
In one example, the result determination unit 540 is specifically used for:
All SIFT when the number of effective SIFT description of the target video image, with the target video image
Ratio between the number of son is described, when being greater than default fractional threshold, determination retrieves the target video image.
Correspondingly, the number of effective SIFT description when the target video image, with the target video image
Ratio between the number of all SIFT description, when no more than default fractional threshold, determination does not retrieve the target video
Image.
Image data base establishment process provided in this embodiment obtains reference video image;Extract the reference video figure
The SIFT of picture describes son;Clustering processing is carried out to SIFT description of the reference video image, obtains the second cluster centre collection
It closes;The difference between the first SIFT description and the corresponding cluster centre of the first SIFT description is calculated, as described the
The difference of one SIFT description;Clustering processing is carried out to the difference of SIFT description of the reference video image, obtains third
Cluster centre set;Each SIFT based on the reference video image describes son, establishes in the second cluster centre set
Corresponding relationship in each cluster centre and the third cluster centre set between each cluster centre;It is poly- according to described second
Class centralization, the third cluster centre set and the corresponding relationship, reference video image can sufficiently be covered by establishing
The image data base of middle characteristic information provides accurate, reliable database and supports for retrieval, the audit of target video image.
Information processing unit provided in an embodiment of the present invention, including processor and memory, above-mentioned target image obtain single
Member 510, effectively describes sub- determination unit 530, result determination unit 540 and Database unit at the sub- extraction unit 520 of description
550 it is equal as program unit storage in memory, by processor execute above procedure unit stored in memory Lai
Realize corresponding function.
Include kernel in processor, is gone in memory to transfer corresponding program unit by kernel.Kernel can be set one
Or more, to solve the review efficiency of current video content and the lower technical problem of precision is audited by adjusting kernel parameter.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/
Or the forms such as Nonvolatile memory, if read-only memory (ROM) or flash memory (flash RAM), memory include that at least one is deposited
Store up chip.
The embodiment of the invention provides a kind of storage mediums, are stored thereon with program, real when which is executed by processor
The step of existing information processing method above-mentioned.
The embodiment of the invention provides a kind of processor, the processor is for running program, wherein described program operation
The step of Shi Zhihang information processing method above-mentioned.
The embodiment of the invention provides a kind of equipment, equipment include processor, memory and storage on a memory and can
The step of program run on a processor, processor realizes information processing method above-mentioned when executing program.
Equipment herein can be server, PC, PAD, mobile phone etc..
Present invention also provides a kind of computer program products, when executing on data processing equipment, are adapted for carrying out just
Beginningization has the program of the step of information processing method above-mentioned.
Finally, it is to be noted that, herein, such as first and first or the like relational terms be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes that
A little elements, but also including other elements that are not explicitly listed, or further include for this process, method, article or
The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged
Except there is also other identical elements in the process, method, article or apparatus that includes the element.
Through the above description of the embodiments, those skilled in the art can be understood that the application can be used
The form of complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects is realized.Based on this
The understanding of sample, the technical solution of the application to background technique contribute in whole or in part can be in the form of software products
It embodies, which can store in storage medium, such as ROM/RAM, magnetic disk, CD, including several
Instruction is used so that a computer equipment (can be personal computer, server or the network equipment etc.) executes the application
Method described in certain parts of each embodiment or embodiment.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For device disclosed in embodiment
For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part
It is bright.
Specific examples are used herein to illustrate the principle and implementation manner of the present application, and above embodiments are said
It is bright to be merely used to help understand the present processes and its core concept;At the same time, for those skilled in the art, foundation
The thought of the application, there will be changes in the specific implementation manner and application range.In conclusion the content of the present specification is not
It is interpreted as the limitation to the application.
Claims (10)
1. a kind of information processing method, which is characterized in that the described method includes:
Obtain target video image;
Extract Scale invariant features transform SIFT description of the target video image;
Using preset image data base, effective SIFT description is determined from the SIFT of target video image description
Son;It wherein, include: cluster centre corresponding with the SIFT of reference video image description in described image database;It is described
It is the SIFT description for meeting pre-determined distance condition at a distance from the cluster centre in described image database that effective SIFT, which describes son,
Son;
When the number of effective SIFT of target video image description meets preset number condition, determine described in retrieving
Target video image.
2. the method as described in claim 1, which is characterized in that the establishment process of described image database includes:
Obtain reference video image;
Extract SIFT description of the reference video image;
SIFT description based on the reference video image carries out clustering processing, obtains cluster result;
According to the cluster result, described image database is established.
3. method according to claim 2, which is characterized in that the acquisition reference video image includes:
Obtain reference video;
The reference video is decoded as at least one video image;
Clustering processing is carried out at least one described video image, obtains the first cluster centre set;First cluster centre
It is included at least in set: the corresponding cluster centre of each video image at least one described video image;
From the corresponding video image of cluster centre each in the first cluster centre set, extracts and meet default mention
The video image for taking rule, as the reference video image.
4. method according to claim 2, which is characterized in that it is described based on the reference video image SIFT description son into
Row clustering processing, obtaining cluster result includes:
Clustering processing is carried out to SIFT description of the reference video image, obtains the second cluster centre set;Described second
It is included at least in cluster centre set: the corresponding cluster centre of each SIFT description of the reference video image;
The difference between the first SIFT description and the corresponding cluster centre of the first SIFT description is calculated, as described the
The difference of one SIFT description;First SIFT describes son and is, any in SIFT description of the reference video image
SIFT description;
Clustering processing is carried out to the difference of SIFT description of the reference video image, obtains third cluster centre set;Institute
State and include at least in third cluster centre set: the difference of each SIFT description of the reference video image is corresponding
Cluster centre;
Wherein, the cluster result includes at least the second cluster centre set and the third cluster centre set.
5. method as claimed in claim 4, which is characterized in that it is described according to the cluster result, establish described image data
Library includes:
Each SIFT based on the reference video image describes son, establishes in the second cluster centre set in each cluster
Corresponding relationship in the heart and the third cluster centre set between each cluster centre;
According to the second cluster centre set, the third cluster centre set and the corresponding relationship, described image is established
Database.
6. the method as described in claim 1, which is characterized in that include: the second cluster centre set in described image database
With third cluster centre set;It is described using preset image data base, from the SIFT of target video image description
Determine that effective SIFT describes son and includes:
From the second cluster centre set, it is determining it is sub with the 2nd SIFT description at a distance from nearest cluster centre, as the
One cluster centre;Wherein, it is any SIFT in SIFT description of the target video image that the 2nd SIFT, which describes son,
Description;
From the third cluster centre set, cluster centre corresponding with first cluster centre is determined, as
Two cluster centres;
When the 2nd SIFT description and the maximum range value of second cluster centre are less than pre-determined distance threshold value, by institute
It states the 2nd SIFT and describes son and be determined as effective SIFT description.
7. the method as described in claim 1, which is characterized in that described when effective SIFT of the target video image describes son
Number when meeting preset number condition, determine that retrieving the target video image includes:
When the number of effective SIFT description of the target video image, described with all SIFT of the target video image
Ratio between the number of son, when being greater than default fractional threshold, determination retrieves the target video image.
8. a kind of information processing unit, which is characterized in that described device includes:
Target image acquiring unit, for obtaining target video image;
Sub- extraction unit is described, the SIFT for extracting the target video image describes son;
Sub- determination unit is effectively described, for utilizing preset image data base, is described from the SIFT of the target video image
Effective SIFT description is determined in son;It wherein, include: to describe son with the SIFT of reference video image in described image database
Corresponding cluster centre;Effective SIFT describes son and is, meets at a distance from the cluster centre in described image database
The SIFT of pre-determined distance condition describes son;
As a result determination unit, the number for effective SIFT description when the target video image meet preset number condition
When, determination retrieves the target video image.
9. device as claimed in claim 8, which is characterized in that described device further include:
The Database unit, for obtaining reference video image;Extract SIFT description of the reference video image;
SIFT description based on the reference video image carries out clustering processing, obtains cluster result;According to the cluster result, build
Vertical described image database.
10. device as claimed in claim 8, which is characterized in that include: the second cluster centre set in described image database
With third cluster centre set;It is described effectively to describe sub- determination unit and be specifically used for:
From the second cluster centre set, it is determining it is sub with the 2nd SIFT description at a distance from nearest cluster centre, as the
One cluster centre;Wherein, it is any SIFT in SIFT description of the target video image that the 2nd SIFT, which describes son,
Description;
From the third cluster centre set, cluster centre corresponding with first cluster centre is determined, as
Two cluster centres;
When the 2nd SIFT description and the maximum range value of second cluster centre are less than pre-determined distance threshold value, by institute
It states the 2nd SIFT and describes son and be determined as effective SIFT description.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810993318.2A CN109086744A (en) | 2018-08-29 | 2018-08-29 | Information processing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810993318.2A CN109086744A (en) | 2018-08-29 | 2018-08-29 | Information processing method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109086744A true CN109086744A (en) | 2018-12-25 |
Family
ID=64795162
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810993318.2A Pending CN109086744A (en) | 2018-08-29 | 2018-08-29 | Information processing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109086744A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102945289A (en) * | 2012-11-30 | 2013-02-27 | 苏州搜客信息技术有限公司 | Image search method based on CGCI-SIFT (consistence index-scale invariant feature transform) partial feature |
CN103902704A (en) * | 2014-03-31 | 2014-07-02 | 华中科技大学 | Multi-dimensional inverted index and quick retrieval algorithm for large-scale image visual features |
CN105183845A (en) * | 2015-09-06 | 2015-12-23 | 华中科技大学 | ERVQ image indexing and retrieval method in combination with semantic features |
-
2018
- 2018-08-29 CN CN201810993318.2A patent/CN109086744A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102945289A (en) * | 2012-11-30 | 2013-02-27 | 苏州搜客信息技术有限公司 | Image search method based on CGCI-SIFT (consistence index-scale invariant feature transform) partial feature |
CN103902704A (en) * | 2014-03-31 | 2014-07-02 | 华中科技大学 | Multi-dimensional inverted index and quick retrieval algorithm for large-scale image visual features |
CN105183845A (en) * | 2015-09-06 | 2015-12-23 | 华中科技大学 | ERVQ image indexing and retrieval method in combination with semantic features |
Non-Patent Citations (2)
Title |
---|
栾悉道 等著: "《多媒体情报处理技术》", 31 May 2016, 国防工业出版社 * |
艾列富: "基于内容的大规模图像索与检索方法研究", 《中国博士学位论文全文数据库信息科技辑》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10163227B1 (en) | Image file compression using dummy data for non-salient portions of images | |
Kwon et al. | Learning jpeg compression artifacts for image manipulation detection and localization | |
AU2015274708B2 (en) | Rule-based video importance analysis | |
RU2608261C2 (en) | Automatic tag generation based on image content | |
JP5934653B2 (en) | Image classification device, image classification method, program, recording medium, integrated circuit, model creation device | |
CN104679818B (en) | A kind of video key frame extracting method and system | |
JP4139615B2 (en) | Event clustering of images using foreground / background segmentation | |
US20120269425A1 (en) | Predicting the aesthetic value of an image | |
WO2017000716A2 (en) | Image management method and device, and terminal device | |
CN104063706B (en) | Video fingerprint extraction method based on SURF algorithm | |
CN109862394A (en) | Checking method, device, equipment and the storage medium of video content | |
CN108647245B (en) | Multimedia resource matching method and device, storage medium and electronic device | |
WO2017000109A1 (en) | Search method, search apparatus, user equipment, and computer program product | |
US20160196478A1 (en) | Image processing method and device | |
Bianco et al. | Predicting image aesthetics with deep learning | |
CN109948734B (en) | Image clustering method and device and electronic equipment | |
CN109284613B (en) | Method, device, equipment and storage medium for identification detection and counterfeit site detection | |
CN110532413B (en) | Information retrieval method and device based on picture matching and computer equipment | |
WO2012114727A1 (en) | Image processing device and image processing method | |
CN111325271A (en) | Image classification method and device | |
CN110942456B (en) | Tamper image detection method, device, equipment and storage medium | |
WO2015027882A1 (en) | Method, apparatus and terminal for image processing | |
CN115424171A (en) | Flame and smoke detection method, device and storage medium | |
JP2014092955A (en) | Similar content search processing device, similar content search processing method and program | |
Feng et al. | A novel saliency detection method for wild animal monitoring images with WMSN |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181225 |