CN102819583B - Network animation method for quickly retrieving - Google Patents
Network animation method for quickly retrieving Download PDFInfo
- Publication number
- CN102819583B CN102819583B CN201210262536.1A CN201210262536A CN102819583B CN 102819583 B CN102819583 B CN 102819583B CN 201210262536 A CN201210262536 A CN 201210262536A CN 102819583 B CN102819583 B CN 102819583B
- Authority
- CN
- China
- Prior art keywords
- network
- animation
- color
- network animation
- image
- 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
Landscapes
- Processing Or Creating Images (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses network animation method for quickly retrieving, the method effectively utilizes architectural feature and the content characteristic of network animation, realizing quick-searching, it comprises the following steps: (1), based on network animation analytical technology, extracts the architectural feature of network animation to be retrieved;(2) network animated construction feature to be retrieved being compared with the architectural feature of storage in storehouse, if finding same characteristic features, then differentiating that the two is identical network animation;(3) network animation is decomposed into some images, utilizes the method combined based on color entropy and color histogram to extract network animated content feature;(4) comparing cell animated content feature, calculates similarity, if similarity is more than certain threshold value, then differentiates that the two is similar network animation.It is demonstrated experimentally that the present invention can realize the retrieval for network animation accurately and efficiently.
Description
Technical field
The present invention relates to network animation based on content retrieval, web advertisement monitoring field, particularly relate to one
Network animation method for quickly retrieving.
Background technology
Network animation be a kind of based on vector graphics, there is interactive function and supported by numerous software platforms
Multimedia form, has been widely applied to video display, the web advertisement, music MTV, cartoon, game, class
The field such as part and cell phone multimedia message.Particularly in terms of the web advertisement, the overwhelming majority of each big main stream website is important
Advertisement is all using network animation as carrier.Flourish, in network animation along with network animation
The demand hold retrieval, monitoring is the most growing.
Summary of the invention
For in prior art, the shortcomings such as network animation retrieval rate is slow, retrieval is inconvenient, the present invention proposes
A kind of method for quickly retrieving that network animated construction feature is combined with content characteristic.
In order to solve above-mentioned technical problem, technical scheme is as follows:
Network animation method for quickly retrieving, comprises the steps:
11) architectural feature of network animation is extracted;
12) calculate the color entropy of network animation and color histogram thus obtain network animated content feature;
13) according to the content characteristic of network animation, the similarity between network animation is calculated, according to the value of similarity
Judge that network animation is the most similar.
Further, comprise the steps: that described step 11) comprises the steps:
21) the frame number frame_num in network animation, shape number shape_num, other number of tags are extracted
Other_num, by the architectural feature of 3 network consisting animations:
flashStructFeature {
int frame_num;
int shape_num;
int other_num;};
22) utilize Hash lookup algorithm that network animated construction feature existing in storehouse is made a look up, if searching successfully,
Then think that the two is just the same, the network animation information that in storehouse, the match is successful is exported to similar network animation collection
Close.
Further, comprise the steps: that described step 12) comprises the steps:
31) network animation is decomposed into some shape image, calculates the color entropy of each shape image, get colors
The m pair shape image that entropy is maximum more than certain threshold value and color entropy, using this m sub-picture as this network animation
Characteristic image;
Described color entropy computing formula:
Wherein, PiRepresent that color i in the percentage ratio shared by entire image, i.e. image of the pixel with color i occurs
Probability, m represents the number of the color being quantized in image;
32) described m pair characteristic image is transformed into hsv color space from RGB color, then to HSV
Three passage H of color space, S, V uniform quantization respectively is 18,3,3 intervals, the face altogether quantified
Chromatic number amount is 18*3*3=162 kind, is calculated the color histogram of m pair characteristic image, and is normalized
Process, obtain the content characteristic of network animation.
Further, comprise the steps: that described step 13) comprises the steps:
41) content characteristic that the color entropy in network animation to be compared approximates is selected;
42) histogram intersection method is utilized to calculate the distance between content characteristic;
Two common factor distances d between rectangular histogram f and g (f, g) is calculated as follows:
Wherein, H, S, V represent that color histogram is in H passage, the quantization district of channel S and V passage respectively
Between, described f and g represents the color histogram of two width images of color entropy approximation in network animation respectively;
43) if (f, g) less than certain threshold value, then it is assumed that network animation is similar for distance d between content characteristic.
The present invention has the beneficial effects that: effectively utilize architectural feature and the content characteristic of network animation, it is achieved quickly examine
Rope, it comprises the following steps: (1), based on network animation analytical technology, extracts network animation to be retrieved
Architectural feature;(2) network animated construction feature to be retrieved is compared with the architectural feature of storage in storehouse,
If discovery same characteristic features, then differentiate that the two is identical network animation;(3) network animation is decomposed into some
Image, utilizes the method combined based on color entropy and color histogram to extract network animated content feature;(4)
Comparing cell animated content feature, calculates similarity, if similarity is more than certain threshold value, then differentiates that the two is
Similar network animation.It is demonstrated experimentally that the present invention can realize the inspection for network animation accurately and efficiently
Rope.The technical field of the application of the present invention includes the necks such as the retrieval of network animated content, network animated content monitoring
Territory.
Accompanying drawing explanation
Fig. 1 is the network animation retrieval flow figure of the present invention;
Fig. 2 is Content Feature Extraction and the comparative approach flow chart of the network animation of the present invention.
Detailed description of the invention
Below in conjunction with the drawings and specific embodiments, the present invention is described further.
As Fig. 1 illustrates the network animation method for quickly retrieving flow chart that the present invention proposes, first the method is analyzed
Network animation internal structure to be retrieved, extracts its architectural feature, and utilizes Hash lookup algorithm to existing in storehouse
Network animated construction feature make a look up, if searching successfully, then it is assumed that the two is just the same, by storehouse
Joining successful network animation information to export to similar network animation collections, this step can achieve quick-searching,
Large quantities of unmatched network animation can be filtered out, but this step is prioritization scheme, it should be noted that
Even if not carrying out this step, the present invention also can be realized.Next the content characteristic of network animation to be retrieved is extracted,
And color combining entropy and color histogram map distance calculate this content characteristic and have network animated content feature in storehouse
Similarity degree, by similarity degree more than certain threshold value network animation information export to similar network animation collection
Closing, return similar network animation collections, to user, is finally completed the work of network animation quick-searching.
The present invention mainly has following two parts: one is the architectural feature extraction of network animation;Two is that network moves
Draw Content Feature Extraction and compare.
Two parts of the present invention will be described in detail respectively below.
1. the architectural feature of network animation is extracted
The overall structure of network animation file is by file header, document body and end of file label three part group
Become.Wherein, document body is made up of some object tag, and each object tag is too by label
Head, label body and label end mark composition.Through analyzing and experimental verification, the frame tagging in network animation
Number, shape number of tags and other number of tags combine can be as differentiating that network animation is whether as
Important evidence.Therefore, the present invention extracts the frame number frame_num in network animation, shape number shape_num,
Other number of tags other_num, by the architectural feature of 3 network consisting animations:
flashStructFeature {
int frame_num;
int shape_num;
int other_num;}
2. the Content Feature Extraction of network animation and Similarity measures
The present invention proposes as shown in Figure 2 the Content Feature Extraction of network animation and comparative approach flow chart.One
Individual complete network animation file can resolve from time domain and two aspects of spatial domain.Network animation literary composition
Part basic composition unit in time domain is frame, and the basic composition unit in spatial domain is then object mark
Sign.The analysis found that network animation has a kind of important object tag: shape (Shape), it is group
Become the important basic element of network animation, and save the most of important informations in network animation.Generally
In the case of, about a ten seconds network animation file at least has frame up to a hundred, the most only comprises about 20
Shape.Therefore, network animation is decomposed into shape, and the content characteristic of extraction network animation from shape,
Recall precision can be significantly increased.
For further speeding up retrieval, available color entropy finds out the most representational shape image in network animation.
Color entropy represents the quantity of information of color in image, and color entropy is the biggest, and the colouring information amount in image is the biggest,
The image contained much information more can effectively reflect the feature of network animation.
Based on above-mentioned, first network animation is decomposed into some shape image by the present invention, calculates each shape graph
The color entropy of picture, the m pair that the entropy that gets colors is maximum more than certain threshold value (general value is 3) and color entropy
Shape image, using this m sub-picture as the characteristic image of this network animation.
Color entropy computing formula:
Wherein, PiRepresent pixel color i in the percentage ratio shared by entire image, i.e. image with color i
The probability occurred, m represents the number of the color being quantized in image.
Then, m pair characteristic image is transformed into hsv color space from RGB color, then to HSV
Color space carries out uniform color quantization, is quantized into 162 kinds of colors.Specifically, hsv color space
Three passage H, S, V uniform quantization respectively is 18,3,3 intervals, the number of colors the most altogether quantified
For 18*3*3=162 kind.It is calculated the color histogram of m pair characteristic image, and is normalized,
Content characteristic as network animation;It follows that select between the color entropy in network animation to be compared poor
Away from the content characteristic less than threshold value (general value 0.8), histogram intersection distance is utilized to calculate therebetween
Distance;
If (h, s, v) (h, s v) represent the color histogram of two width images, then two rectangular histograms f and g to f respectively with g
Between common factor distance d (f, g) is defined as follows:
Wherein, H, S, V represent that color histogram is in H passage, the quantization district of channel S and V passage respectively
Between.
Finally, if (f g) less than certain threshold value (general value 0.5), then recognizes distance d between content characteristic
It is similar for network animation.
The above is only the preferred embodiment of the present invention, it is noted that common for the art
Technical staff, without departing from the inventive concept of the premise, it is also possible to make some improvements and modifications, these
Improvements and modifications also should be regarded as in scope.
Claims (5)
1. network animation method for quickly retrieving, it is characterised in that comprise the steps:
11) architectural feature of network animation file is extracted, including time domain characteristic and space domain characteristic;
12) network animation being decomposed into some shape image, calculate the color entropy of each shape image, the entropy that gets colors is more than certain threshold
Value and the maximum m pair shape image of color entropy, using this m sub-picture as the characteristic image of this network animation, described m is secondary
Characteristic image is transformed into hsv color space from RGB color, is calculated the color histogram of m pair characteristic image, and
It is normalized, thus obtains network animated content feature;
13) according to the content characteristic of network animation, utilize histogram intersection method to calculate the distance between content characteristic, calculate network animation
Between similarity, judge that network animation is the most similar according to the value of similarity.
Network animation method for quickly retrieving the most according to claim 1, it is characterised in that comprise the steps: described step 11)
Comprise the steps:
21) extracting frame number frame_num, the shape number shape_num in network animation, other number of tags other_num, by 3
The architectural feature of network consisting animation:
flashStructFeature{
int frame_num;
int shape_num;
int other_num;};
22) Hash lookup algorithm is utilized to make a look up, network animated construction feature existing in storehouse if searching successfully, then it is assumed that the two
Just the same, the network animation information that in storehouse, the match is successful is exported to similar network animation collections.
Network animation method for quickly retrieving the most according to claim 1, it is characterised in that
Described color entropy computing formula:
Wherein, PiRepresent the probability that color i occurs in the percentage ratio shared by entire image, i.e. image of the pixel with color i, m table
The number of the color being quantized in diagram picture.
Network animation method for quickly retrieving the most according to claim 1, it is characterised in that described step 12) comprise the steps:
Three passage H to hsv color space, S, V uniform quantization respectively is 18,3,3 intervals, the number of colours altogether quantified
Amount is 18*3*3=162 kind, is calculated the color histogram of m pair characteristic image.
Network animation method for quickly retrieving the most according to claim 1, it is characterised in that comprise the steps:
Described step 13) comprise the steps:
41) content characteristic that the color entropy in network animation to be compared approximates is selected;
42) histogram intersection method is utilized to calculate the distance between content characteristic;
Two common factor distances d between rectangular histogram f and g (f, g) is calculated as follows:
Wherein, H, S, V represent that color histogram divides at H passage, channel S and the quantized interval of V passage, described f and g respectively
Biao Shi the color histogram of two width images of color entropy approximation in network animation;
43) if (f, g) less than certain threshold value, then it is assumed that network animation is similar for distance d between content characteristic.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210262536.1A CN102819583B (en) | 2012-07-26 | 2012-07-26 | Network animation method for quickly retrieving |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210262536.1A CN102819583B (en) | 2012-07-26 | 2012-07-26 | Network animation method for quickly retrieving |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102819583A CN102819583A (en) | 2012-12-12 |
CN102819583B true CN102819583B (en) | 2016-08-24 |
Family
ID=47303694
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210262536.1A Active CN102819583B (en) | 2012-07-26 | 2012-07-26 | Network animation method for quickly retrieving |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102819583B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9514560B2 (en) * | 2013-03-15 | 2016-12-06 | Dreamworks Animation Llc | Arbitrary hierarchical tagging of computer-generated animation assets |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1728161A (en) * | 2005-07-28 | 2006-02-01 | 上海交通大学 | Method for filtering sensing images based on heteropic quantized color feature vectors |
CN101276417A (en) * | 2008-04-17 | 2008-10-01 | 上海交通大学 | Method for filtering internet cartoon medium rubbish information based on content |
CN102254006A (en) * | 2011-07-15 | 2011-11-23 | 上海交通大学 | Method for retrieving Internet video based on contents |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH1011454A (en) * | 1996-06-20 | 1998-01-16 | Nippon Telegr & Teleph Corp <Ntt> | Video subject matter retrieval method and device therefor |
-
2012
- 2012-07-26 CN CN201210262536.1A patent/CN102819583B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1728161A (en) * | 2005-07-28 | 2006-02-01 | 上海交通大学 | Method for filtering sensing images based on heteropic quantized color feature vectors |
CN101276417A (en) * | 2008-04-17 | 2008-10-01 | 上海交通大学 | Method for filtering internet cartoon medium rubbish information based on content |
CN102254006A (en) * | 2011-07-15 | 2011-11-23 | 上海交通大学 | Method for retrieving Internet video based on contents |
Non-Patent Citations (2)
Title |
---|
基于内容的视频检索关键技术的研究与实现;王忠;《中国优秀硕士学位论文全文数据库》;20120215;正文第2和4章 * |
基于颜色空间的图像特征提取的研究;孙滔;《中国优秀硕士学位论文全文数据库》;20061015;正文第2-3章 * |
Also Published As
Publication number | Publication date |
---|---|
CN102819583A (en) | 2012-12-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10324977B2 (en) | Searching method and apparatus | |
CN103347167B (en) | A kind of monitor video content based on segmentation describes method | |
CN100550038C (en) | Image content recognizing method and recognition system | |
CN104376105A (en) | Feature fusing system and method for low-level visual features and text description information of images in social media | |
CN105678309B (en) | A kind of image multi-tag dimensioning algorithm based on more example packet feature learnings | |
CN105138991A (en) | Video emotion identification method based on emotion significant feature integration | |
CN101374234A (en) | Method and apparatus for monitoring video copy base on content | |
CN102968637A (en) | Complicated background image and character division method | |
CN102184242A (en) | Cross-camera video abstract extracting method | |
CN102098449B (en) | A kind of method utilizing Mark Detection to carry out TV programme automatic inside segmentation | |
CN104636761A (en) | Image semantic annotation method based on hierarchical segmentation | |
CN113888754B (en) | Vehicle multi-attribute identification method based on radar vision fusion | |
CN104661037A (en) | Tampering detection method and system for compressed image quantization table | |
CN106228157A (en) | Coloured image word paragraph segmentation based on image recognition technology and recognition methods | |
CN104978565A (en) | Universal on-image text extraction method | |
CN108664951A (en) | Pedestrian's recognition methods again based on color name feature | |
CN102567738A (en) | Rapid detection method for pornographic videos based on Gaussian distribution | |
CN106203448A (en) | A kind of scene classification method based on Nonlinear Scale Space Theory | |
CN102819583B (en) | Network animation method for quickly retrieving | |
Zheng et al. | Study on image retrieval based on image texture and color statistical projection | |
CN108986103A (en) | A kind of image partition method merged based on super-pixel and more hypergraphs | |
CN108304588A (en) | A kind of image search method and system based on k neighbours and Fuzzy Pattern Recognition | |
CN107357834A (en) | A kind of image search method of view-based access control model conspicuousness fusion | |
CN115063800B (en) | Text recognition method and electronic equipment | |
Ritter et al. | An extensible tool for the annotation of videos using segmentation and tracking |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |