CN101196994A - Image content recognizing method and recognition system - Google Patents
Image content recognizing method and recognition system Download PDFInfo
- Publication number
- CN101196994A CN101196994A CNA2007103042044A CN200710304204A CN101196994A CN 101196994 A CN101196994 A CN 101196994A CN A2007103042044 A CNA2007103042044 A CN A2007103042044A CN 200710304204 A CN200710304204 A CN 200710304204A CN 101196994 A CN101196994 A CN 101196994A
- Authority
- CN
- China
- Prior art keywords
- image content
- primary image
- similarity
- picture
- key parameter
- 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.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The present invention discloses a picture content identification method and system, wherein the method comprises the following steps that basic image content examples and corresponding marking information are stored in an example base beforehand. During identification, one or more than one item of basic image content is split from a to-be-identified picture by a split module; the split basic image content is compared with the basic image content examples in the example base by a similarity comparison module, obtaining corresponding similarity; a result output module determines the basic image content example having the highest similarity with each item of the basic image content, and outputs the marking information corresponding to the basic image content examples as a picture content identification result. The present invention can decrease the computation cost of computers and the occupancy to system resources, and can enlarge the species of recognizable content.
Description
Technical field
The present invention relates to Computerized Information Processing Tech, relate in particular to a kind of recognition system and recognition methods that computing machine picture content is discerned automatically.
Background technology
In the present computer internet authentication control system, often adopt the picture validation code technology to realize client identity checking and frequency of utilization control.Fig. 1 is a kind of interface synoptic diagram of realizing the authentication login by picture validation code.Referring to Fig. 1, the user not only needs to input correct account number 101 and password 102, but also needs correctly to read identifying code character 103 from described identifying code picture 100, and is input to system, could and sign in to system by checking.
In test process to the identifying code system, in order to assess the risk of identifying code system, need a kind ofly from the identifying code picture, to identify verification code information automatically and to be input to test macro in the system, with the input behavior of automatic simulation real user, and carry out risk assessment according to the whole identifying code of the reaction pair of identifying code system system.In this test macro, at first need computing machine automatically the content in the identifying code picture to be carried out discriminance analysis, identify identifying code character wherein, and then carry out subsequent treatment.
Present identifying code image content recognition technology generally is to adopt the content recognition extractive technique, for example literal identification (OCR) mode.Its major programme is to comprise an image content recognition unit, the employing intelligent algorithm is realized, need carry out learning training to the literary style of every kind of literal as hand-written discrimination system common in the mobile phone, obtain applicable model of cognition according to large-scale training set, when identification, need utilize intelligent algorithm according to features such as the stroke of distinguishing out each literal, the order of strokes observed in calligraphy, company's literary styles, provide candidate collection, obtain identifier word output after the hand picking, the entire process process need carries out the calculating of large amount of complex.In addition, when having interfere informations such as wave, noise in the identifying code picture, also need to comprise an auxiliary noise elimination unit, be used to remove interfere informations such as wave, noise, the image content of eliminating after handling through noise is cleaner, and the image content recognition unit is easier to identify correct Word message.
But present this content recognition extractive technique has following shortcoming:
1) described image content recognition unit needs comprehensive multiple statistic algorithm and intelligent algorithm to realize that calculation processes complexity, calculated amount are very huge, need expend a large amount of computer CPU computing expense and system resource.
2) can only from picture, identify specific literal, for the bigger difficulty of identification existence of image information.For example, if also comprise the pattern that some is specific in the picture, for example comprise the pattern of a sheep, then corresponding identifying code input may be " sheep " this word, and for this situation, existing recognition technology can't be discerned.
Summary of the invention
In view of this, technical matters to be solved by this invention is to provide a kind of image content recognizing method, with the computing cost that reduces computing machine with to the taking of system resource, and can enlarge discernible content type.
Another technical matters to be solved by this invention is to provide a kind of image content recognition system, with the computing cost that reduces computing machine with to the taking of system resource, and can enlarge discernible content type.
In order to realize the foregoing invention purpose, main technical schemes of the present invention is:
A kind of image content recognizing method, this method are stored primary image content instance and corresponding markup information thereof in advance;
When discerning, comprising:
A, from picture to be identified, split out one or more primary image contents;
B, the primary image content that splits out and the primary image content instance in the described case library are compared, obtain corresponding similarity;
C, determine and the highest primary image content instance of described each primary image content similarity, and the markup information of described primary image content instance correspondence is exported as the image content recognition result.
Preferably, the concrete grammar of the markup information of described storage primary image content instance and correspondence thereof is:
The example picture is split, therefrom split out one or more primary image content instances and storage respectively;
Be described each primary image content instance mark corresponding markup information and storage.
Preferably, described example picture is the identifying code example picture in the identifying code system.
Preferably, among the described step B, determine that the concrete grammar of similarity between the primary image content instance of a primary image content and contrast thereof is:
The key parameter of the primary image content instance of B1, the described primary image content of extraction and contrast, the similarity index of relatively more definite both sides' key parameter;
B2, determine the similarity between the primary image content instance of described primary image content and contrast thereof according to the similarity index of described key parameter.
Preferably, further comprise before the described step B1: the primary image content instance to described primary image content and/or its contrast carries out normalized, unified both sides' form index.
Preferably, the key parameter of described step B1 is one; Be specially among the described step B2: with the similarity index of this key parameter as the similarity between the primary image content instance of described primary image content and contrast thereof.
Preferably, the key parameter of described step B1 is more than one; Be specially among the described step B2: the similarity index to all key parameters between the primary image content instance of described primary image content and contrast thereof is weighted calculating, with result of calculation as the final similarity of both sides.
Preferably, described key parameter comprises following any key parameter or plants the combination of key parameter arbitrarily:
Original picture size, original color matrix or gray matrix, through color matrices or gray matrix, the distributed intelligence of picture lightness, picture special-effect statistical parameter and the original image form and the embedded information of conversion.
Preferably, this method is further added up the probability of occurrence of the primary image content instance in the described case library; And after step B obtains the similarity of described each primary image content and its contrast primary image content instance, further the probability of occurrence according to each contrast primary image content instance is weighted calculating to relevant similarity, with the definite foundation of the similarity after the weighted calculation as step C.
A kind of image content recognition system comprises:
Case library is used to store primary image content instance and corresponding markup information thereof;
First splits module, is used for splitting out one or more primary image contents from picture to be identified;
The similarity comparison module is used for splitting primary image content that module splits out and the primary image content instance of described case library compares with first, obtains corresponding similarity;
Output module as a result, be used for comparative result according to described picture analogies degree comparison module, select and the highest primary image content instance of described each primary image content similarity, and the markup information of described primary image content instance correspondence is exported as the image content recognition result.
Preferably, described system further comprises:
Second splits module, is used for splitting out one or more primary image content instances from the example picture, and described primary image example is deposited in the described case library;
Labeling module is used to provide the mark interface, receives the markup information at each primary image content instance of user's input by the mark interface, and the markup information of described primary image content correspondence is deposited in the described case library.
Preferably, described similarity comparison module specifically comprises:
Key parameter similarity index determination module is used for the primary image content instance at a primary image content and contrast thereof, extracts both sides' key parameter, relatively determines the similarity index of described key parameter;
The similarity determination module is used for determining similarity between the primary image content instance of described primary image content and contrast thereof according to the similarity index of described key parameter.
Preferably, described similarity comparison module further comprises: the normalization module is used for the described first primary image content that splits the module fractionation is normalized to and the corresponding to primary image content of the example format index of described case library.
Preferably, described key parameter similarity index determination module comprises following any module or plants the combination of module arbitrarily:
Be used for determining original picture size similarity index cover half piece really;
Be used for determining original color matrix or gray matrix similarity index cover half piece really;
Be used for determining through the color matrices of conversion or gray matrix similarity index cover half piece really;
Be used for determining picture lightness similarity index cover half piece really;
Be used for determining picture special-effect similarity index cover half piece really;
Be used for determining original image form and embedded information similarity index cover half piece really.
With respect to prior art, the present invention adopts and discerns image content based on the similarity manner of comparison of picture example, not needing to utilize intelligent algorithm to carry out contents extraction calculates, have the characteristics simple, that renewal is simple, computational complexity is low of using, can reduce the computing cost of computer CPU, and reduce taking resource for computer system.Simultaneously, the present invention is primary image content and markup information thereof owing to what collect storage, described primary image content can be a word content, it also can be graphical content, what export is the markup information of primary image content instance, therefore both can identify the word content in the picture, can identify the graphical content in the picture again, can enlarge discernible content type.
When the present invention is applied to the identification of identifying code picture, can collect identifying code picture example in advance, splitting into the primary image content instance rower of going forward side by side annotates, the case library of forming an identifying code picture, because the picture validation code system generally all uses the character set of limited number, as monogram, combination of numbers, the identifying code picture is formed in Chinese character combinations etc., therefore the present invention can more easily collect the set of a primary image content instance such as literal that often occurs, thereby be easy to build environment-identification at some picture validation codes system, and carry out to finish at an easy rate after identifying code upgrades the renewal of corresponding case library in the picture validation code system, the required artificial and computer resource expense of the maintenance of whole recognition system is all less, does not need the professional to operate.
Description of drawings
Fig. 1 is a kind of interface synoptic diagram of realizing the authentication login by picture validation code;
Fig. 2 is a process flow diagram of collecting storage fundamental figure content instance and corresponding markup information thereof in case library of the present invention;
Fig. 3 is the synoptic diagram that an identifying code example picture is split into a plurality of primary image content instances of the present invention;
Fig. 4 is the process flow diagram of a kind of specific embodiment that image content is discerned of the present invention;
Fig. 5 is a kind of primary structure synoptic diagram of image content recognition system of the present invention;
Fig. 6 is for comprising a kind of structural representation of example collection subsystem in the image content recognition system of the present invention;
Fig. 7 is the structural representation of similarity comparison module of the present invention.
Embodiment
Below by specific embodiments and the drawings the present invention is described in further details.
It is that example describes that following examples are applied in the identifying code picture recognition with image content recognition system of the present invention.
Method of the present invention need set in advance case library, wherein collects storage primary image content instance and corresponding markup information thereof in advance.Fig. 2 is a process flow diagram of collecting storage fundamental figure content instance and corresponding markup information thereof in case library of the present invention.Referring to Fig. 2, this flow process specifically comprises:
The described split process of this step can utilize Boundary Recognition method commonly used to realize that for example described concrete method for splitting can be: the pixel count fixed length by picture splits picture, is a primary image content instance every 20 * 30 pixels for example; Perhaps cut apart according to method complete, the partial continuous background colour, when for example having the continuous background look of strip, then the gap according to word of vertically cutting apart and word splits.
All pictures in step 203, the described pending picture set that judges whether to finish dealing with, if, process ends then; Otherwise returning step 201 continues to handle.
By the described step of Fig. 2, can in case library, collect the storage primary image content instance that decomposed of a series of identifying code example picture and the set of markup information thereof, the picture to be identified that all are similar to these identifying code example pictures utilizes the present invention to discern.
Fig. 4 is the process flow diagram of a kind of specific embodiment that image content is discerned of the present invention.In the present embodiment, described picture to be identified is the identifying code picture that an identifying code system generates, and this flow process specifically comprises:
If described identifying code picture to be identified splits into an above primary image content, then need export the markup information of described correspondence according to the order of described primary image content in former identifying code picture.
In described step 402, the concrete grammar of determining similarity between the primary image content instance of a primary image content and contrast thereof comes down to adopt picture analogies degree comparison techniques, represent similarity degree with similarity index herein, promptly use [0,1] interval decimal to represent similarity degree.For described two given pictures, the i.e. primary image content instance (i.e. example picture that comprises substance) of a primary image content (picture that comprises substance that promptly splits out) and contrast thereof, the comparison procedure of similarity comprises the steps 421 to step 423:
Step 421, the primary image content instance of described primary image content and/or its contrast is carried out normalized, unified contrast both sides' form index.Promptly indexs such as both sides' picture size, COLOR COMPOSITION THROUGH DISTRIBUTION are carried out conversion, obtain the unified picture of technical indicator as pending picture.For example, specifying unified dimension of picture is 12 * 12 pixels, adopts 256 grades of gray-scale maps to represent, finishes both sides' level and vertical normalized such as rectification.If both sides' form index is unified, then skip this step 421.
Step 422, extract the key parameter of described primary image content and contrast primary image content instance thereof, obtain the similarity index of both sides' key parameter one by one.
Step 423, determine the bipartite similarity of described contrast according to the similarity index of described contrast both sides key parameter.Herein, if the key parameter of described contrast has only one, then with the similarity index of this key parameter as the similarity between the primary image content instance of described primary image content and contrast thereof.If the key parameter of described contrast is more than one, then need the similarity index of all key parameters between the primary image content instance of described primary image content and contrast thereof is weighted calculating, with result of calculation as the final similarity of both sides.
Described key parameter is the crucial comparative parameter that is used to calculate similarity, comprises the combination of one of following key parameter or following any kind of key parameter at least:
A) original picture size; B) original color matrix if adopt the gray scale picture, then is a gray matrix; C) through the color matrices or the gray matrix of conversion, wherein can discard some trifling information, for example adopt wavelet transformation scheduling algorithm or graphics compression algorithm scheduling algorithm that color matrices or gray matrix are carried out conversion process; D) picture lightness distributed intelligence; E) picture special-effect statistical parameter, for example be outstanding prospect literal, background colour often adopts the color that obvious difference is arranged with the prospect literal, identifies real Word message otherwise human eye is difficult, then can do distribution statistics to the color that obvious conflict is arranged, with statistics as key parameter; F) embedded information (meta-data) in original image form and the picture.
For above-mentioned every kind of key parameter, adopt existing ripe comparison algorithm can determine the similarity index of this kind key parameter of described contrast both sides, be specifically as follows:
For contrast both sides' original picture size, its similarity index concrete determine method can for:
(a) the original picture size comparison is made up of length and width comparison two parts, supposes that the length and width that 2 bands compare picture p1, p2 are respectively X1, Y1 and X2, Y2, and then big or small similarity index computing method commonly used are as follows:
SIM(p1，p2)＝W1*(X1+X2)/(2*max(X1，X2))+W2*(Y1-Y2)/(2*max(Y1，Y2))。
Wherein W1, W2 are the length and width weighted index, generally all are made as 0.5.
(b) for contrast both sides' original color matrix or gray matrix, this sentences 256 grades of gray matrixs is the algorithm that example illustrates its similarity index.Suppose that 2 bands gray matrix relatively all gets for being added up by the correspondence image of 20 * 30 sizes, then can regard a gray scale segment that has 20 * 30 square dot matrix to form as, the gray-scale value of each point has 256 grades of variations.Then the acquisition of similarity index can be adopted following way commonly used: the grey scale values that compares 2 matrix correspondence positions successively, if the diversity factor of grey scale values less than some assign thresholds (as 200, difference between 205 is 5), then the similar counting of correspondence position is designated as 1, after traversal is complete, adds up similar counting and can obtain similarity index.
(c) for contrast both sides' color, gray matrix through conversion, the purpose of conversion is to reduce the size of matrix, and as the matrix of direct comparison 100000 * 10000000 sizes, computing cost is obviously higher.At first adopt Flame Image Process and artificial intelligence scheduling algorithm herein, some trifling information are abandoned, then can obtain the less relatively matrix of information loss, to obtain size after Wavelet Transformation Algorithm or other image compression algorithm conversion be 256 * 256 matrix as adopting.Similarity calculating method after the conversion is identical with the method for above-mentioned (b).
(d) for the lightness distributed intelligence of the picture that contrasts both sides, the lightness distributed intelligence can be stored in the matrix, can adopt the computing method of similar color matrices to obtain similarity index.
(e), generally be used for assisting relatively the comparative parameter of non-certain necessity for contrast both sides' special-effect statistical parameter.Here come for example with conflict color counting.So-called color conflict, when being meant visual inspection, the color of the existing significantly neighborhood pixels of conflict of color table is easier to distinguish as adjacent the putting together of reddish blue to counting, and pink and red placed adjacent then is not easy to distinguish.According to the conflict color that predefined is good statistics is gathered, scanned 2 images respectively, can obtain corresponding counting, can obtain corresponding similarity index thus.
(f) for contrast both sides' original image form, the picture format with same type is considered as unanimity exactly, and similarity index is 1; Dissimilar picture formats are considered as dissmilarity, and similarity index is 0, and for example bmp form and jpg form are dissimilar.The embedded information of picture is the optional content of picture, comprises as the embedded information (meta-data) of jpg type: the process software title and the version information of this picture; Date of formation; The picture copyright information; Camera parameter such as aperture, shutter data; Color space parameter or the like.For embedded information, can adopt the above-mentioned embedded information of contrast whether consistent, and with conforming probability as the embedded information similarity index.
Below by concrete computing formula described calculation of similarity degree process is described:
SIM(P1，P2)＝W1×S1+W2×S2+W3×S3+…+Wn×Sn
In the above-mentioned formula, the primary image content picture of described P1 for from picture to be identified, splitting out, P2 is a primary image content instance picture in the case library, SIM (P1, P2) be similarity between the P2 of P1, S1, S2 ..., Sn is the similarity index of different key parameter, W1, W2 ..., Wn is the weighted index of each key parameter correspondence, can adopt decimal or integer, realize deciding on concrete system.
Suppose now to use three kinds of critical datas, and set W1=W2=W3=1/3, S1 represents the original image size similarity degree, the COLOR COMPOSITION THROUGH DISTRIBUTION similarity degree that the S2 representative calculates based on color matrices, and the S3 representative has the counting of obvious vision conflict color.Wherein, the calculating of each similarity index can be adopted the algorithm in any image processing field, and unique requirement is that computing velocity is wanted to reach the minimum requirements that real-time is used, and for example finishes in some ms.For example, obtain S1=0.9 this moment, S2=0.8, and S3=0.9, then final SIM (P1, P2)=1/3*0.9+1/3*0.8+1/3*0.9=0.86 is the final similarity of 2 pictures.
Comparative approach by above-mentioned similarity, can compare described primary image content to be compared and all the example pictures in the described case library and draw corresponding similarity, sort according to the height of similarity markup information, get the identification content of the markup information of the example picture correspondence that wherein similarity is the highest as described primary image content correspondence to described example picture correspondence.If above-mentioned P2 is an example picture the most similar to described P1 in the case library, it is designated alphabetical A, and the probability that then can obtain P1 and be alphabetical A is 86%, and other identifications that split part by that analogy.
The practical application that provides an above-mentioned recognition methods is below given an example, for an identifying code picture that 5 literal (i.e. 5 primary image contents) are arranged, obtain 5 pictures to be compared by fractionation, to these 5 pictures one by one with case library in primary image content instance picture relatively, can obtain the markup information formation that 5 row are arranged according to similarity, each markup information formation is respectively according to similarity degree series arrangement from high in the end, and block or simply get preceding ten as pending set according to a certain similar threshold value, generally select similarity is the highest in the 5 row markup information formations markup information as final recognition result.Further, the present invention can also add up the probability of occurrence of the concrete identifying code of institute's application verification code system, and store this probability of occurrence, after drawing described 5 row markup information formations, can also do weighted calculation respectively to the similarity of each markup information correspondence according to the probability of occurrence of concrete identifying code, according to described 5 markup information formations being rearranged, take out the highest final recognition result of markup information conduct of similarity in described each mark formation again through the similarity after the weighted calculation.
Fig. 5 is a kind of primary structure synoptic diagram of image content recognition system of the present invention.Referring to Fig. 5, this image content recognition system comprises:
Collect storage primary image content instance for convenience from the example picture, referring to Fig. 6, also further comprise the example collection subsystem among a kind of embodiment of described system, this example collection subsystem comprises that specifically second splits module 505 and labeling module 506.
Described second splits module 505 is used for splitting out one or more primary image content instances from the example picture, and described primary image example is deposited in the described case library 501.
Described labeling module 506 is used to provide the mark interface, receives the markup information at each primary image content instance of user's input by the mark interface, and the markup information of described primary image content correspondence is deposited in the described case library 501.
Fig. 7 is the structural representation of described similarity comparison module.Referring to Fig. 7, primary image content and primary image content instance that similarity comparison module 503 is treated contrast carry out similarity analysis calculating, specifically comprise:
Key parameter similarity index determination module 702 is used for the primary image content instance at a primary image content and contrast thereof, extracts both sides' key parameter, relatively determines the similarity index of described key parameter.
Described key parameter is the crucial comparative parameter that is used to calculate similarity, and is corresponding with above-mentioned recognition methods, and described key parameter similarity index determination module can comprise following any module or plant the combination of module arbitrarily:
Be used for determining original picture size similarity index cover half piece really;
Be used for determining original color matrix or gray matrix similarity index cover half piece really;
Be used for determining through the color matrices of conversion or gray matrix similarity index cover half piece really;
Be used for determining picture lightness similarity index cover half piece really;
Be used for determining picture special-effect similarity index cover half piece really;
Be used for determining original image form and embedded information similarity index cover half piece really.
Certainly, described similarity determination module 703 can also adopt the similarity between the definite contrast of other key parameter picture, and this present invention is not limited.
The above; only for the preferable embodiment of the present invention, but protection scope of the present invention is not limited thereto, and anyly is familiar with the people of this technology in the disclosed technical scope of the present invention; the variation that can expect easily or replacement all should be encompassed within protection scope of the present invention.
Claims (14)
1. an image content recognizing method is characterized in that, this method is stored primary image content instance and corresponding markup information thereof in advance;
When discerning, comprising:
A, from picture to be identified, split out one or more primary image contents;
B, the primary image content that splits out and the primary image content instance in the described case library are compared, obtain corresponding similarity;
C, determine and the highest primary image content instance of described each primary image content similarity, and the markup information of described primary image content instance correspondence is exported as the image content recognition result.
2. image content recognizing method according to claim 1 is characterized in that, the concrete grammar of the markup information of described storage primary image content instance and correspondence thereof is:
The example picture is split, therefrom split out one or more primary image content instances and storage respectively;
Be described each primary image content instance mark corresponding markup information and storage.
3. image content recognizing method according to claim 2 is characterized in that, described example picture is the identifying code example picture in the identifying code system.
4. image content recognizing method according to claim 1 is characterized in that, among the described step B, determines that the concrete grammar of similarity between the primary image content instance of a primary image content and contrast thereof is:
The key parameter of the primary image content instance of B1, the described primary image content of extraction and contrast, the similarity index of relatively more definite both sides' key parameter;
B2, determine the similarity between the primary image content instance of described primary image content and contrast thereof according to the similarity index of described key parameter.
5. image content recognizing method according to claim 4 is characterized in that, further comprise before the described step B1: the primary image content instance to described primary image content and/or its contrast carries out normalized, unified both sides' form index.
6. image content recognizing method according to claim 4 is characterized in that, the key parameter of described step B1 is one; Be specially among the described step B2: with the similarity index of this key parameter as the similarity between the primary image content instance of described primary image content and contrast thereof.
7. image content recognizing method according to claim 4 is characterized in that, the key parameter of described step B1 is more than one; Be specially among the described step B2: the similarity index to all key parameters between the primary image content instance of described primary image content and contrast thereof is weighted calculating, with result of calculation as the final similarity of both sides.
8. image content recognizing method according to claim 4 is characterized in that, described key parameter comprises following any key parameter or plants the combination of key parameter arbitrarily:
Original picture size, original color matrix or gray matrix, through color matrices or gray matrix, the distributed intelligence of picture lightness, picture special-effect statistical parameter and the original image form and the embedded information of conversion.
9. image content recognizing method according to claim 1 is characterized in that, this method is further added up the probability of occurrence of the primary image content instance in the described case library; And after step B obtains the similarity of described each primary image content and its contrast primary image content instance, further the probability of occurrence according to each contrast primary image content instance is weighted calculating to relevant similarity, with the definite foundation of the similarity after the weighted calculation as step C.
10. an image content recognition system is characterized in that, comprising:
Case library is used to store primary image content instance and corresponding markup information thereof;
First splits module, is used for splitting out one or more primary image contents from picture to be identified;
The similarity comparison module is used for splitting primary image content that module splits out and the primary image content instance of described case library compares with first, obtains corresponding similarity;
Output module as a result, be used for comparative result according to described picture analogies degree comparison module, select and the highest primary image content instance of described each primary image content similarity, and the markup information of described primary image content instance correspondence is exported as the image content recognition result.
11. image content recognition system according to claim 10 is characterized in that, described system further comprises:
Second splits module, is used for splitting out one or more primary image content instances from the example picture, and described primary image example is deposited in the described case library;
Labeling module is used to provide the mark interface, receives the markup information at each primary image content instance of user's input by the mark interface, and the markup information of described primary image content correspondence is deposited in the described case library.
12. image content recognition system according to claim 10 is characterized in that, described similarity comparison module specifically comprises:
Key parameter similarity index determination module is used for the primary image content instance at a primary image content and contrast thereof, extracts both sides' key parameter, relatively determines the similarity index of described key parameter;
The similarity determination module is used for determining similarity between the primary image content instance of described primary image content and contrast thereof according to the similarity index of described key parameter.
13. image content recognition system according to claim 12, it is characterized in that, described similarity comparison module further comprises: the normalization module is used for the described first primary image content that splits the module fractionation is normalized to and the corresponding to primary image content of the example format index of described case library.
14. image content recognition system according to claim 12 is characterized in that, described key parameter similarity index determination module comprises following any module or plants the combination of module arbitrarily:
Be used for determining original picture size similarity index cover half piece really;
Be used for determining original color matrix or gray matrix similarity index cover half piece really;
Be used for determining through the color matrices of conversion or gray matrix similarity index cover half piece really;
Be used for determining picture lightness similarity index cover half piece really;
Be used for determining picture special-effect similarity index cover half piece really;
Be used for determining original image form and embedded information similarity index cover half piece really.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007103042044A CN100550038C (en) | 2007-12-26 | 2007-12-26 | Image content recognizing method and recognition system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007103042044A CN100550038C (en) | 2007-12-26 | 2007-12-26 | Image content recognizing method and recognition system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101196994A true CN101196994A (en) | 2008-06-11 |
| CN100550038C CN100550038C (en) | 2009-10-14 |
Family
ID=39547386
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2007103042044A Active CN100550038C (en) | 2007-12-26 | 2007-12-26 | Image content recognizing method and recognition system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100550038C (en) |
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101635763A (en) * | 2008-07-23 | 2010-01-27 | 深圳富泰宏精密工业有限公司 | Picture classification system and method |
| CN102467653A (en) * | 2010-10-29 | 2012-05-23 | 方正国际软件(北京)有限公司 | Image-text recognition method and system thereof |
| CN102867025A (en) * | 2012-08-23 | 2013-01-09 | 百度在线网络技术（北京）有限公司 | Method and device for acquiring picture marking data |
| CN102890761A (en) * | 2011-08-24 | 2013-01-23 | 北京文海思创科技有限公司 | Method for verifying through graphical verification code |
| CN103186781A (en) * | 2011-12-31 | 2013-07-03 | 北京新媒传信科技有限公司 | Text recognition method |
| CN103258280A (en) * | 2012-02-17 | 2013-08-21 | 盛趣信息技术（上海）有限公司 | Price comparative method and system |
| CN103426191A (en) * | 2012-05-26 | 2013-12-04 | 百度在线网络技术（北京）有限公司 | Method and system for picture marking |
| CN101859368B (en) * | 2009-04-09 | 2013-12-04 | 普诚科技股份有限公司 | Image identification device and method |
| CN103428515A (en) * | 2012-05-15 | 2013-12-04 | 索尼公司 | Video format determination device, video format determination method, and video display device |
| CN103914996A (en) * | 2014-04-24 | 2014-07-09 | 广东小天才科技有限公司 | Method and device for acquiring character learning materials from picture |
| CN104200204A (en) * | 2014-09-02 | 2014-12-10 | 福建富士通信息软件有限公司 | Picture processing device and method |
| CN104252446A (en) * | 2013-06-27 | 2014-12-31 | 鸿富锦精密工业（深圳）有限公司 | Computing device, and verification system and method for consistency of contents of files |
| CN104462152A (en) * | 2013-09-23 | 2015-03-25 | 深圳市腾讯计算机系统有限公司 | Webpage recognition method and device |
| WO2015123214A1 (en) | 2014-02-11 | 2015-08-20 | Alibaba Group Holding Limited | Generating barcode and authenticating based on barcode |
| CN105046140A (en) * | 2015-06-09 | 2015-11-11 | 苏州德锐朗智能科技有限公司 | Automatic input method for character verification code |
| CN105094760A (en) * | 2014-04-28 | 2015-11-25 | 小米科技有限责任公司 | Picture marking method and device |
| CN105138867A (en) * | 2014-06-09 | 2015-12-09 | 北大方正集团有限公司 | Method and device for protecting image copyright |
| CN105160236A (en) * | 2015-08-31 | 2015-12-16 | 小米科技有限责任公司 | Method and device for inputting verification code |
| CN105225103A (en) * | 2014-07-02 | 2016-01-06 | 中国银联股份有限公司 | Continue payment system and method |
| CN105809096A (en) * | 2014-12-31 | 2016-07-27 | 中兴通讯股份有限公司 | Figure labeling method and terminal |
| CN106155994A (en) * | 2016-06-30 | 2016-11-23 | 广东小天才科技有限公司 | The comparative approach of a kind of content of pages and device, terminal unit |
| CN106203435A (en) * | 2016-07-13 | 2016-12-07 | 广州安望信息科技有限公司 | Picture and text recognition methods and device thereof |
| CN106529380A (en) * | 2015-09-15 | 2017-03-22 | 阿里巴巴集团控股有限公司 | Image identification method and device |
| CN106845323A (en) * | 2015-12-03 | 2017-06-13 | 阿里巴巴集团控股有限公司 | A kind of collection method of marking data, device and certificate recognition system |
| CN107958264A (en) * | 2017-11-20 | 2018-04-24 | 奕响（大连）科技有限公司 | A kind of similar decision method of picture |
| CN108052944A (en) * | 2017-12-27 | 2018-05-18 | 深圳市大熊动漫文化有限公司 | A kind of image-recognizing method and device |
| CN108734556A (en) * | 2018-05-18 | 2018-11-02 | 广州优视网络科技有限公司 | Recommend the method and device of application |
| CN109376746A (en) * | 2018-10-25 | 2019-02-22 | 黄子骞 | A kind of image identification method and system |
| CN110019898A (en) * | 2017-08-08 | 2019-07-16 | 航天信息股份有限公司 | A kind of animation image processing system |
| CN110414645A (en) * | 2018-04-28 | 2019-11-05 | 深圳果力智能科技有限公司 | A kind of pattern recognition method based on Match of elemental composition |
| CN110780789A (en) * | 2019-10-25 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Game application starting method and device, storage medium and electronic device |
| CN110796715A (en) * | 2019-08-26 | 2020-02-14 | 腾讯科技（深圳）有限公司 | Electronic map labeling method, device, server and storage medium |
| CN111259366A (en) * | 2020-01-22 | 2020-06-09 | 支付宝(杭州)信息技术有限公司 | Verification code recognizer training method and device based on self-supervision learning |
| CN110598390B (en) * | 2018-06-13 | 2021-07-13 | 南宁富桂精密工业有限公司 | Verification code method, server and verification code system based on picture |
| CN114996785A (en) * | 2022-06-13 | 2022-09-02 | 华侨大学 | Intelligent material selection method for slate typesetting and slate typesetting method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100516289B1 (en) * | 2000-11-02 | 2005-09-21 | 주식회사 케이티 | Content based image reference apparatus and method for relevance feedback using fussy integral |
| CN101034442A (en) * | 2006-03-08 | 2007-09-12 | 刘欣融 | System for judging between identical and proximate goods appearance design based on pattern recognition |
-
2007
- 2007-12-26 CN CNB2007103042044A patent/CN100550038C/en active Active
Cited By (51)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101635763A (en) * | 2008-07-23 | 2010-01-27 | 深圳富泰宏精密工业有限公司 | Picture classification system and method |
| CN101859368B (en) * | 2009-04-09 | 2013-12-04 | 普诚科技股份有限公司 | Image identification device and method |
| CN102467653A (en) * | 2010-10-29 | 2012-05-23 | 方正国际软件(北京)有限公司 | Image-text recognition method and system thereof |
| CN102890761A (en) * | 2011-08-24 | 2013-01-23 | 北京文海思创科技有限公司 | Method for verifying through graphical verification code |
| CN102890761B (en) * | 2011-08-24 | 2015-06-10 | 北京文海思创科技有限公司 | Method for verifying through graphical verification code |
| CN103186781A (en) * | 2011-12-31 | 2013-07-03 | 北京新媒传信科技有限公司 | Text recognition method |
| CN103258280A (en) * | 2012-02-17 | 2013-08-21 | 盛趣信息技术（上海）有限公司 | Price comparative method and system |
| CN103428515A (en) * | 2012-05-15 | 2013-12-04 | 索尼公司 | Video format determination device, video format determination method, and video display device |
| CN103428515B (en) * | 2012-05-15 | 2016-01-20 | 索尼公司 | Video format judgment device and method and video display apparatus |
| US9967536B2 (en) | 2012-05-15 | 2018-05-08 | Saturn Licensing Llc | Video format determination device, video format determination method, and video display device |
| CN103426191A (en) * | 2012-05-26 | 2013-12-04 | 百度在线网络技术（北京）有限公司 | Method and system for picture marking |
| CN103426191B (en) * | 2012-05-26 | 2016-04-27 | 百度在线网络技术（北京）有限公司 | A kind of picture mask method and system |
| CN102867025A (en) * | 2012-08-23 | 2013-01-09 | 百度在线网络技术（北京）有限公司 | Method and device for acquiring picture marking data |
| CN104252446A (en) * | 2013-06-27 | 2014-12-31 | 鸿富锦精密工业（深圳）有限公司 | Computing device, and verification system and method for consistency of contents of files |
| CN104462152B (en) * | 2013-09-23 | 2019-04-09 | 深圳市腾讯计算机系统有限公司 | A kind of recognition methods of webpage and device |
| CN104462152A (en) * | 2013-09-23 | 2015-03-25 | 深圳市腾讯计算机系统有限公司 | Webpage recognition method and device |
| WO2015123214A1 (en) | 2014-02-11 | 2015-08-20 | Alibaba Group Holding Limited | Generating barcode and authenticating based on barcode |
| CN103914996B (en) * | 2014-04-24 | 2016-11-23 | 广东小天才科技有限公司 | A kind of method and apparatus obtaining Words study data from picture |
| CN103914996A (en) * | 2014-04-24 | 2014-07-09 | 广东小天才科技有限公司 | Method and device for acquiring character learning materials from picture |
| CN105094760A (en) * | 2014-04-28 | 2015-11-25 | 小米科技有限责任公司 | Picture marking method and device |
| CN105094760B (en) * | 2014-04-28 | 2019-10-29 | 小米科技有限责任公司 | A kind of picture indicia method and device |
| CN105138867A (en) * | 2014-06-09 | 2015-12-09 | 北大方正集团有限公司 | Method and device for protecting image copyright |
| CN105225103A (en) * | 2014-07-02 | 2016-01-06 | 中国银联股份有限公司 | Continue payment system and method |
| CN105225103B (en) * | 2014-07-02 | 2020-05-22 | 中国银联股份有限公司 | Continuous payment system and method |
| CN104200204A (en) * | 2014-09-02 | 2014-12-10 | 福建富士通信息软件有限公司 | Picture processing device and method |
| CN104200204B (en) * | 2014-09-02 | 2017-10-03 | 福建富士通信息软件有限公司 | A kind of picture processing device and method |
| CN105809096A (en) * | 2014-12-31 | 2016-07-27 | 中兴通讯股份有限公司 | Figure labeling method and terminal |
| CN105046140A (en) * | 2015-06-09 | 2015-11-11 | 苏州德锐朗智能科技有限公司 | Automatic input method for character verification code |
| CN105160236A (en) * | 2015-08-31 | 2015-12-16 | 小米科技有限责任公司 | Method and device for inputting verification code |
| CN105160236B (en) * | 2015-08-31 | 2018-04-06 | 小米科技有限责任公司 | A kind of method and apparatus of input validation code |
| CN106529380A (en) * | 2015-09-15 | 2017-03-22 | 阿里巴巴集团控股有限公司 | Image identification method and device |
| CN106529380B (en) * | 2015-09-15 | 2019-12-10 | 阿里巴巴集团控股有限公司 | Image recognition method and device |
| CN106845323B (en) * | 2015-12-03 | 2020-04-28 | 阿里巴巴集团控股有限公司 | Marking data collection method and device and certificate identification system |
| CN106845323A (en) * | 2015-12-03 | 2017-06-13 | 阿里巴巴集团控股有限公司 | A kind of collection method of marking data, device and certificate recognition system |
| CN106155994A (en) * | 2016-06-30 | 2016-11-23 | 广东小天才科技有限公司 | The comparative approach of a kind of content of pages and device, terminal unit |
| CN106155994B (en) * | 2016-06-30 | 2019-04-26 | 广东小天才科技有限公司 | A kind of comparative approach and device, terminal device of content of pages |
| CN106203435A (en) * | 2016-07-13 | 2016-12-07 | 广州安望信息科技有限公司 | Picture and text recognition methods and device thereof |
| CN110019898A (en) * | 2017-08-08 | 2019-07-16 | 航天信息股份有限公司 | A kind of animation image processing system |
| CN107958264A (en) * | 2017-11-20 | 2018-04-24 | 奕响（大连）科技有限公司 | A kind of similar decision method of picture |
| CN108052944A (en) * | 2017-12-27 | 2018-05-18 | 深圳市大熊动漫文化有限公司 | A kind of image-recognizing method and device |
| CN110414645A (en) * | 2018-04-28 | 2019-11-05 | 深圳果力智能科技有限公司 | A kind of pattern recognition method based on Match of elemental composition |
| CN110414645B (en) * | 2018-04-28 | 2023-05-30 | 深圳果力智能科技有限公司 | Pattern recognition method based on element matching |
| CN108734556A (en) * | 2018-05-18 | 2018-11-02 | 广州优视网络科技有限公司 | Recommend the method and device of application |
| CN110598390B (en) * | 2018-06-13 | 2021-07-13 | 南宁富桂精密工业有限公司 | Verification code method, server and verification code system based on picture |
| CN109376746A (en) * | 2018-10-25 | 2019-02-22 | 黄子骞 | A kind of image identification method and system |
| CN110796715A (en) * | 2019-08-26 | 2020-02-14 | 腾讯科技（深圳）有限公司 | Electronic map labeling method, device, server and storage medium |
| CN110796715B (en) * | 2019-08-26 | 2023-11-24 | 腾讯科技（深圳）有限公司 | Electronic map labeling method, device, server and storage medium |
| CN110780789A (en) * | 2019-10-25 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Game application starting method and device, storage medium and electronic device |
| CN110780789B (en) * | 2019-10-25 | 2023-01-06 | 腾讯科技（深圳）有限公司 | Game application starting method and device, storage medium and electronic device |
| CN111259366A (en) * | 2020-01-22 | 2020-06-09 | 支付宝(杭州)信息技术有限公司 | Verification code recognizer training method and device based on self-supervision learning |
| CN114996785A (en) * | 2022-06-13 | 2022-09-02 | 华侨大学 | Intelligent material selection method for slate typesetting and slate typesetting method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100550038C (en) | 2009-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100550038C (en) | Image content recognizing method and recognition system | |
| CN103955660B (en) | Method for recognizing batch two-dimension code images | |
| CN106951832B (en) | Verification method and device based on handwritten character recognition | |
| CN113283446B (en) | Method and device for identifying object in image, electronic equipment and storage medium | |
| CN104809481A (en) | Natural scene text detection method based on adaptive color clustering | |
| CN106845513A (en) | Staff detector and method based on condition random forest | |
| CN105139041A (en) | Method and device for recognizing languages based on image | |
| CN111552966A (en) | Malicious software homology detection method based on information fusion | |
| CN112926379A (en) | Method and device for constructing face recognition model | |
| CN110399760A (en) | A kind of batch two dimensional code localization method, device, electronic equipment and storage medium | |
| CN104966109A (en) | Medical laboratory report image classification method and apparatus | |
| Obaidullah et al. | Structural feature based approach for script identification from printed Indian document | |
| CN110766010A (en) | Information identification method, model training method and related device | |
| CN113935880A (en) | Policy recommendation method, device, equipment and storage medium | |
| CN112613367A (en) | Bill information text box acquisition method, system, equipment and storage medium | |
| CN111680669A (en) | Test question segmentation method and system and readable storage medium | |
| CN111626313A (en) | Feature extraction model training method, image processing method and device | |
| CN115601768A (en) | Method, device and equipment for judging written characters and storage medium | |
| CN104504385A (en) | Recognition method of handwritten connected numerical string | |
| CN110414471B (en) | Video identification method and system based on double models | |
| CN109871910B (en) | Handwritten character recognition method and device | |
| CN113822521A (en) | Method and device for detecting quality of question library questions and storage medium | |
| CN105512655A (en) | Face recognition method and face recognition device | |
| CN112200184B (en) | Calligraphy area detection and author identification method in natural scene | |
| CN111539390A (en) | Small target image identification method, equipment and system based on Yolov3 |
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 | ||
| C41 | Transfer of patent application or patent right or utility model | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20151223 Address after: The South Road in Guangdong province Shenzhen city Fiyta building 518057 floor 5-10 Nanshan District high tech Zone Patentee after: Shenzhen Tencent Computer System Co., Ltd. Address before: Shenzhen Futian District City, Guangdong province 518044 Zhenxing Road, SEG Science Park 2 East Room 403 Patentee before: Tencent Technology (Shenzhen) Co., Ltd. |