CN103955713A - Icon recognition method and device - Google Patents
Icon recognition method and device Download PDFInfo
- Publication number
- CN103955713A CN103955713A CN201410168987.8A CN201410168987A CN103955713A CN 103955713 A CN103955713 A CN 103955713A CN 201410168987 A CN201410168987 A CN 201410168987A CN 103955713 A CN103955713 A CN 103955713A
- Authority
- CN
- China
- Prior art keywords
- icon
- station
- maximum
- similarity
- station symbol
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The embodiment of the invention relates to the technical field of image recognition, in particular to an icon recognition method and the device. The icon recognition method and device are used for solving the problem that a station caption recognition method in the prior art is low in recognition accuracy. The method includes the steps of matching an icon with a first set of stored icons to obtain a plurality of first similarity values, matching the change portion of the icon with the change portion of each icon in an icon set corresponding to the maximum first similarity value respectively if the maximum first similarity value is smaller than a first threshold value so as to obtain a plurality of second similarity values, and determining the recognition result according to the first similarity values and the second similarity values. Thus, the station caption recognition accuracy is improved.
Description
Technical field
The present invention relates to image recognition technology field, particularly a kind of icon identification method and apparatus.
Background technology
TV station's station symbol be a TV station indicate, the important semantic information such as the platform that comprises Liao Gai TV station is bright, program orientation, is one of important speech source of realizing video analysis, understanding and retrieval.The major function of TV is to watch live telecast, comprises a large amount of information in live telecast.
Have now bulk information service provider, more difficult but live telecast is connected with information service, the content information that causes obtaining live telecast is still more difficult.Station symbol detection and Identification technology is the key that overcomes above problem, has very important research and using value.
Existing TV station symbol recognition technology is: the size of obtaining all TV station's station symbols region, intercept station symbol in picture as masterplate according to this area size, masterplate is extracted to HOG (Histogram of Oriented Gradient, histograms of oriented gradients) feature, and by PCA (Principal Component Analysis, principal component analysis (PCA)) by data training SVM (the Support Vector Machine after dimension, support vector machine) sorter, when identification, classify by sorter.But current this mode is because video image content is more chaotic, and edge complexity, and piecemeal coupling, do not consider overall feature, destroyed the overall permanence of station symbol image.
In sum, the lower problem of station identification method for distinguishing recognition accuracy in prior art.
Summary of the invention
The embodiment of the present invention provides a kind of icon identification method and apparatus, in order to solve the lower problem of station identification method for distinguishing recognition accuracy existing in prior art.
A kind of icon recognition methods that the embodiment of the present invention provides, described icon comprises main part and changing unit, the method comprises:
Described icon is mated with the first group icon of storage respectively, obtain multiple the first similarity values;
If the first maximum similarity value is less than first threshold, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values;
Determine recognition result according to the first similarity value and the second similarity value.
Preferably, if the second maximum similarity is greater than the first maximum similarity, and the second maximum similarity value is greater than Second Threshold, icon corresponding to the second similarity value that definite described icon is maximum;
If the second maximum similarity is not more than the first maximum similarity, and the first similarity of described maximum is greater than Second Threshold, icon corresponding to the first similarity value that definite described icon is maximum.
Preferably, if the first maximum similarity value and the second maximum similarity are all less than Second Threshold, described icon is mated with the second group icon of storage respectively, obtain multiple third phases like degree value;
If maximum third phase is greater than Second Threshold like degree value, determine that described icon is maximum seemingly icon corresponding to degree value of third phase.
Preferably, described icon is TV station's station symbol, and described icon image is from television picture capturing.
Preferably, according to the image parameter of the first group icon, the described icon intercepting is normalized.
Preferably, according to changing unit image parameter, intercept the changing unit of described icon.
A kind of icon recognition device that the embodiment of the present invention provides, this device comprises:
The first icon matching unit, for described icon is mated with the first group icon of storage respectively, obtains multiple the first similarity values;
The second icon matching unit, if be less than first threshold for the first maximum similarity value, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values;
Icon recognition unit, for determining recognition result according to the first similarity value and the second similarity value.
Preferably, described icon recognition unit specifically for:
If the second maximum similarity is greater than the first maximum similarity, and the second maximum similarity value is greater than Second Threshold, icon corresponding to the second similarity value that definite described icon is maximum;
If the second maximum similarity is not more than the first maximum similarity, and the first similarity of described maximum is greater than Second Threshold, icon corresponding to the first similarity value that definite described icon is maximum.
Preferably, described image identification unit also for:
If the first maximum similarity value and the second maximum similarity are all less than Second Threshold, described icon is mated with the second group icon of storage respectively, obtain multiple third phases like degree value;
If maximum third phase is greater than Second Threshold like degree value, determine that described icon is maximum seemingly icon corresponding to degree value of third phase.
Preferably, described the first icon matching unit specifically for:
Described icon is TV station's station symbol, and described icon image is from television picture capturing.
Preferably, described the first icon matching unit also for:
According to the image parameter of the first group icon, the described icon intercepting is normalized.
Preferably, described the second icon matching unit specifically for: according to changing unit image parameter, intercept the changing unit of described icon.
Embodiment of the present invention icon comprises main part and changing unit, and described icon is mated with the first group icon of storage respectively, obtains multiple the first similarity values; If the first maximum similarity value is less than first threshold, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values; Determine recognition result according to the first similarity value and the second similarity value.Distinguish large class because the embodiment of the present invention adopts global characteristics, utilize local feature to carry out meticulous identification, improved the accuracy rate of TV station symbol recognition.
Brief description of the drawings
The image matching method process flow diagram that Fig. 1 provides for the embodiment of the present invention one;
Fig. 2 is TV station symbol recognition method particular flow sheet in the embodiment of the present invention two;
The image matching apparatus schematic diagram that Fig. 3 provides for the embodiment of the present invention three;
The image matching apparatus schematic diagram that Fig. 4 provides for the embodiment of the present invention four;
The image matching system schematic diagram that Fig. 5 provides for the embodiment of the present invention five.
Embodiment
Embodiment of the present invention icon comprises main part and changing unit, and described icon is mated with the first group icon of storage respectively, obtains multiple the first similarity values; If the first maximum similarity value is less than first threshold, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values; Determine recognition result according to the first similarity value and the second similarity value.Distinguish large class because the embodiment of the present invention adopts global characteristics, utilize local feature to carry out meticulous identification, strengthen the discrimination of similar station symbol, dwindle matching range, improve recognition accuracy and recognition efficiency.
Below in conjunction with Figure of description, the preferred embodiments of the present invention are described, should be appreciated that preferred embodiment described herein, only for description and interpretation the present invention, is not intended to limit the present invention.And in the situation that not conflicting, the feature in embodiment and embodiment in the application can combine mutually.
Embodiment mono-
As shown in Figure 1, a kind of icon recognition methods that the embodiment of the present invention one provides, described icon comprises main part and changing unit, comprises the following steps:
Step 101, by described icon respectively with storage the first group icon mate, obtain multiple the first similarity values.
If the first maximum similarity value of step 102 is less than first threshold, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values.
Step 103, determine recognition result according to the first similarity value and the second similarity value.
In step 101, the intercept method of icon can have multiple.
In enforcement, for obtaining of live telecast channel (video playback channel) station symbol, can adopt the methods such as picture download or target intercepting.
At present, TV station's station symbol has multiple, specifically comprises:
For the different channel that belongs to same TV station, main station symbol pattern is consistent, be that sub-station symbol has word difference, such as the channel station symbol CCTV-1 of the Chinese Central Television (CCTV), CCTV-2, CCTV-3, CCTV-4 ... CCTV-N etc. are many, the main station symbol CCTV of Chinese Central Television's channel is the same, only has the digital beings difference of sub-station symbol.
The station symbol of TV station of domestic each province and first class city TV station is diverse, such as Beijing, east, Shenzhen, Zhejiang, Hunan, phoenix, TVB (Television Broadcasts Limited, TVB) etc., exist and reach tens of different TV station's station symbols.
In enforcement, according to each television station channel resource, each television station channel is intercepted to a video image in advance, utilize image handling implement to intercept the image that comprises station symbol, and the image that comprises station symbol is carried out to the background process of going except station symbol, only on image, retain station symbol.Then set up TV station's station symbol pattern library.
In enforcement, can divide according to the similarity value of the sub-station symbol of the main station symbol of TV station's station symbol and TV station's station symbol.
Such as Chinese Central Television's channel station symbol comprises: CCTV-1, CCTV-2, CCTV-3, CCTV-4 ... CCTV-N etc., wherein the main station symbol CCTV of Chinese Central Television's station symbol is the same, only has the digital beings difference of sub-station symbol.So can the Chinese Central Television (CCTV) be divided into a class according to the similarity value of the main station symbol of TV station's station symbol.Be defined as one of image collection of the first group icon.
Such as the smart hawk cartoon in Hunan, good good cartoon, the turn cartoon in Shanghai etc. in Guangdong.Wherein the sub-station symbol of TV station's station symbol all comprises the same content, and different, the main station symbol of TV station's station symbol is different.So can above-mentioned TV station station symbol be divided into a class according to the similarity value of the sub-station symbol of TV station's station symbol.Be defined as one of image collection of the first group icon.
In enforcement, image parameter in TV station's station symbol icon thesaurus obtains according to training under line, specifically comprises what pixel, the coordinate position of TV station's station symbol image etc. of relative position scope, TV station's station symbol image of changing unit in TV station's icon station symbol obtained.
Such as, in Chinese Central Television's channel station symbol, CCTV-1, CCTV-2, CCTV-3, wherein changing unit is the numerical portion of sub-station symbol, and the letter of main station symbol be do not have vicissitudinous, image parameter in this TV station's icon thesaurus is to determine according to the coordinate position of the pixel of the relative position of TV station's station symbol image change part, TV station's station symbol image change and TV station's station symbol changing unit etc., so the changing unit in station symbol image should be chosen respectively " 1 ", " 2 " and " 3 ".
Such as the smart hawk cartoon in Hunan, good good cartoon, the turn cartoon in Shanghai etc. in Guangdong.Wherein the sub-station symbol of TV station's station symbol all comprises the same content, and different, the main station symbol of TV station's station symbol is different.So here the changing unit in station symbol image is respectively " smart hawk ", " Jia Jia " and " turn ".
In enforcement, what the first group icon was corresponding is common TV station station symbol, and what the second group icon was corresponding is uncommon TV station station symbol.The division of common TV station station symbol and uncommon TV station station symbol, specifically determined by the audience ratings of TV station, such as TV station's station symbol image high audience ratings of TV station is stored in to one, be divided into common TV station station symbol, and TV station's station symbol image relatively low TV station's audience ratings is stored in to one, be divided into uncommon TV station station symbol.
Such as, the higher TV station's station symbol as Hunan, phoenix, Zhejiang, east, Anhui, Beijing, CCTV etc. of audience ratings of TV station can be stored in common TV station's station symbol storehouse; And can be stored as in uncommon TV station's station symbol storehouse as TV station symbols such as formation, Hebei, Shaanxi, Shanxi lower audience ratings accordingly.Wherein the stored number in common TV station's station symbol storehouse is less than the stored number in uncommon TV station's station symbol storehouse.
In force, some television station channel station symbol can change in the time playing, and the station symbol a set of such as central TV has: CCTV-1, cctv1, central authorities' 1 cover, 1, central authorities, one, central authorities, the multiple variation station symbol such as CCTV is a set of.
Preferably, described icon is TV station's station symbol, and described icon image is from television picture capturing.
Preferably, according to the image parameter of the first group icon, the described icon intercepting is normalized.
In enforcement, according to the image parameter of the common TV station station symbol of storage, the described TV station station symbol image intercepting is normalized, obtains TV station's station symbol image that need to mate.
In enforcement, intercepting TV station station symbol image is reduced according to training parameter under line, obtain the TV station station symbol image consistent with training parameter.
In enforcement, according to the proportionate relationship of station symbol image in TV station's station symbol pattern library, such as 4:3,16:9 etc., by intercept after reduction TV station's station symbol image normalized of being correlated with, on normalized basis, intercept TV station's station symbol image region, obtain with TV station station symbol pattern library in the station symbol image of station symbol image scaled relationship consistency.
In enforcement, by with TV station station symbol pattern library in TV station's station symbol image of TV station's station symbol image scaled relationship consistency, mate with the each common TV station station symbol image in TV station station symbol pattern library, can obtain multiple the first similarity value S1, it can be the arbitrary value between 0-100%.
In step 102, first threshold H1 is pre-set.It can be the arbitrary value between 0-100%.
In enforcement, if the first maximum similarity value is not less than first threshold, directly export the common TV station station symbol image with the TV station's station symbol images match intercepting.
Such as the TV station's station symbol image intercepting is CCTV-1, and a set of station symbol of central TV that TV station's station symbol pattern library comprises has: the multiple variation station symbols such as CCTV-1, CCTV-2, CCTV-3, CCTV1, central authorities' 1 cover, 1, central authorities, one, central authorities, CCTV are a set of.After TV station's station symbol image of intercepting is mated with TV station's station symbol image of above-mentioned storage respectively, obtain maximum the first similarity value S1; Maximum the first similarity value is herein 100%, is not less than predefined first threshold H1, the common station symbol CCTV-1 of TV station of direct output matching.
In step 102, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values.
In enforcement, the image set credit union of common TV station station symbol image divides according to the similarity value of the sub-station symbol of the main station symbol of TV station's station symbol and TV station's station symbol.
Such as, the TV station's station symbol image intercepting is CCTV-1, as Chinese Central Television's channel station symbol comprises: cctv1, CCTV-2, CCTV-3, CCTV-4 ... CCTV-N etc., and the TV station's station symbol image intercepting is the sub-station symbol " 1 " changing, changing unit " 1 ", " 2 ", " 3 ", " 4 " and " N " in each common TV station station symbol image in TV station's station symbol image change part " 1 " of above-mentioned intercepting and definite TV station's station symbol image collection are mated, obtain each the second similarity value S2, it can be the arbitrary value between 0-100%.
In step 103, determine recognition result according to the first similarity value and the second similarity value.Wherein recognition methods has multiplely, enumerates several below.
Mode one
If the second maximum similarity is greater than the first maximum similarity, and the second maximum similarity value is greater than Second Threshold, icon corresponding to the second similarity value that definite described icon is maximum;
In enforcement, Second Threshold H2 is pre-set, and it can be the arbitrary value between 0-100%.
Such as the TV station's station symbol image intercepting is CCTV-1, and the station symbol of the Chinese Central Television (CCTV) that TV station's station symbol pattern library comprises has: the multiple variation station symbols such as CCTV-2, CCTV-3, CCTV1, central authorities' 1 cover, 1, central authorities, one, central authorities, CCTV are a set of.TV station's station symbol image of intercepting is mated with above-mentioned storage TV station station symbol image respectively, obtain maximum the first similarity value S1.
According to the similarity value between TV station's station symbol image, the station symbol image C CTV-2 of TV station, CCTV-3, the CCTV1 of above-mentioned storage are divided into a class image collection, determine that the changing unit in image collection is " 2 ", " 3 " and " 1 ".Intercept the changing unit " 1 " of the station symbol image C CTV-1 of TV station.
The changing unit " 2 ", " 3 " and " 1 " that intercept in changing unit " 1 " and definite TV station's station symbol image collection " CCTV-2, CCTV-3, CCTV1 " of TV station station symbol image are mated, obtain the second similarity value S2 of maximum.
If the second maximum similarity value S2 is greater than the first maximum similarity value S1, and the second maximum similarity value S2 is greater than Second Threshold H2, determining maximum common TV station's station symbol image " CCTV1 " corresponding to the second similarity value S2, is the common TV station station symbol image mating with TV station's station symbol of described intercepting.
Mode two
If the second maximum similarity is not more than the first maximum similarity, and the first similarity of described maximum is greater than Second Threshold, icon corresponding to the first similarity value that definite described icon is maximum.
In force, Second Threshold H2 is pre-set, and it can be the arbitrary value between 0-100%.
Such as the TV station's station symbol image intercepting is cctv1, and a set of station symbol of central TV that TV station's station symbol pattern library comprises has: the multiple variation station symbols such as CCTV-1, CCTV-2, CCTV-3, CCTV1, central authorities' 1 cover, 1, central authorities, one, central authorities, CCTV are a set of.TV station's station symbol image of intercepting is mated with above-mentioned storage TV station station symbol image respectively, obtain maximum the first similarity value S1; According to the similarity value between the station symbol figure figure of TV station, the station symbol image C CTV-1 of TV station, CCTV-2, CCTV-3, the CCTV1 of above-mentioned storage are divided into a class image collection.
Then the changing unit " T-1 ", " T-2 ", " T-3 " and " t1 " that intercept in changing unit " t1 " and definite TV station's station symbol image collection " CCTV-1, CCTV-2, CCTV-3, CCTV1 " of TV station's station symbol image mate, and obtain the second similarity value S2.
If the second maximum similarity value S2 is not more than the first maximum similarity value S1, and the first similarity value S1 of described maximum is greater than Second Threshold H2, determine maximum common TV station's station symbol image " CCTV-1 " corresponding to the first similarity value S1, for the common TV station station symbol image of TV station's station symbol images match of described intercepting.
After TV station's station symbol image of intercepting is mated with the each common TV station station symbol image of storage respectively, also comprise:
If the first maximum similarity value is less than first threshold, and the image collection at common station symbol image place corresponding to the first maximum similarity value does not have other common station symbol images, TV station's station symbol image of intercepting is mated with the each uncommon TV station station symbol image of storage respectively, obtain multiple third phases like degree value.
According to third phase like degree the relation with the first similarity and Second Threshold, judge that whether TV station's station symbol image has and have multiplely with the mode of the common TV station station symbol image of TV station's station symbol images match of intercepting, enumerates several below.
Method one
If maximum third phase is greater than the first maximum similarity value like degree value, and maximum third phase is greater than Second Threshold like degree value, determine that maximum third phase is like uncommon station symbol image corresponding to degree value, for TV station's station symbol image of TV station's station symbol images match of described intercepting.
In force, Second Threshold H2 is pre-set, and it can be the arbitrary value between 0-100%.
Such as the TV station's station symbol image intercepting is satellite TV of formation, TV station's station symbol image of satellite TV of formation is mated with the common TV station station symbol image of storage respectively, obtain the first similarity value S1; Common TV station station symbol image corresponding to maximum similarity value generally includes herein: the higher TV station's station symbol image collections of audience ratings such as HNTV, ZTV, Dragon TV, BTV, Anhui STV.
And TV station's station symbol image of satellite TV of formation is not stored in common TV station station symbol image library, so the result of coupling is unsuccessfully, then TV station's station symbol image of the satellite TV of formation of intercepting and the lower uncommon TV station station symbol image of audience ratings are mated, obtain multiple third phases like degree value S3.
Wherein, intercepting TV station's station symbol image of satellite TV of formation and the each common TV station station symbol image (HNTV, ZTV, Dragon TV, BTV, Anhui STV etc.) of storage mates, obtain multiple the first similarity value S1, multiple similarity values herein may be " satellite TV ", " soldier satellite TV " and " satellite TV of group " etc.
The each uncommon TV station station symbol image (satellite TV of formation, Hebei satellite TV, Shaanxi satellite TV, Shanxi satellite TV etc.) that wherein intercepts TV station of satellite TV of formation station symbol image and storage mates, obtain multiple third phases like degree value S3, maximum third phase is " satellite TV of formation " like degree value herein.
Determine that maximum third phase is like uncommon TV station station symbol image corresponding to degree value S3, for TV station's station symbol image of TV station's station symbol images match of described intercepting.
Method two
If maximum third phase is not more than the first maximum similarity value like degree value, and the first similarity value of described maximum is greater than Second Threshold, determine maximum common station symbol image corresponding to the first similarity value, for the station symbol image of TV station's station symbol images match of described intercepting.
Such as the TV station's station symbol image intercepting is TVB-1, TV station's station symbol image of TVB-1 is mated respectively at the common TV station station symbol image of storage, obtain the first similarity value; Common TV station station symbol image corresponding to maximum similarity value generally includes herein: the higher TV station's station symbol image collections of audience ratings such as CCTV-1, CCTV-2, HNTV, ZTV, Dragon TV, BTV, Anhui STV.
And TV station's station symbol image of TVB-1 is not stored in common TV station station symbol image library, so the result of coupling is unsuccessfully, then TV station's station symbol image of the TVB-1 of intercepting and the lower uncommon TV station station symbol of audience ratings are mated, obtain multiple third phases like degree value S3.
Wherein, the each common TV station station symbol image (CCTV-1, CCTV-2, HNTV, ZTV, Dragon TV, BTV, Anhui STV etc.) that intercepts TVB-1 and storage mates, obtain multiple the first similarity value S1, multiple similarity values herein may be " TV-1 " and " TV-" etc.
The each uncommon TV station station symbol image (satellite TV of formation, Hebei satellite TV, Shaanxi satellite TV, Shanxi satellite TV etc.) that wherein intercepts TVB-1 TV station station symbol image and storage mates, obtain multiple third phases like degree value S3, but similarity value S3 herein equals zero.
If this first similarity value S1 is greater than default Second Threshold H2, determine maximum common TV station station symbol image corresponding to the first similarity value S1, for TV station's station symbol image of the station symbol images match of described intercepting.
The embodiment of the present invention is mated TV station's station symbol image of intercepting respectively with the each common TV station station symbol image library of storage, obtain multiple the first similarity values; If maximum the first similarity value is greater than first threshold, the common TV station station symbol image of direct output matching, the images match efficiency of raising.If the first maximum similarity value is less than first threshold, determine the image collection at common TV station's station symbol image place corresponding to the first maximum similarity value, changing unit in TV station's station symbol image of intercepting is mated with the changing unit in each common TV station station symbol image in definite TV station's image collection respectively, and the TV station's station symbol image whether having in TV station's station symbol image collection with the TV station's station symbol images match intercepting is determined in judgement.Adopt the method, station symbol image overall feature and local feature are identified in conjunction with carrying out multi-level coupling, improved the accuracy rate of TV station symbol recognition.
Embodiment bis-
The technical scheme that embodiment mono-provides for a better understanding of the present invention, below TV station symbol recognition method idiographic flow be example, as shown in Figure 2, illustrate a kind of icon recognition methods of the present invention.
Step 201, intercept the Target Station picture of marking on a map, and carry out pre-service, with the Target Station mark picture noise that reduces to intercept.
Step 202, according to the image parameter of common station symbol image of storage, the described station symbol image intercepting is normalized, obtain the station symbol image that need to mate.
Concrete, intercepting station symbol image is reduced according to training parameter under line, obtain the station symbol image consistent with training parameter, and according to the proportionate relationship (as 4:3,16:9 etc.) of station symbol image in station symbol pattern library, by the station symbol image normalized of being correlated with after reduction, on normalized basis, intercept station symbol image region, obtain the station symbol image consistent with station symbol image scaled in station symbol pattern library.
The edge image of the image that step 203, calculating intercept out, by the outline map of canny algorithm extraction image.
Step 204, by the station symbol image of intercepting respectively with storage each common station symbol image mate.
After step 205, coupling, obtain multiple the first similarity value S1, if maximum similarity value S1, higher than the threshold value H1 setting, directly exports and the common station symbol image that intercepts station symbol images match.
If the first maximum similarity value S1 of step 206 is less than first threshold H1, judge whether the image collection at common station symbol image place has other common station symbol images.If have other common station symbol images, carry out step 207; If there is no other common station symbol images, carry out step 213.
Step 207, determine the image collection at the common station symbol image place that the first maximum similarity value S1 is corresponding, the changing unit in the station symbol image of intercepting is mated with the changing unit in each common station symbol image in definite image collection respectively.
Changing unit in step 208, the station symbol image that intercepts is mated with the changing unit in each common station symbol image in definite image collection respectively, obtains multiple the second similarity value S2.If press the related coefficient of method calculation template shown in formula (1) and cut-away view picture.If input picture size is n*m, template image size is also n*m, x
ijexpression input picture is located the gray-scale value of pixel, y at (i, j)
ijexpression input picture is located the gray-scale value of pixel at (i, j), the related coefficient of two width images is:
Step 209, according to the relation between the first similarity value S1, the second similarity value S2 and Second Threshold H2, judge the common station symbol image whether having in definite image collection with the station symbol images match intercepting.
If the highest similarity S2 of step 210 coupling is higher than setting threshold H2, and higher than S1, export the matching result that S2 is corresponding.
If the highest similarity S2 of step 211 coupling is higher than setting threshold H2, and lower than S1, export the matching result that S1 is corresponding.
If step 212 highest similarity S2 is lower than the threshold value H2 setting, and S1 is also lower than setting threshold H2, output matching failure or identify invalid.
Step 213, determine that the image collection at common station symbol image place corresponding to the first maximum similarity value does not have other common station symbol images.
Step 214, the station symbol image of intercepting is mated with the each uncommon station symbol image of storage respectively, obtain multiple third phases like degree value S3.
Step 215, according to the first similarity value S1, third phase like the relation between degree value S3 and Second Threshold H2, judge whether common station symbol image has and the common station symbol image of the station symbol images match intercepting.
If seemingly spending S3, the maximum third phase of step 216 is greater than the first maximum similarity S1, and maximum third phase is seemingly spent S3 value and is greater than Second Threshold H2, determine that maximum third phase is like uncommon station symbol image corresponding to degree value S3, for the common station symbol image of the station symbol images match of described intercepting.
If seemingly spending S3, the maximum third phase of step 217 is not more than the first maximum similarity S1, and the first similarity S1 of described maximum is greater than Second Threshold H2, determine maximum common station symbol image corresponding to the first similarity value S1, for the common station symbol image of the station symbol images match of described intercepting.
Be not more than the first maximum similarity S1 if the maximum third phase of step 218 is seemingly spent S3, and the first similarity S1 of described maximum is also not more than Second Threshold H2, output matching failure or identify invalid.
The embodiment of the present invention is mated the station symbol image of intercepting respectively with the each common station symbol image of storage, obtain multiple the first similarity values; If maximum the first similarity value is greater than first threshold, the common station symbol image of direct output matching, the images match efficiency of raising.
If the first maximum similarity value is less than first threshold, according to whether storing common station symbol image in image collection, if store common station symbol image in image collection, the changing unit in the station symbol image of intercepting is mated with the changing unit in each common station symbol image in definite image collection respectively, obtain multiple the second similarity values; If do not store common station symbol image in image collection, the station symbol image of intercepting is mated with the each uncommon station symbol image of storage respectively, obtain multiple third phases like degree value; According to similarity value and predefined threshold value relation, judgement and the station symbol image that intercepts station symbol images match.Adopt the method, station symbol image overall feature and local feature are identified in conjunction with carrying out multi-level coupling, finally determine the common station symbol image of coupling in conjunction with both matching results, strengthened the discrimination of similar station symbol, improved the accuracy rate of TV station symbol recognition.
Embodiment tri-
The embodiment of the present invention also provides a kind of equipment corresponding with said method, and the detailed content of this equipment can, referring to method part, no longer repeat to introduce at this.
As shown in Figure 3, this image matching apparatus comprises: the first icon matching unit 30, the second icon matching unit 31 and icon recognition unit 32.
The first icon matching unit 30, for described icon is mated with the first group icon of storage respectively, obtains multiple the first similarity values;
The second icon matching unit 31, if be less than first threshold for the first maximum similarity value, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values;
Icon recognition unit 32, for determining recognition result according to the first similarity value and the second similarity value.
Preferably, described icon recognition unit 32, specifically for:
If the second maximum similarity is greater than the first maximum similarity, and the second maximum similarity value is greater than Second Threshold, icon corresponding to the second similarity value that definite described icon is maximum;
If the second maximum similarity is not more than the first maximum similarity, and the first similarity of described maximum is greater than Second Threshold, icon corresponding to the first similarity value that definite described icon is maximum.
Preferably, described image recognition list 32, also for:
If the first maximum similarity value and the second maximum similarity are all less than Second Threshold, described icon is mated with the second group icon of storage respectively, obtain multiple third phases like degree value;
If maximum third phase is greater than Second Threshold like degree value, determine that described icon is maximum seemingly icon corresponding to degree value of third phase.
Preferably, described the first icon matching unit 30 specifically for:
Described icon is TV station's station symbol, and described icon image is from television picture capturing.
Preferably, described the first icon matching unit 30 also for:
According to the image parameter of the first group icon, the described icon intercepting is normalized.
Preferably, described the second icon matching unit 31 specifically for: according to changing unit image parameter, intercept the changing unit of described icon.
The embodiment of the present invention provides a kind of icon recognition device, by the first icon matching unit, the second icon matching unit and icon recognition unit, adopt the method, utilize global characteristics to distinguish large class and local feature carries out meticulous identification, station symbol image overall feature and local feature are identified in conjunction with carrying out multi-level coupling, improved the accuracy rate of TV station symbol recognition.
Embodiment tetra-
As shown in Figure 4, the embodiment of the present application four provides a kind of icon recognition device, comprises processor 40, storer 41, user interface 42 and bus interface 43.Wherein, between processor 40, storer 41 and user interface 42, be connected by bus interface 43.
Processor 40, described icon comprises main part and changing unit, for described icon is mated with the first group icon of storage respectively, obtains multiple the first similarity values; If the first maximum similarity value is less than first threshold, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values; Determine recognition result according to the first similarity value and the second similarity value.
Preferably, processor 40, specifically for: if the second maximum similarity is greater than the first maximum similarity, and the second maximum similarity value is greater than Second Threshold, icon corresponding to the second similarity value that definite described icon is maximum;
If the second maximum similarity is not more than the first maximum similarity, and the first similarity of described maximum is greater than Second Threshold, icon corresponding to the first similarity value that definite described icon is maximum.
Preferably, processor 40, also for: if the first maximum similarity value and the second maximum similarity are all less than Second Threshold, described icon is mated with the second group icon of storage respectively, obtains multiple third phases like degree value;
If maximum third phase is greater than Second Threshold like degree value, determine that described icon is maximum seemingly icon corresponding to degree value of third phase.
Preferably, processor 40, specifically for: described icon is TV station's station symbol, and described icon image is from television picture capturing.
Preferably, processor 40, specifically for: according to the image parameter of the first group icon, the described icon intercepting is normalized.
Preferably, processor 40, specifically for: according to changing unit image parameter, intercept the changing unit of described icon.
Wherein, in Fig. 4, bus architecture can comprise interconnected bus and the bridge of any amount, and the various electrical chains of the storer that the one or more processors that specifically represented by processor 40 and storer 41 represent are connected together.Bus architecture can also be connected together various other electrical chains such as peripherals, voltage stabilizer and management circuit etc., and these are all known in the field, therefore, no longer it are described further herein.Bus interface 43 provides interface.Processor 40 is in charge of bus architecture and common processing, the data that storer 41 can storage of processor 41 uses in the time of executable operations.For different subscriber equipmenies, user interface 42 can also be to connect the interface that needs equipment in can be external, and the equipment of connection includes but not limited to keypad, display, loudspeaker, microphone, operating rod etc.
Processor 40 is in charge of bus architecture and common processing, the data that storer 41 can storage of processor 40 uses in the time of executable operations.
By the description of above-described process flow diagram and scheme, global characteristics and local feature are carried out to combination, divide large class by global characteristics, carry out meticulous identification by local feature, increase the discrimination of similar station symbol, image recognition accuracy rate is provided.
Should be appreciated that the only logical partitioning for carrying out according to the function of this apparatus realization of unit that above image matching apparatus comprises, in practical application, can carry out stack or the fractionation of said units.And the image matching method that the function that the image matching apparatus that this embodiment provides is realized provides with above-described embodiment is corresponding one by one, the more detailed treatment scheme realizing for this equipment, in said method embodiment mono-, be described in detail, be not described in detail herein.
Embodiment five
The embodiment of the present invention four provides a kind of image matching system, and as shown in Figure 5, this image matching apparatus comprises: client device 50 and server end equipment 51.
Client device 50, for by the changing unit of icon and described icon, sends to server.
Server end equipment 51, mates with the first group icon of storage respectively for icon, obtains multiple the first similarity values; If the first maximum similarity value is less than first threshold, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values; Determine recognition result according to the first similarity value and the second similarity value.
Preferably, client device 50, also for by icon image from television picture capturing.
Preferably, client device 50, the icon matching result also sending for reception server end equipment 51.
Preferably, server end equipment 51, also, for according to the image parameter of the first group icon, is normalized the described icon intercepting.
Preferably, server end equipment 51, also for the picture matching result of marking on a map to client device 50 sending stations.
For the image matching system of the present embodiment, client device wherein and server apparatus can be arranged in same entity apparatus, can be also multiple independent entity devices that arrange.In the time that client device and server apparatus are multiple independent entity device arranging, client device can be televisor, and server apparatus can be the webserver etc.
Those skilled in the art should understand, embodiments of the invention can be provided as method, system or computer program.Therefore, the present invention can adopt complete hardware implementation example, completely implement software example or the form in conjunction with the embodiment of software and hardware aspect.And the present invention can adopt the form at one or more upper computer programs of implementing of computer-usable storage medium (including but not limited to magnetic disk memory, CD-ROM, optical memory etc.) that wherein include computer usable program code.
The present invention is with reference to describing according to process flow diagram and/or the block scheme of the method for the embodiment of the present invention, equipment (system) and computer program.Should understand can be by the flow process in each flow process in computer program instructions realization flow figure and/or block scheme and/or square frame and process flow diagram and/or block scheme and/or the combination of square frame.Can provide these computer program instructions to the processor of multi-purpose computer, special purpose computer, Embedded Processor or other programmable data processing device to produce a machine, the instruction that makes to carry out by the processor of computing machine or other programmable data processing device produces the device for realizing the function of specifying at flow process of process flow diagram or multiple flow process and/or square frame of block scheme or multiple square frame.
These computer program instructions also can be stored in energy vectoring computer or the computer-readable memory of other programmable data processing device with ad hoc fashion work, the instruction that makes to be stored in this computer-readable memory produces the manufacture that comprises command device, and this command device is realized the function of specifying in flow process of process flow diagram or multiple flow process and/or square frame of block scheme or multiple square frame.
These computer program instructions also can be loaded in computing machine or other programmable data processing device, make to carry out sequence of operations step to produce computer implemented processing on computing machine or other programmable devices, thereby the instruction of carrying out is provided for realizing the step of the function of specifying in flow process of process flow diagram or multiple flow process and/or square frame of block scheme or multiple square frame on computing machine or other programmable devices.
Although described the preferred embodiments of the present invention, once those skilled in the art obtain the basic creative concept of cicada, can make other change and amendment to these embodiment.So claims are intended to be interpreted as comprising preferred embodiment and fall into all changes and the amendment of the scope of the invention.
Obviously, those skilled in the art can carry out various changes and modification and the spirit and scope that do not depart from the application to the application.Like this, if these amendments of the application and within modification belongs to the scope of the application's claim and equivalent technologies thereof, the application is also intended to comprise these changes and modification interior.
Claims (12)
1. an icon recognition methods, described icon comprises main part and changing unit, it is characterized in that, the method comprises:
Described icon is mated with the first group icon of storage respectively, obtain multiple the first similarity values;
If the first maximum similarity value is less than first threshold, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values;
Determine recognition result according to the first similarity value and the second similarity value.
2. the method for claim 1, is characterized in that:
If the second maximum similarity is greater than the first maximum similarity, and the second maximum similarity value is greater than Second Threshold, icon corresponding to the second similarity value that definite described icon is maximum;
If the second maximum similarity is not more than the first maximum similarity, and the first similarity of described maximum is greater than Second Threshold, icon corresponding to the first similarity value that definite described icon is maximum.
3. method as claimed in claim 2, is characterized in that:
If the first maximum similarity value and the second maximum similarity are all less than Second Threshold, described icon is mated with the second group icon of storage respectively, obtain multiple third phases like degree value;
If maximum third phase is greater than Second Threshold like degree value, determine that described icon is maximum seemingly icon corresponding to degree value of third phase.
4. the method as described in as arbitrary in claim 1-3, is characterized in that:
Described icon is TV station's station symbol, and described icon image is from television picture capturing.
5. the method as described in as arbitrary in claim 1-4, is characterized in that: described method also comprises:
According to the image parameter of the first group icon, described icon is normalized.
6. the method as described in as arbitrary in claim 1-3, is characterized in that:
According to changing unit image parameter, intercept the changing unit of described icon.
7. an icon recognition device, is characterized in that, this device comprises:
The first icon matching unit, for described icon is mated with the first group icon of storage respectively, obtains multiple the first similarity values;
The second icon matching unit, if be less than first threshold for the first maximum similarity value, by the changing unit of described icon respectively the changing unit of the each icon in the icon set corresponding with the first similarity value of maximum mate, obtain multiple the second similarity values;
Icon recognition unit, for determining recognition result according to the first similarity value and the second similarity value.
8. device as claimed in claim 7, is characterized in that, described icon recognition unit specifically for:
If the second maximum similarity is greater than the first maximum similarity, and the second maximum similarity value is greater than Second Threshold, icon corresponding to the second similarity value that definite described icon is maximum;
If the second maximum similarity is not more than the first maximum similarity, and the first similarity of described maximum is greater than Second Threshold, icon corresponding to the first similarity value that definite described icon is maximum.
As claimed in claim 8 device, it is characterized in that, described image identification unit also for:
If the first maximum similarity value and the second maximum similarity are all less than Second Threshold, described icon is mated with the second group icon of storage respectively, obtain multiple third phases like degree value;
If maximum third phase is greater than Second Threshold like degree value, determine that described icon is maximum seemingly icon corresponding to degree value of third phase.
10. as described in as arbitrary in claim 7-9, device, is characterized in that, described the first icon matching unit specifically for:
Described icon is TV station's station symbol, and described icon image is from television picture capturing.
11. as arbitrary in claim 7-10 as described in device, it is characterized in that, described the first icon matching unit also for:
According to the image parameter of the first group icon, described icon is normalized.
12. as arbitrary in claim 7-9 as described in device, it is characterized in that, described the second icon matching unit specifically for:
According to changing unit image parameter, intercept the changing unit of described icon.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410168987.8A CN103955713B (en) | 2014-04-24 | 2014-04-24 | A kind of icon-based programming method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410168987.8A CN103955713B (en) | 2014-04-24 | 2014-04-24 | A kind of icon-based programming method and apparatus |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103955713A true CN103955713A (en) | 2014-07-30 |
| CN103955713B CN103955713B (en) | 2017-08-18 |
Family
ID=51332987
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410168987.8A Active CN103955713B (en) | 2014-04-24 | 2014-04-24 | A kind of icon-based programming method and apparatus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103955713B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105868683A (en) * | 2015-11-24 | 2016-08-17 | 乐视致新电子科技（天津）有限公司 | Channel logo identification method and apparatus |
| CN106446850A (en) * | 2016-09-30 | 2017-02-22 | 中国传媒大学 | Station logo recognition method and device |
| CN107133553A (en) * | 2016-02-29 | 2017-09-05 | 无锡天脉聚源传媒科技有限公司 | A kind of method and device for recognizing TV station's icon |
| CN107516097A (en) * | 2017-08-10 | 2017-12-26 | 青岛海信电器股份有限公司 | TV station symbol recognition method and apparatus |
| CN109886324A (en) * | 2019-02-01 | 2019-06-14 | 广州云测信息技术有限公司 | Icon-based programming method and apparatus |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009125596A1 (en) * | 2008-04-11 | 2009-10-15 | パナソニック株式会社 | Image processing apparatus, method, and storage medium |
| CN102663454A (en) * | 2012-04-18 | 2012-09-12 | 安徽科大讯飞信息科技股份有限公司 | Method and device for evaluating character writing standard degree |
| CN103024552A (en) * | 2012-12-07 | 2013-04-03 | 康佳集团股份有限公司 | Television channel background recognition method |
-
2014
- 2014-04-24 CN CN201410168987.8A patent/CN103955713B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009125596A1 (en) * | 2008-04-11 | 2009-10-15 | パナソニック株式会社 | Image processing apparatus, method, and storage medium |
| CN102663454A (en) * | 2012-04-18 | 2012-09-12 | 安徽科大讯飞信息科技股份有限公司 | Method and device for evaluating character writing standard degree |
| CN103024552A (en) * | 2012-12-07 | 2013-04-03 | 康佳集团股份有限公司 | Television channel background recognition method |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105868683A (en) * | 2015-11-24 | 2016-08-17 | 乐视致新电子科技（天津）有限公司 | Channel logo identification method and apparatus |
| CN107133553A (en) * | 2016-02-29 | 2017-09-05 | 无锡天脉聚源传媒科技有限公司 | A kind of method and device for recognizing TV station's icon |
| CN106446850A (en) * | 2016-09-30 | 2017-02-22 | 中国传媒大学 | Station logo recognition method and device |
| CN107516097A (en) * | 2017-08-10 | 2017-12-26 | 青岛海信电器股份有限公司 | TV station symbol recognition method and apparatus |
| CN107516097B (en) * | 2017-08-10 | 2020-03-24 | 青岛海信电器股份有限公司 | Station caption identification method and device |
| CN109886324A (en) * | 2019-02-01 | 2019-06-14 | 广州云测信息技术有限公司 | Icon-based programming method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103955713B (en) | 2017-08-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103235956B (en) | A kind of commercial detection method and device | |
| CN103927387A (en) | Image retrieval system, method and device | |
| CN108780507A (en) | System and method for providing Image Classifier | |
| CN102713938A (en) | Scale space normalization technique for improved feature detection in uniform and non-uniform illumination changes | |
| CN103955713A (en) | Icon recognition method and device | |
| US10373014B2 (en) | Object detection method and image search system | |
| CN110942061A (en) | Character recognition method, device, equipment and computer readable medium | |
| CN108563651B (en) | Multi-video target searching method, device and equipment | |
| CN105956631A (en) | On-line progressive image classification method facing electronic image base | |
| CN103198311A (en) | Method and apparatus for recognizing a character based on a photographed image | |
| CN105260428A (en) | Picture processing method and apparatus | |
| CN108536467A (en) | Location processing method, device, terminal device and the storage medium of code | |
| KR102018046B1 (en) | Method and apparatus for extracting image feature | |
| CN103995864B (en) | A kind of image search method and device | |
| Soni et al. | Keypoints based enhanced multiple copy‐move forgeries detection system using density‐based spatial clustering of application with noise clustering algorithm | |
| CN114187333A (en) | Image alignment method, image alignment device and terminal equipment | |
| KR20170126300A (en) | Method, apparatus and computer program for matching between the images | |
| CN109697240A (en) | A kind of image search method and device based on feature | |
| CN104966109A (en) | Medical laboratory report image classification method and apparatus | |
| CN110321759B (en) | Video feature extraction method and device | |
| CN114119964A (en) | Network training method and device, and target detection method and device | |
| CN105022480A (en) | Input method and terminal | |
| CN112883827A (en) | Method and device for identifying designated target in image, electronic equipment and storage medium | |
| CN112348008A (en) | Certificate information identification method and device, terminal equipment and storage medium | |
| CN109685079B (en) | Method and device for generating characteristic image category information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |