CN104954854A - Station logo identification method and electronic equipment - Google Patents
Station logo identification method and electronic equipment Download PDFInfo
- Publication number
- CN104954854A CN104954854A CN201410120709.5A CN201410120709A CN104954854A CN 104954854 A CN104954854 A CN 104954854A CN 201410120709 A CN201410120709 A CN 201410120709A CN 104954854 A CN104954854 A CN 104954854A
- Authority
- CN
- China
- Prior art keywords
- image
- station symbol
- model
- component
- signal
- 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
Abstract
The embodiment of the invention provides a station logo identification method and electronic equipment, which relates to the field of image processing. The station logo and a background are distinguished on the basis of a pre-stored model base for station logo identification. The method comprises steps: a first image is acquired, wherein the first image is a frame of image transmitted by a first channel in television signals, and the first image contains station logo information and background information; first processing is carried out on the first image, and a first component related to the station logo information is acquired; the first component and a model component in the pre-stored model base are compared, a first model component matched with the first component is acquired, and the station logo corresponding to the first model component serves as an identification result.
Description
Technical field
The present invention relates to field of information processing, particularly relate to a kind of TV station symbol recognition method and electronic equipment.
Background technology
Station symbol is the mark of TV station and TV column, is one of academia and industrial quarters study hotspot in recent years.TV, by identifying the programme content of user's viewing, can provide the service of more property for user.This identifies with regard to needing the station symbol of TV to user's view content, is user's recommending television according to the station symbol identified, thus improves the experience of user.
In existing TV tech, be carry out TV station symbol recognition based on characteristics of image, adopt identification timestamp in this way, in the background content in picture except station symbol usually can be calculated in, thus cause discrimination to reduce.
Summary of the invention
Embodiments provide a kind of TV station symbol recognition method and electronic equipment, the model library based on pre-stored carries out TV station symbol recognition, can distinguish station symbol and background more exactly, improves discrimination.
For achieving the above object, embodiments of the invention adopt following technical scheme:
First aspect, discloses a kind of station identification method for distinguishing, is applied to electronic equipment, comprises:
Obtain the first image, described first image is the two field picture that in TV signal, the first channel transmits, and described first image comprises station symbol information and background information;
First process is carried out to described first image, obtains first component relevant to described station symbol information;
Contrast the model component in the model library of described first component and pre-stored, obtain and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result.
In conjunction with first aspect, in the implementation that the first is possible, describedly carry out the first process to described first image, before obtaining the first component relevant to described station symbol information, described method also comprises:
According to M the channel existed, Modling model storehouse; Described model library comprises M model component, the corresponding channel of each described model component, described M be greater than 1 positive integer.
In conjunction with the first possible implementation of first aspect, in the implementation that the second is possible, described M the channel according to existing, Modling model storehouse, is specially,
For each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer;
Signal transacting is carried out to described N number of sample image, obtains at least one the independent signal unit all existed in described N number of sample image;
Be normalized at least one independent signal unit described, obtain at least one station symbol signal element, utilize at least one station symbol signal element component model component described, described model component corresponds to channel corresponding to described N number of sample image;
Structure discrete cosine signal group, is normalized described discrete cosine group of functions, obtains background signal tuple;
Obtain M model component, a described M model component and described background signal tuple are merged and forms described model library.
In conjunction with the implementation that the second of first aspect is possible, in the implementation that the third is possible, the first process is carried out to described first image, obtains first component relevant to described station symbol information, be specially,
Based on described model library, sparse signal decomposition is carried out to described first image, obtain described first image corresponding to the signal decomposition result in each described station symbol signal element in described model library and described background signal tuple; Described signal decomposition result is described first component.
In conjunction with the third possible implementation of first aspect, in the 4th kind of possible implementation, contrast all model component in the model library of described first component and pre-stored, obtain and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result, be specially
Contrast the station symbol signal element corresponding to each model component in described signal decomposition result and described model library, using meeting model component corresponding at least one pre-conditioned station symbol signal element as the first model component mated with described signal decomposition result, using the station symbol corresponding with the first model component that described signal decomposition result is mated as recognition result; Describedly pre-conditionedly be greater than threshold value for response.
In conjunction with the first possible implementation of first aspect, in the 5th kind of possible implementation, described M the channel according to existing, Modling model storehouse, is specially,
For each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer, described M be greater than 1 positive integer;
For each subsample image in described N number of sample image, detect the region, black border in the image of described subsample, determine the ratio of described subsample image, and determine the station symbol region in the image of described subsample; Extract the image border point in described station symbol region;
The image border point exceeding predetermined threshold value in all image borders point utilizing described N number of sample image corresponding forms the station symbol edge image of channel corresponding to described N number of sample image, using described station symbol edge image as model component; Described model component corresponds to channel corresponding to described N number of sample image;
Obtain M model component, described M model component is merged and forms described model library.
In conjunction with the 5th kind of possible implementation of first aspect, in the 6th kind of possible implementation, the first process is carried out to described first image, obtains first component relevant to described station symbol information, be specially,
Based on described first model library, detect the black border in described first image, determine the ratio of described first image, determine the station symbol region in described first image, and extract the marginal point in the described station symbol region in described first image, obtain the first station symbol edge image; Described first station symbol edge image is described first component.
In conjunction with the 6th kind of possible implementation of first aspect, in the 7th kind of possible implementation, contrast all model component in the model library of described first component and pre-stored, obtain and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result, be specially
Contrast the station symbol edge image that described first station symbol edge image is corresponding with model component each in model library, using model component corresponding for station symbol edge image maximum for similarity as the first model component mated with described first station symbol edge image, using the station symbol corresponding with the first model component that described first station symbol edge image mates as recognition result.
Second aspect, discloses a kind of electronic equipment, it is characterized in that, comprising:
Obtain unit, for obtaining the first image, described first image is the two field picture that in TV signal, the first channel transmits, and described first image comprises station symbol information and background information;
Processing unit, for carrying out the first process to described first image, obtains first component relevant to described station symbol information;
Recognition unit, for contrasting the model component in the model library of described first component and pre-stored, obtains and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result.
In conjunction with second aspect, in the first implementation, described electronic equipment also comprises sets up unit,
Described set up unit for, at described processing unit, the first process is carried out to described first image, before obtaining the first component relevant to described station symbol information, according to exist M channel, Modling model storehouse; Described model library comprises M model component, the corresponding channel of each described model component, described M be greater than 1 positive integer.
In conjunction with the first possible implementation of second aspect, in the implementation that the second is possible, described unit of setting up comprises and chooses subelement, signal transacting subelement and constructor unit,
Describedly choose subelement, for for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer;
Described signal transacting subelement is used for, and carries out signal transacting to described N number of sample image, obtains at least one the independent signal unit all existed in described N number of sample image;
Described signal transacting subelement also for, at least one independent signal unit described is normalized, obtains at least one station symbol signal element;
Described constructor unit, for utilizing at least one station symbol signal element component model component described, described model component corresponds to channel corresponding to described N number of sample image;
Described constructor unit is also for, structure discrete cosine signal group;
Described signal transacting subelement also for, described discrete cosine group of functions is normalized, obtains background signal tuple;
Described constructor unit also for, M the model component obtained and described background signal tuple are merged and form described model library.
In conjunction with the implementation that the second of second aspect is possible, in the implementation that the third is possible,
Described processing unit specifically for, based on described model library, sparse signal decomposition is carried out to described first image, obtains described first image and correspond to signal decomposition result in described model library on each described station symbol signal element and described background signal tuple; Described signal decomposition result is described first component.
In conjunction with the third possible implementation of second aspect, in the 4th kind of possible implementation,
Described recognition unit specifically for, contrast the station symbol signal element corresponding to each model component in described signal decomposition result and described model library, using meeting model component corresponding at least one pre-conditioned station symbol signal element as the first model component mated with described signal decomposition result, using the station symbol corresponding with the first model component that described signal decomposition result is mated as recognition result; Describedly pre-conditionedly be greater than threshold value for response.
In conjunction with the first possible implementation of second aspect, in the 5th kind of possible implementation, described unit of setting up also comprises: detection sub-unit and extraction subelement,
Describedly choose subelement, for for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer;
Described detection sub-unit, for for each subsample image in described N number of sample image, detects the region, black border in the image of described subsample, determines the ratio of described subsample image, and determine the station symbol region in the image of described subsample;
Described extraction subelement, for extracting the image border point in described station symbol region;
Described constructor unit is used for, the image border point exceeding predetermined threshold value in all image borders point utilizing described N number of sample image corresponding forms the station symbol edge image of channel corresponding to described N number of sample image, using described station symbol edge image as model component; Described model component corresponds to channel corresponding to described N number of sample image;
Described constructor unit also for, M the model component obtained is merged the described model library of formation.
In conjunction with the 5th kind of possible implementation of second aspect, in the 6th kind of possible implementation,
Described processing unit specifically for, based on described first model library, detect the black border in described first image, determine the ratio of described first image, determine the station symbol region in described first image, and extract the marginal point in the described station symbol region in described first image, obtain the first station symbol edge image; Described first station symbol edge image is described first component.
In conjunction with the 6th kind of possible implementation of second aspect, in the 7th kind of possible implementation,
Described recognition unit specifically for, contrast the station symbol edge image that described first station symbol edge image is corresponding with model component each in model library, using model component corresponding for station symbol edge image maximum for similarity as the first model component mated with described first station symbol edge image, using the station symbol corresponding with the first model component that described first station symbol edge image mates as recognition result.
TV station symbol recognition method provided by the invention and electronic equipment, for the channel existed, Modling model storehouse in advance, this model library comprises the model value of the corresponding station symbol of all channels, when carrying out TV station symbol recognition, sparse signal decomposition carried out to image or extracts the marginal point in image station symbol region, finally the result obtained and model library being contrasted, using station symbol corresponding for the model value that mates in model library as recognition result.With prior art, carry out TV station symbol recognition based on characteristics of image and compare, station symbol and background can be distinguished more exactly, get rid of background to the impact of TV station symbol recognition, improve discrimination.
Accompanying drawing explanation
The schematic flow sheet of the TV station symbol recognition method that Fig. 1 provides for the embodiment of the present invention 1;
The schematic flow sheet of the TV station symbol recognition method that Fig. 2 provides for the embodiment of the present invention 2;
The schematic flow sheet of the TV station symbol recognition method that Fig. 3 provides for the embodiment of the present invention 3;
The structured flowchart of the electronic equipment that Fig. 4 provides for the embodiment of the present invention 4;
Another structured flowchart of the electronic equipment that Fig. 5 provides for the embodiment of the present invention 4.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
Embodiment 1:
Embodiments provide a kind of TV station symbol recognition method, be applied to the electronic equipment that intelligent television, Set Top Box, computer etc. have disposal ability, as shown in Figure 1, said method comprising the steps of:
101, obtain the first image, described first image is the two field picture that in TV signal, the first channel transmits, and described first image comprises station symbol information and background information.
Here the first image obtained comprises station symbol corresponding to described first channel, and the background image of the current display of television image.In specific implementation, TV can obtain the first image by wireless receiving, also can be obtain the first image from Set Top Box.Set Top Box just can obtain the first image by receiving the signal sent from wire cable, satellite antenna, broadband network and terrestrial broadcasting etc.Computer can pass through Network Capture first image.
In addition, M the channel according to existing also is needed before this, Modling model storehouse; Described model library comprises M model component, the corresponding channel of each described model component, described M be greater than 1 positive integer.Here, there is M channel if current, so just need to set up the model library that comprises M station symbol, when identifying a two field picture of the video screen display obtained, the station symbol model of coupling can be found in described model library.Here, can be identification target equipment Modling model storehouse be stored in self in advance, also can be that other external equipment sets up described model library in advance, more described model library is stored in described identification target equipment afterwards.
In specific implementation, can the different model library of correspondence establishment two kinds, there are two kinds of different TV station symbol recognition schemes for these two kinds of model libraries.The method in the first Modling model storehouse comprises: for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer; Signal transacting is carried out to described N number of sample image, obtains at least one the independent signal unit all existed in described N number of sample image; Structure discrete cosine signal group; Utilize at least one independent signal unit component model component described, described model component corresponds to channel corresponding to described N number of sample image.
Like this, just can obtain M model component, and all independent signals unit comprised a described M model component and described discrete cosine signal group are normalized, a described M model component and described discrete cosine signal combination are formed the first model library.
The method in another kind of Modling model storehouse comprises: for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer; For each subsample image in described N number of sample image, detect the region, black border in the image of described subsample, determine the ratio of described subsample image, and determine the station symbol region in the image of described subsample; Extract the image border point in described station symbol region; The image border point exceeding predetermined threshold value in all image borders point utilizing described N number of sample image corresponding forms the station symbol edge image of channel corresponding to described N number of sample image, using described station symbol edge image as model component; Described model component corresponds to channel corresponding to described N number of sample image.
Like this, also just obtain M model component, described M model component is merged the first model library described in formation the.
102, the first process is carried out to described first image, obtain first component relevant to described station symbol information.
After described first image of acquisition, need to process described first image, to obtain component relevant to described station symbol information in described first image, could identify according to the station symbol information of component to described first image obtained like this.As mentioned above, owing to there is the mode in two kinds of Modling model storehouses, therefore when processing described first image, also there are two kinds of analytical methods accordingly.The method in corresponding the first Modling model storehouse described in step 102, sparse signal decomposition is carried out to described first image, obtains described first image corresponding to the signal decomposition result on each described model component in described first model library and described discrete cosine signal; Described signal decomposition result is described first component.The method in corresponding the second Modling model storehouse described in step 102, detect the black border in described first image, determine the station symbol region in the ratio of described first image and described first image, and extract the marginal point in described station symbol region, obtain station symbol edge image; Described station symbol edge image is described first component.Although two kinds of processing modes are different here, core concept is all by distinguishing the station symbol information in described first image and background information, obtains first component relevant to described station symbol information.
103, contrast the model component in the model library of described first component and pre-stored, obtain and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result.
Owing to comprising M model component corresponding to M channel in the model library of pre-stored, if therefore contrast described first component with in described model library important station symbol that just can be corresponding using the model component mated as recognition result.Based on described model library, if adopt the first image described in the method process of signal Its Sparse Decomposition to obtain signal decomposition result in a step 102, here the independent signal unit that described signal decomposition result is corresponding with each model component in described model library is just contrasted, response is greater than the model component of at least one independent signal unit correspondence of threshold value as the first model component mated with described signal decomposition result, and using station symbol corresponding for described first model component as recognition result.Based on described model library, if carry out process acquisition first station symbol edge image by the black border detected in described first image to described first image in a step 102, here the station symbol edge image that described first station symbol edge image is corresponding with each model component in described model library is just contrasted, using model component corresponding for station symbol edge image maximum for similarity as the first model component mated with described first station symbol edge image, using station symbol corresponding for described first model component as recognition result.
TV station symbol recognition method provided by the invention, for the channel existed, Modling model storehouse in advance, this model library comprises the model value of the corresponding station symbol of all channels, when carrying out TV station symbol recognition, sparse signal decomposition carried out to image or extracts the marginal point in image station symbol region, finally the result obtained and model library being contrasted, using station symbol corresponding for the model value that mates in model library as recognition result.With prior art, carry out TV station symbol recognition based on characteristics of image and compare, station symbol and background can be distinguished more exactly, get rid of background to the impact of TV station symbol recognition, improve discrimination.
Embodiment 2:
Embodiments provide a kind of TV station symbol recognition method, be applied to the electronic equipment that intelligent television, Set Top Box, computer etc. have disposal ability, as shown in Figure 2, said method comprising the steps of:
201, for each channel in M channel, N number of sample image is chosen.
Wherein, described N be greater than 1 positive integer.The TV station symbol recognition method that the embodiment of the present invention provides needs to set up the model library that comprises the corresponding station symbol of all channels in advance.Therefore just need to set up station symbol model corresponding to each channel, this station symbol model storage to be used in model library station symbol coupling.When setting up station symbol model for each channel, for ensureing that model is large as much as possible with the similarity of true station symbol, needing, first for this channel chooses N number of sample image, to process this N number of sample image more afterwards, obtaining stable station symbol model.In specific implementation, under guarantee is the prerequisite of same channel, intercepts N and open television image, as sample image.
202, signal transacting is carried out to described N number of sample image, obtain at least one the independent signal unit all existed in described N number of sample image.
Gray processing process is carried out to each sample image, represents this sample image by the mode of vector afterwards, so just can obtain a two-dimensional matrix.Then carry out independent component analysis to this two-dimensional matrix, this analytical method can analyze mutual independently main information component in the middle of matrix.Because these images all comprise same station symbol, so when carrying out independent component analysis, prospect station symbol can be decomposed out by as most important component.Main information component described here is exactly the independent signal unit described in the embodiment of the present invention, and can there is multiple independent signal unit for a station symbol, in addition, independent signal unit can be cosine function.The method is repeatedly applied to N number of sample image that each channel is corresponding, just can obtains the station symbol model of station symbol corresponding to each channel.
203, be normalized at least one independent signal unit described, obtain at least one station symbol signal element, utilize at least one station symbol signal element component model component described, described model component corresponds to channel corresponding to described N number of sample image.
Described in obtaining after carrying out signal transacting to N number of sample image, at least one independent signal unit is normalized, at least one station symbol signal element can be obtained, using the station symbol model of at least one station symbol signal element described as channel corresponding to described N number of sample image.Because at least one independent signal unit described is by carrying out independent component analysis acquisition to described N number of sample image, the station symbol of the channel corresponding with described N number of sample image is extremely similar, therefore can be used for expressing the station symbol of channel corresponding to described N number of sample image completely with the model component that at least one independent signal unit at least one station symbol signal element normalized described is formed.
204, construct discrete cosine signal group, described discrete cosine group of functions is normalized, obtain background signal tuple.
After obtaining station symbol model corresponding to all channels, also need structure one group of discrete cosine function, for rebuilding the complex background of image.In addition, also need, described discrete cosine group of functions is normalized, obtain background signal tuple.The independent signal that the discrete cosine signal constructed is corresponding with any station symbol is not identical, the image of its correspondence is only similar to natural image texture, therefore, when carrying out analyzing and processing to image, the complex background in image except station symbol is expressed by the background signal unit obtained after can organizing cosine function normalization with this.It should be noted that, to the normalization of each independent signal unit and discrete cosine signal can together with carry out, also can carry out as step described in the embodiment of the present invention, not limit the order to the two normalized.
In addition, what each channel obtained in step 202 was corresponding does not have constitutive relations between at least one independent signal unit and discrete cosine signal element, and they are different signal elements.At least one independent signal unit described is for expressing the station symbol in image, and the discrete cosine function of structure is mainly in order to express the complex background in image.
205, obtain M model component, a described M model component and described background signal tuple are merged and forms described model library.
There is M channel if current, all respectively step 201-204 is carried out to this M channel, just can obtain M the model component that M channel is corresponding.Here, all station symbol signal elements that this M model component comprises and background signal tuple be all signal is normalized after result, a described M model component and described background signal tuple are merged component model storehouse.The model library of such foundation had both contained the station symbol model of the corresponding station symbol of all channels, can express again the complex background in image.
206, obtain the first image, described first image is the two field picture that in TV signal, the first channel transmits.
Wherein, described first image comprises station symbol information and background information.It can be random the first image obtained.In specific implementation, TV can obtain the first image by wireless receiving, also can be obtain the first image from Set Top Box.Set Top Box just can obtain the first image by receiving the signal sent from wire cable, satellite antenna, broadband network and terrestrial broadcasting etc.Computer can pass through Network Capture first image.
207, based on model library, sparse signal decomposition is carried out to described first image, obtain described first image corresponding to the signal decomposition result in each described station symbol signal element in described model library and described background signal tuple; Described signal decomposition result is described first component.
In order to identify the station symbol comprising and comprise in the image of complex background, need the model library set up in step 205 by this image does Its Sparse Decomposition.Openness the referring to of signal can represent signal with the linear combination of minority signal element.So-called sparse signal decomposes, in addition sparsity constraints, makes decomposition coefficient comprise neutral element as much as possible, can represent after namely decomposing with the independent signal unit of minority.After decomposing, the station symbol content in image can be decomposed on the model component in model library, and complex background content can be decomposed on discrete cosine signal element.So, the station symbol in image and complex background just can be separated naturally.Here, after carrying out sparse signal decomposition to described image, the signal decomposition result of acquisition is exactly described first component, and then can determine according to described first component combination model storehouse the station symbol that described image is corresponding.To sum up can draw, described signal decomposition result also contains at least one independent signal unit, and described first image is corresponding also can be at least one independent signal unit with the decomposition result of each model component.
208, the station symbol signal element in described signal decomposition result and described model library corresponding to each model component is contrasted, using meeting model component corresponding at least one pre-conditioned station symbol signal element as the first model component mated with described signal decomposition result, using the station symbol corresponding with the first model component that described signal decomposition result is mated as recognition result.
After decomposing, the station symbol content in image can be decomposed on the model component in model library, and namely station symbol content can be decomposed on each station symbol signal element of model library, and background content to be also decomposed in model library on discrete cosine function.Contrast the station symbol signal element corresponding to each model component in described signal decomposition result and described model library afterwards, using meeting model component corresponding at least one pre-conditioned station symbol signal element as the first model component mated with described signal decomposition result, using the station symbol corresponding with the first model component that described signal decomposition result is mated as recognition result.Here because station symbol content can corresponding be expressed by least one station symbol signal element, therefore sparse signal decomposition is being carried out to the first image, the signal decomposition result obtained is except comprising discrete cosine function, also comprise at least one signal element, the station symbol signal element that contrast at least one signal element described is corresponding with model component each in model library again, the matching result therefore obtained also is at least one station symbol signal element that response is greater than threshold value.
TV station symbol recognition method provided by the invention, for the channel existed, Modling model storehouse in advance, this model library comprises the model value of the corresponding station symbol of all channels, when carrying out TV station symbol recognition, sparse signal decomposition carried out to image or extracts the marginal point in image station symbol region, finally the result obtained and model library being contrasted, using station symbol corresponding for the model value that mates in model library as recognition result.With prior art, carry out TV station symbol recognition based on characteristics of image and compare, station symbol and background can be distinguished more exactly, get rid of background to the impact of TV station symbol recognition, improve discrimination.
Embodiment 3:
Embodiments provide a kind of TV station symbol recognition method, as shown in Figure 3, said method comprising the steps of:
301, for each channel in M channel, N number of sample image is chosen.
Here, described M, N are the positive integer being greater than 1.The TV station symbol recognition method that the embodiment of the present invention provides needs to set up the model library that comprises the corresponding station symbol of all channels in advance.Therefore just need to set up station symbol model corresponding to each channel, this station symbol model storage to be used in model library station symbol coupling.When setting up station symbol model for each channel, for ensureing that model is large as much as possible with the similarity of true station symbol, needing, first for this channel chooses N number of sample image, to process this N number of sample image more afterwards, obtaining stable station symbol model.In specific implementation, under guarantee is the prerequisite of same channel, intercepts N and open television image, as sample image.
302, for each subsample image in described N number of sample image, detect the region, black border in the image of described subsample, determine the ratio of described subsample image, and determine the station symbol region in the image of described subsample; Extract the image border point in described station symbol region.
Carrying out black surround detection to each subsample image in N number of sample image, is in fact exactly that detected image the right and left exists how many 0 pixel.Aspect ratio refers to image aspect ratio, can carry out the picture that image stretch obtains standard proportional.In fact, be do not comprise black surround in the middle of the signal that TV station sends, image exists black surround to be caused because operator's Set Top Box and TV resolution arrange different.In fact the picture of sample image and process obtains from TV, be need according to the resolution of picture and the right and left comprise black picture element number process.Generally, giving tacit consent to station symbol region is the upper left corner of picture.But must ensure that picture does not exist black surround, and picture be standard proportional could using the upper left corner of picture as station symbol region.Therefore when processing image, first determining whether this image exists black surround, respective handling just should be taked to eliminate black surround if there is black surround.And then detecting whether the ratio of this image is standard proportional, if not standard proportional, the image that image stretch obtains standard proportional can be carried out.Like this using the upper left corner of image as station symbol region, extract the marginal point in described station symbol region, just can obtain the station symbol edge image of accurate stable.
303, the image border point exceeding predetermined threshold value in all image borders point utilizing described N number of sample image corresponding forms the station symbol edge image of channel corresponding to described N number of sample image, using described station symbol edge image as model component; Described model component corresponds to channel corresponding to described N number of sample image.
Find in practice, compared to the color of station symbol, the marginal point of station symbol is not easy the impact being subject to background content, is applicable to doing the model mating and identify.Therefore, using the image border point that exceedes predetermined threshold value in all image borders point corresponding for the N number of sample image station symbol edge image as channel corresponding to described N number of sample image, give up remaining edge point, just can obtain the stable edge image model of channel corresponding to described N number of sample image.
304, M model component is obtained, by the first model library described in described M model component merging formation the.
There is M channel if current, all respectively step 301-304 is carried out to this M channel, just can obtain M the model component that M channel is corresponding.Afterwards, this M model component component model storehouse is also utilized.The model library of such foundation just contains the stable edge image model of station symbol corresponding to M channel, for mating in TV station symbol recognition, obtains recognition result accurately.
305, based on described model library, detect the black border in described first image, determine the station symbol region in the ratio of described first image and described first image, and extract the marginal point in described station symbol region, obtain the first station symbol edge image.
Wherein, described station symbol edge image be exactly decompose after obtain first component relevant to the station symbol content of described first image.Here, need equally to determine whether this image exists black surround, respective handling just should be taked to eliminate black surround if there is black surround.And then detecting whether the ratio of this image is standard proportional, if not standard proportional, the image that image stretch obtains standard proportional can be carried out.Like this using the upper left corner of image as station symbol region, extract the marginal point in described station symbol region, just can obtain the first station symbol edge image of accurate stable.Again this first station symbol edge image is carried out contrast with station symbol edge image each in model library to mate, just can obtain recognition result accurately.
306, the described first station symbol edge image station symbol edge image corresponding with model component each in model library is contrasted, using model component corresponding for station symbol edge image maximum for similarity as the first model component mated with described first station symbol edge image, using the station symbol corresponding with the first model component that described first station symbol edge image mates as recognition result.
In specific implementation, pointwise coupling is carried out to described first station symbol edge, the described marginal point that the station symbol edge image that each marginal point comprised by described first station symbol edge image is corresponding with model component each in model library comprises carries out contrast one by one and mates, using first model component of model component corresponding for station symbol edge image maximum for similarity as coupling, using station symbol corresponding for this first model component as recognition result.Wherein, similarity can account for edges matched point the percentage that edge of table edge image comprises marginal point and represents.Example, if described first station symbol edge image mates with 80 marginal points in station symbol edge image A in model library, and comprise 100 marginal points in station symbol edge image A, therefore the similarity of described first station symbol edge image and station symbol edge image A is 80 percent.If described first station symbol edge image mates with 90 marginal points in station symbol edge image B in model library, and comprises 100 marginal points in station symbol edge image B, therefore the similarity of described first station symbol edge image and station symbol edge image B is 90 percent.Therefore using model component corresponding for station symbol edge image B larger for similarity as the first model component mated with described first station symbol edge image, using the station symbol corresponding with the first model component that described first station symbol edge image mates as recognition result.
TV station symbol recognition method provided by the invention, for the channel existed, Modling model storehouse in advance, this model library comprises the model value of the corresponding station symbol of all channels, when carrying out TV station symbol recognition, sparse signal decomposition carried out to image or extracts the marginal point in image station symbol region, finally the result obtained and model library being contrasted, using station symbol corresponding for the model value that mates in model library as recognition result.With prior art, carry out TV station symbol recognition based on characteristics of image and compare, station symbol and background can be distinguished more exactly, get rid of background to the impact of TV station symbol recognition, improve discrimination.
Embodiment 4:
Embodiments provide a kind of electronic equipment, as shown in Figure 4, described electronic equipment comprises: obtain unit 401, processing unit 402 and recognition unit 403.
Obtain unit 401, for obtaining the first image, described first image is the two field picture that in TV signal, the first channel transmits, and described first image comprises station symbol information and background information;
Processing unit 402, for carrying out the first process to described first image, obtains first component relevant to described station symbol information.
Recognition unit 403, for contrasting the model component in the model library of described first component and pre-stored, obtains and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result.
As shown in Figure 5, described electronic equipment also comprises sets up unit 404, described set up unit 404 for, at described processing unit 401, first process is carried out to described first image, before obtaining the first component relevant to described station symbol information, according to M the channel existed, Modling model storehouse; Described model library comprises M model component, the corresponding channel of each described model component, described M be greater than 1 positive integer.
Described unit 404 of setting up comprises and chooses subelement, signal transacting subelement and constructor unit.Describedly choose subelement, for for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer.Described signal transacting subelement is used for, and carries out signal transacting to described N number of sample image, obtains at least one the independent signal unit all existed in described N number of sample image.Described signal transacting subelement also for, at least one independent signal unit described is normalized, obtains at least one station symbol signal element.Described constructor unit, for utilizing at least one station symbol signal element component model component described, described model component corresponds to channel corresponding to described N number of sample image.Described constructor unit is also for, structure discrete cosine signal group.Described signal transacting subelement also for, described discrete cosine group of functions is normalized, obtains background signal tuple.Described constructor unit also for, M the model component obtained and described background signal tuple are merged and form described model library.
Described processing unit 402 specifically for, based on described model library, sparse signal decomposition is carried out to described first image, obtains described first image and correspond to signal decomposition result in described model library on each described station symbol signal element and described background signal tuple; Described signal decomposition result is described first component.
Described recognition unit 403 specifically for, contrast the station symbol signal element corresponding to each model component in described signal decomposition result and described model library, using meeting model component corresponding at least one pre-conditioned station symbol signal element as the first model component mated with described signal decomposition result, using the station symbol corresponding with the first model component that described signal decomposition result is mated as recognition result; Describedly pre-conditionedly be greater than threshold value for response.
Described unit 404 of setting up also comprises: detection sub-unit and extraction subelement.Describedly choose subelement, for for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer.Described detection sub-unit, for for each subsample image in described N number of sample image, detects the region, black border in the image of described subsample, determines the ratio of described subsample image, and determine the station symbol region in the image of described subsample.Described extraction subelement, for extracting the image border point in described station symbol region.Described constructor unit is used for, the image border point exceeding predetermined threshold value in all image borders point utilizing described N number of sample image corresponding forms the station symbol edge image of channel corresponding to described N number of sample image, using described station symbol edge image as model component; Described model component corresponds to channel corresponding to described N number of sample image.Described constructor unit also for, M the model component obtained is merged the described model library of formation.
Described processing unit 402 specifically for, based on described first model library, detect the black border in described first image, determine the ratio of described first image, determine the station symbol region in described first image, and extract the marginal point in the described station symbol region in described first image, obtain the first station symbol edge image; Described first station symbol edge image is described first component.
Described recognition unit 403 specifically for, contrast the station symbol edge image that described first station symbol edge image is corresponding with model component each in model library, using model component corresponding for station symbol edge image maximum for similarity as the first model component mated with described first station symbol edge image, using the station symbol corresponding with the first model component that described first station symbol edge image mates as recognition result.
Electronic equipment provided by the invention, for the channel existed, Modling model storehouse in advance, this model library comprises the model value of the corresponding station symbol of all channels, when carrying out TV station symbol recognition, sparse signal decomposition carried out to image or extracts the marginal point in image station symbol region, finally the result obtained and model library being contrasted, using station symbol corresponding for the model value that mates in model library as recognition result.With prior art, carry out TV station symbol recognition based on characteristics of image and compare, station symbol and background can be distinguished more exactly, get rid of background to the impact of TV station symbol recognition, improve discrimination.
One of ordinary skill in the art will appreciate that: all or part of step realizing said method embodiment can have been come by the hardware that program command is relevant, aforesaid program can be stored in a computer read/write memory medium, this program, when performing, performs the step comprising said method embodiment; And aforesaid storage medium comprises: ROM, RAM, magnetic disc or CD etc. various can be program code stored medium.
The above; be only the specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, is anyly familiar with those skilled in the art in the technical scope that the present invention discloses; change can be expected easily or replace, all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should described be as the criterion with the protection range of claim.
Claims (16)
1. a TV station symbol recognition method, is applied to electronic equipment, it is characterized in that, comprising:
Obtain the first image, described first image is the two field picture that in TV signal, the first channel transmits, and described first image comprises station symbol information and background information;
First process is carried out to described first image, obtains first component relevant to described station symbol information;
Contrast the model component in the model library of described first component and pre-stored, obtain and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result.
2. method according to claim 1, is characterized in that, describedly carries out the first process to described first image, and before obtaining the first component relevant to described station symbol information, described method also comprises:
According to M the channel existed, Modling model storehouse; Described model library comprises M model component, the corresponding channel of each described model component, described M be greater than 1 positive integer.
3. method according to claim 2, is characterized in that, described M the channel according to existing, and Modling model storehouse, is specially,
For each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer;
Signal transacting is carried out to described N number of sample image, obtains at least one the independent signal unit all existed in described N number of sample image;
Be normalized at least one independent signal unit described, obtain at least one station symbol signal element, utilize at least one station symbol signal element component model component described, described model component corresponds to channel corresponding to described N number of sample image;
Structure discrete cosine signal group, is normalized described discrete cosine group of functions, obtains background signal tuple;
Obtain M model component, a described M model component and described background signal tuple are merged and forms described model library.
4. method according to claim 3, is characterized in that, carries out the first process to described first image, obtains first component relevant to described station symbol information, is specially,
Based on described model library, sparse signal decomposition is carried out to described first image, obtain described first image corresponding to the signal decomposition result in each described station symbol signal element in described model library and described background signal tuple; Described signal decomposition result is described first component.
5. method according to claim 4, is characterized in that, contrasts all model component in the model library of described first component and pre-stored, obtain and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result, be specially
Contrast the station symbol signal element corresponding to each model component in described signal decomposition result and described model library, using meeting model component corresponding at least one pre-conditioned station symbol signal element as the first model component mated with described signal decomposition result, using the station symbol corresponding with the first model component that described signal decomposition result is mated as recognition result; Describedly pre-conditionedly be greater than threshold value for response.
6. method according to claim 2, is characterized in that, described M the channel according to existing, and Modling model storehouse, is specially,
For each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer, described M be greater than 1 positive integer;
For each subsample image in described N number of sample image, detect the region, black border in the image of described subsample, determine the ratio of described subsample image, and determine the station symbol region in the image of described subsample; Extract the image border point in described station symbol region;
The image border point exceeding predetermined threshold value in all image borders point utilizing described N number of sample image corresponding forms the station symbol edge image of channel corresponding to described N number of sample image, using described station symbol edge image as model component; Described model component corresponds to channel corresponding to described N number of sample image;
Obtain M model component, described M model component is merged and forms described model library.
7. method according to claim 6, is characterized in that, carries out the first process to described first image, obtains first component relevant to described station symbol information, is specially,
Based on described first model library, detect the black border in described first image, determine the ratio of described first image, determine the station symbol region in described first image, and extract the marginal point in the described station symbol region in described first image, obtain the first station symbol edge image; Described first station symbol edge image is described first component.
8. method according to claim 7, is characterized in that, contrasts all model component in the model library of described first component and pre-stored, obtain and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result, be specially
Contrast the station symbol edge image that described first station symbol edge image is corresponding with model component each in model library, using model component corresponding for station symbol edge image maximum for similarity as the first model component mated with described first station symbol edge image, using the station symbol corresponding with the first model component that described first station symbol edge image mates as recognition result.
9. an electronic equipment, is characterized in that, comprising:
Obtain unit, for obtaining the first image, described first image is the two field picture that in TV signal, the first channel transmits, and described first image comprises station symbol information and background information;
Processing unit, for carrying out the first process to described first image, obtains first component relevant to described station symbol information;
Recognition unit, for contrasting the model component in the model library of described first component and pre-stored, obtains and described first point of flux matched first model component, using the station symbol corresponding with described first model component as recognition result.
10. the electronic equipment according to right 9, is characterized in that, also comprises and sets up unit,
Described set up unit for, at described processing unit, the first process is carried out to described first image, before obtaining the first component relevant to described station symbol information, according to exist M channel, Modling model storehouse; Described model library comprises M model component, the corresponding channel of each described model component, described M be greater than 1 positive integer.
11. electronic equipments according to claim 10, is characterized in that, described unit of setting up comprises and chooses subelement, signal transacting subelement and constructor unit,
Describedly choose subelement, for for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer;
Described signal transacting subelement is used for, and carries out signal transacting to described N number of sample image, obtains at least one the independent signal unit all existed in described N number of sample image;
Described signal transacting subelement also for, at least one independent signal unit described is normalized, obtains at least one station symbol signal element;
Described constructor unit, for utilizing at least one station symbol signal element component model component described, described model component corresponds to channel corresponding to described N number of sample image;
Described constructor unit is also for, structure discrete cosine signal group;
Described signal transacting subelement also for, described discrete cosine group of functions is normalized, obtains background signal tuple;
Described constructor unit also for, M the model component obtained and described background signal tuple are merged and form described model library.
12. electronic equipments according to claim 11, is characterized in that,
Described processing unit specifically for, based on described model library, sparse signal decomposition is carried out to described first image, obtains described first image and correspond to signal decomposition result in described model library on each described station symbol signal element and described background signal tuple; Described signal decomposition result is described first component.
13. electronic equipments according to claim 12, is characterized in that,
Described recognition unit specifically for, contrast the station symbol signal element corresponding to each model component in described signal decomposition result and described model library, using meeting model component corresponding at least one pre-conditioned station symbol signal element as the first model component mated with described signal decomposition result, using the station symbol corresponding with the first model component that described signal decomposition result is mated as recognition result; Describedly pre-conditionedly be greater than threshold value for response.
14. electronic equipments according to claim 10, is characterized in that, described unit of setting up also comprises: detection sub-unit and extraction subelement,
Describedly choose subelement, for for each channel in M channel, choose N number of sample image, described N be greater than 1 positive integer;
Described detection sub-unit, for for each subsample image in described N number of sample image, detects the region, black border in the image of described subsample, determines the ratio of described subsample image, and determine the station symbol region in the image of described subsample;
Described extraction subelement, for extracting the image border point in described station symbol region;
Described constructor unit is used for, the image border point exceeding predetermined threshold value in all image borders point utilizing described N number of sample image corresponding forms the station symbol edge image of channel corresponding to described N number of sample image, using described station symbol edge image as model component; Described model component corresponds to channel corresponding to described N number of sample image;
Described constructor unit also for, M the model component obtained is merged the described model library of formation.
15. electronic equipments according to claim 14, is characterized in that,
Described processing unit specifically for, based on described first model library, detect the black border in described first image, determine the ratio of described first image, determine the station symbol region in described first image, and extract the marginal point in the described station symbol region in described first image, obtain the first station symbol edge image; Described first station symbol edge image is described first component.
16. electronic equipments according to claim 15, is characterized in that,
Described recognition unit specifically for, contrast the station symbol edge image that described first station symbol edge image is corresponding with model component each in model library, using model component corresponding for station symbol edge image maximum for similarity as the first model component mated with described first station symbol edge image, using the station symbol corresponding with the first model component that described first station symbol edge image mates as recognition result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410120709.5A CN104954854B (en) | 2014-03-27 | 2014-03-27 | A kind of TV station symbol recognition method and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410120709.5A CN104954854B (en) | 2014-03-27 | 2014-03-27 | A kind of TV station symbol recognition method and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104954854A true CN104954854A (en) | 2015-09-30 |
| CN104954854B CN104954854B (en) | 2019-01-15 |
Family
ID=54169141
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410120709.5A Active CN104954854B (en) | 2014-03-27 | 2014-03-27 | A kind of TV station symbol recognition method and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104954854B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106851397A (en) * | 2017-02-28 | 2017-06-13 | 青岛海信电器股份有限公司 | A kind of station symbol replacing options and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080317123A1 (en) * | 2007-06-21 | 2008-12-25 | Samsung Electronics Co., Ltd. | System and method for still object detection based on normalized cross correlation |
| CN101626465A (en) * | 2009-08-03 | 2010-01-13 | 深圳创维－Rgb电子有限公司 | Image display method of flat television |
| CN101739561A (en) * | 2008-11-11 | 2010-06-16 | 中国科学院计算技术研究所 | TV station logo training method and identification method |
| CN101950366A (en) * | 2010-09-10 | 2011-01-19 | 北京大学 | Method for detecting and identifying station logo |
-
2014
- 2014-03-27 CN CN201410120709.5A patent/CN104954854B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080317123A1 (en) * | 2007-06-21 | 2008-12-25 | Samsung Electronics Co., Ltd. | System and method for still object detection based on normalized cross correlation |
| CN101739561A (en) * | 2008-11-11 | 2010-06-16 | 中国科学院计算技术研究所 | TV station logo training method and identification method |
| CN101626465A (en) * | 2009-08-03 | 2010-01-13 | 深圳创维－Rgb电子有限公司 | Image display method of flat television |
| CN101950366A (en) * | 2010-09-10 | 2011-01-19 | 北京大学 | Method for detecting and identifying station logo |
Non-Patent Citations (1)
| Title |
|---|
| ALEX REIS DOS SANTOS 等: "Real-Time Opaque and Semi-Transparent TV Logos Detection", 《PROC. 5TH INTERNATIONAL INFORMATION AND TELECOMMUNICATION TECHNOLOGIES SYMPOSIUM》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106851397A (en) * | 2017-02-28 | 2017-06-13 | 青岛海信电器股份有限公司 | A kind of station symbol replacing options and device |
| CN106851397B (en) * | 2017-02-28 | 2020-03-10 | 青岛海信电器股份有限公司 | Station caption replacing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104954854B (en) | 2019-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106254933B (en) | Subtitle extraction method and device | |
| CN102509118B (en) | Method for monitoring video retrieval | |
| CN107862315B (en) | Subtitle extraction method, video searching method, subtitle sharing method and device | |
| US9094714B2 (en) | Systems and methods for on-screen graphics detection | |
| US8773430B2 (en) | Method for distinguishing a 3D image from a 2D image and for identifying the presence of a 3D image format by feature correspondence determination | |
| US9076076B1 (en) | Image similarity determination | |
| CN104023249A (en) | Method and device of identifying television channel | |
| KR101395822B1 (en) | Method of selective removal of text in video and apparatus for performing the same | |
| CN104978565B (en) | A kind of pictograph extracting method of universality | |
| CN107480670A (en) | A kind of method and apparatus of caption extraction | |
| CN111124888A (en) | Method and device for generating recording script and electronic device | |
| CN103886027A (en) | Television and method for acquiring article information by scanning visual area | |
| CN105657547A (en) | Detection method and device for similar video and pirated video | |
| CN103544467B (en) | Method for distinguishing and its device are known in a kind of station symbol detection | |
| CN110986889A (en) | High-voltage substation panoramic monitoring method based on remote sensing image technology | |
| CN113435438B (en) | Image and subtitle fused video screen plate extraction and video segmentation method | |
| CN114267029A (en) | Lane line detection method, device, equipment and storage medium | |
| Kaur et al. | Image segmentation based on color | |
| CN104954854A (en) | Station logo identification method and electronic equipment | |
| CN109919164B (en) | User interface object identification method and device | |
| Losson et al. | CFA local binary patterns for fast illuminant-invariant color texture classification | |
| Yang et al. | Caption detection and text recognition in news video | |
| CN110399867B (en) | Text image area identification method, system and related device | |
| CN105120335A (en) | A method and apparatus for processing television program pictures | |
| CN105678298A (en) | Station logo recognition method |
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 |