CN109977859A

CN109977859A - A kind of map logo method for distinguishing and relevant apparatus

Info

Publication number: CN109977859A
Application number: CN201910228432.0A
Authority: CN
Inventors: 黎伟
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2019-03-25
Filing date: 2019-03-25
Publication date: 2019-07-05
Anticipated expiration: 2039-03-25
Also published as: CN109977859B

Abstract

This application discloses a kind of map logo method for distinguishing, comprising: obtains P frame image to be detected at random from video to be detected；Edge detection is carried out to image to be detected in P frame image to be detected, obtains object edge detection set of graphs, it includes M object edge detection figure that object edge, which detects in set of graphs, and each object edge detection figure is obtained after merging to P edge detection graph；Set of graphs, which is detected, according to object edge determines icon area；The icon in video to be detected is determined according to P frame image to be detected and icon area；Icon is matched with preset icon set, obtains the icon-based programming result of video to be detected, wherein preset icon set includes at least one preset icon.Disclosed herein as well is devices.The application can increase the diversity of logo background variation using stochastical sampling, reach better sample effect, suitable for the detection to static logo and dynamic logo, to promote recognition accuracy.

Description

A kind of map logo method for distinguishing and relevant apparatus

Technical field

This application involves field of image processing more particularly to a kind of map logo method for distinguishing and relevant apparatus.

Background technique

As multitude of video information pours in the actual life of people, the detection of video logo is had become to source video sequence analysis One effective means.The publisher of determination video that can be relatively easy by the logo of video passes through the logo energy in program Navigate to specific program.These important semantic informations are for providing accurate video search.In addition, passing through detection video program In logo can also remove advertising segment, and then improve ornamental value.In Video security field, video logo detection technique can be with It is effective to determine source video sequence.

Currently, can be using the side based on optical character identification (Optical Character Recognition, OCR) Method is detected and is identified to logo.When user switches program, it will appear the logo with text on video pictures.? It shows that there is also the delays of a bit of time before logo, OCR Text region can be carried out to logo within the section time, based on text Word directly differentiates logo type.

However, being continuously increased with video type, more and more logos are emerged.These logos often have one A little special effects, for example, A class logo can constantly be shaken, the subtitle of B class logo disappears again after rolling appearance gradually, and alternately Appear in the upper left corner and the lower right corner of video, the image and text of C class logo all can constantly rotate etc..The logo of these types It changes over time and changes, be referred to as cardon logo.Cardon logo is identified based on the method for OCR, it is quasi- True rate is lower, and is not suitable for the dynamic logo of pure image, causes adaptation range smaller.

Summary of the invention

The embodiment of the present application provides a kind of map logo method for distinguishing and relevant apparatus, on the one hand utilizes stochastical sampling energy The diversity for enough increasing the variation of logo background, reaches better sample effect, and on the other hand, multi-frame video image is merged Dynamic logo can be become to metastable static logo, and then static logo is identified, be thus suitable for static state The detection of logo and dynamic logo, to promote recognition accuracy.

In view of this, the application first aspect provides a kind of map logo method for distinguishing, comprising:

Obtain P frame image to be detected at random from video to be detected, wherein the video to be detected includes Q frame video figure Picture, the Q are the integer greater than 1, and the P is the integer more than or equal to 1, and less than or equal to the Q；

Edge detection is carried out to image to be detected in described P frame image to be detected, obtains object edge detection set of graphs, It wherein, include M object edge detection figure in the object edge detection set of graphs, each object edge detection figure is to P It is obtained after edge detection graph fusion, the M is the integer more than or equal to 1；

Set of graphs, which is detected, according to the object edge determines icon area；

The icon in the video to be detected is determined according to described P frame image to be detected and the icon area；

The icon is matched with preset icon set, obtain the icon-based programming of the video to be detected as a result, its In, the preset icon set includes at least one preset icon.

The application second aspect provides a kind of icon-based programming device, comprising:

Module is obtained, for obtaining P frame image to be detected at random from video to be detected, wherein the video to be detected Including Q frame video image, the Q is integer greater than 1, the P be more than or equal to 1, and it is whole less than or equal to the Q Number；

Detection module, image to be detected in described P frame image to be detected for obtaining to the acquisition module carry out Edge detection obtains object edge detection set of graphs, wherein includes M object edge in the object edge detection set of graphs Detection figure, each object edge detection figure are obtained after merging to P edge detection graph, and the M is whole more than or equal to 1 Number；

Determining module, the object edge detection set of graphs for being detected according to the detection module determine logo Region；

The determining module is also used to be determined according to described P frame image to be detected and the icon area described to be checked Survey the icon in video；

Identification module, the icon for determining the determining module are matched with preset icon set, are obtained The icon-based programming result of the video to be detected, wherein the preset icon set includes at least one preset icon.

In a kind of possible design, in the first implementation of the second aspect of the embodiment of the present application, the figure Marking identification device further includes division module and extraction module；

The division module, for being carried out in the detection module to image to be detected in described P frame image to be detected Edge detection, before obtaining object edge detection set of graphs, to described every frame image to be detected in described P frame image to be detected It is divided, obtains multiple images region corresponding to described every frame image to be detected；

The extraction module is more corresponding to described every frame image to be detected for dividing from the division module M image-region corresponding to described every frame image to be detected is extracted in a image-region, wherein the M image-region is used In progress edge detection.

In a kind of possible design, in second of implementation of the second aspect of the embodiment of the present application,

The detection module, specifically for the target figure in the M image-region to described every frame image to be detected As region progress edge detection, P edge detection graph corresponding to the object region is obtained, wherein the target figure As region belongs to any one image-region in the M image-region；

The P edge detection graph according to corresponding to the object region, determines corresponding to the object region Object edge detect figure；

When getting object edge detection figure corresponding to the M image-region, the object edge detection is obtained Set of graphs.

In a kind of possible design, in the third implementation of the second aspect of the embodiment of the present application,

The determining module, specifically for the object edge detection set of graphs in each object edge detect figure into Column hisgram statistics, obtains M statistical result, wherein the statistics with histogram is used to detect object edge figure progress horizontal The statistics of direction and vertical direction；

Judge whether each statistical result in the M statistical result meets logo extracted region condition respectively；

If at least one statistical result in the M statistical result meets the statistical threshold, it is determined that the target There are logo regions for edge detection set of graphs；

If meeting the statistical threshold without statistical result in the M statistical result, it is determined that the object edge inspection Logo region is not present in mapping set.

In a kind of possible design, in the 4th kind of implementation of the second aspect of the embodiment of the present application,

The determining module, specifically for determining figure to be matched in the logo region according to described P frame image to be detected The image score value of picture；

If the image score value of the image to be matched is greater than or equal to logo image threshold, it is determined that the video to be detected In there are the logo images.

In a kind of possible design, in the 5th kind of implementation of the second aspect of the embodiment of the present application,

The identification module, specifically for obtaining the local feature set to be matched of the logo image, wherein it is described to Matching local feature set includes at least one local feature to be matched；

Obtain the local feature set of each preset icon in the preset icon set, wherein the local feature collection Closing includes at least one local feature；

It is close by k according to the local feature set to be matched and the local feature set of each preset icon Adjacent algorithm determines candidate's logo image collection from the preset icon set, wherein candidate's logo image collection includes N A candidate's logo image, the N are the integer more than or equal to 1；

Each candidate logo image in the logo image and the candidate logo image collection is compared, is obtained The pairing point set of each candidate logo image, wherein the pairing point set includes at least one match point, described to match To the characteristic point of expression candidate logo image and the logo image successful matching；

Part according to the pairing point set of each candidate logo image and each candidate logo image is special Collection is closed, and N number of similarity score is calculated；

According to the maximum value of similarity score in N number of similarity score, from the candidate logo image collection really The Target Station logo image of fixed video to be detected.

In a kind of possible design, in the 6th kind of implementation of the second aspect of the embodiment of the present application,

The identification module, a part to be matched for being specifically used for 1) obtaining in the local feature set to be matched are special Sign；

2) according to the local feature to be matched, obtained from the local feature set of each preset icon with it is described K nearest candidate feature of local feature to be matched, wherein the K is the integer more than or equal to 1；

Step 1) is repeated to step 2), until it is to be matched to get each of described local feature set to be matched The candidate feature of local feature；

According to the candidate feature of each of the local feature set to be matched local feature to be matched, the time is obtained Channel selection logo image set.

In a kind of possible design, in the 7th kind of implementation of the second aspect of the embodiment of the present application,

The identification module, specifically for by each of logo image local feature to be matched and each time Each local feature of channel selection logo image is matched, and projection matrix is obtained, wherein the projection matrix indicates the logo figure As the position coordinates after projection；

According to the logo image and the projection matrix, the pairing point set of each candidate logo image is determined It closes.

In a kind of possible design, in the 8th kind of implementation of the second aspect of the embodiment of the present application,

The identification module, specifically for calculating similarity score in the following way:

Wherein, the score indicates that the similarity score, the A indicate the pairing point set institute of candidate logo image Corresponding area union, the B indicate area union corresponding to the local feature set of the candidate logo image.

In a kind of possible design, in the 9th kind of implementation of the second aspect of the embodiment of the present application, the figure Marking identification device further includes processing module and extraction module；

The acquisition module is also used to match the icon with preset icon set in the identification module, obtain Before the icon-based programming result for taking the video to be detected, video collection to be processed is obtained, wherein the video collection to be processed In include at least one video to be processed；

The detection module, each of described video collection to be processed for being also used to obtain the acquisition module is wait locate Reason video is detected, and obtains logo image collection to be processed, wherein the logo image collection to be processed includes at least one Logo image to be processed, at least one of described logo image collection to be processed logo image to be processed correspond to the same mark Know；

The processing module, in the logo image collection to be processed for being detected to the detection module to Treatment bench logo image is handled, and the preset icon set is obtained；

The extraction module, each of described preset icon set for handling the processing module are default Icon carries out feature extraction, obtains the local feature set of each preset icon, wherein the local feature set includes At least one local feature, the local feature include characteristic point position coordinate and characteristic information.

In a kind of possible design, in the tenth kind of implementation of the second aspect of the embodiment of the present application,

The processing module, specifically for when receiving the first process instruction, according to first process instruction from institute State the first logo image to be processed of rejecting in logo image collection to be processed, wherein described in carrying in first process instruction The mark of first logo image to be processed；

When receiving second processing instruction, instructed according to the second processing in the logo image collection to be processed The second logo image to be processed be adjusted, obtain the preset icon in the preset icon set, wherein at described second The mark of the described second logo image to be processed is carried in reason instruction.

The application third aspect provides a kind of server, comprising: memory, transceiver, processor and bus system；

Wherein, the memory is for storing program；

The processor is used to execute the program in the memory, includes the following steps:

The icon is matched with preset icon set, obtain the icon-based programming of the video to be detected as a result, its In, the preset icon set includes at least one preset icon；

The bus system is for connecting the memory and the processor, so that the memory and the place Reason device is communicated.

The application fourth aspect provides a kind of terminal device, comprising: memory, transceiver, processor and bus system；

Wherein, the memory is for storing program；

The 5th aspect of the application provides a kind of computer readable storage medium, in the computer readable storage medium It is stored with instruction, when run on a computer, so that computer executes method described in above-mentioned various aspects.

As can be seen from the above technical solutions, the embodiment of the present application has the advantage that

In the embodiment of the present application, a kind of map logo method for distinguishing is provided, it is necessary first to obtain at random from video to be detected P frame image to be detected is taken, edge detection then is carried out to image to be detected in P frame image to be detected, obtains object edge inspection Mapping set, it includes M object edge detection figure that object edge, which detects in set of graphs, and each object edge detection figure is to P Obtained after edge detection graph fusion, set of graphs detected according to object edge and determines icon area, according to P frame image to be detected with And icon area determines the icon in video to be detected, and icon is matched with preset icon set, obtains video to be detected Icon-based programming result, wherein preset icon set includes at least one preset icon.By the above-mentioned means, stochastical sampling is more Frame video image is merged, and on the one hand can be increased the diversity of logo background variation using stochastical sampling, be reached better Sample effect, on the other hand, multi-frame video image, which is carried out fusion, can become dynamic icon metastable static icon, And then static icon is identified, thus it is suitable for the detection to static icon and dynamic icon, so that it is quasi- to promote identification True rate.

Detailed description of the invention

Fig. 1 is a configuration diagram of icon-based programming system in the embodiment of the present application；

Fig. 2 is an overall flow schematic diagram of icon-based programming system in the embodiment of the present application；

Fig. 3 is map logo method for distinguishing one embodiment schematic diagram in the embodiment of the present application；

Fig. 4 is one embodiment schematic diagram of the M image-region of image to be detected in the embodiment of the present application；

Fig. 5 is another embodiment schematic diagram of the M image-region of image to be detected in the embodiment of the present application；

Fig. 6 is another embodiment schematic diagram of the M image-region of image to be detected in the embodiment of the present application；

Fig. 7 is one embodiment schematic diagram for generating object edge in the embodiment of the present application and detecting figure；

Fig. 8 is to detect one embodiment schematic diagram that figure carries out statistics with histogram to object edge in the embodiment of the present application；

Fig. 9 is a schematic diagram of logo image in the embodiment of the present application；

Figure 10 is one embodiment schematic diagram that similarity compares in the embodiment of the present application；

Figure 11 is the schematic diagram of the same logo difference display format in the embodiment of the present application；

Figure 12 is icon-based programming device one embodiment schematic diagram in the embodiment of the present application；

Figure 13 is another embodiment schematic diagram of icon-based programming device in the embodiment of the present application；

Figure 14 is another embodiment schematic diagram of icon-based programming device in the embodiment of the present application；

Figure 15 is server one embodiment schematic diagram in the embodiment of the present application；

Figure 16 is terminal device one embodiment schematic diagram in the embodiment of the present application.

Specific embodiment

The embodiment of the present application provides a kind of map logo method for distinguishing and relevant apparatus, on the one hand utilizes stochastical sampling energy The diversity for enough increasing the variation of icon background, reaches better sample effect, and on the other hand, multi-frame video image is merged Dynamic icon can be become to metastable static icon, and then static icon is identified, be thus suitable for static state The detection of icon and dynamic icon, to promote recognition accuracy.

The description and claims of this application and term " first ", " second ", " third ", " in above-mentioned attached drawing The (if present)s such as four " are to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should manage The data that solution uses in this way are interchangeable under appropriate circumstances, so that embodiments herein described herein for example can be to remove Sequence other than those of illustrating or describe herein is implemented.In addition, term " includes " and " corresponding to " and their times What is deformed, it is intended that cover it is non-exclusive include, for example, contain the process, method of a series of steps or units, system, Product or equipment those of are not necessarily limited to be clearly listed step or unit, but may include be not clearly listed or for The intrinsic other step or units of these process, methods, product or equipment.

It should be understood that TV station symbol recognition method provided herein can be used for artificial intelligence field, video image analysis neck Domain and searching field etc..Logo (logo) refers to the mark and mark of TV station or video website, usually suspension and video The corner locations such as the upper left corner, the upper right corner or the lower right corner.Logo detection and identification be using image procossing and identification technology automatically from Logo image is obtained in video, and is automatically classified into certain known logo.It is at full speed with TV tech and Internet technology Development, the program up to tens or even set up to a hundred of TV stations at different levels are transferred to by modes such as satellite and microwaves, thousands of Video content be transferred to by network mode on the terminal device that user uses.

In order to carry out real-time monitoring to TV programme and network video, the application proposes a kind of map logo method for distinguishing, It is all higher to the verification and measurement ratio of static logo and dynamic logo, and manual operation is not necessarily in detection process, save manpower money Source.It is well known that ensureing the safety of TV programme and network video, guarding against illegal inter-cut and interference are the one of safe broadcast The very important task of item.It can be monitored and illegally be intercutted and illegal signals invasion by the identification to logo, and then according to platform It marks testing result and Realtime Alerts is carried out to TV programme by all kinds of means and network video, illegally intercut and enter to be effectively prevented It invades, reduces the labor intensity of staff, and avoid operation error.

Specifically, the illegal video intercepting system with identification is detected based on logo, it can be by artificially collecting logo Related illegal video, and index database is generated, automatic logo detection and identification are carried out to the video for needing to audit on network, if view Logo in frequency hit index database, then intercept the video, the video of otherwise letting pass.In addition, network video can be according to logo point Class collects the video (every class only needs a small amount of video) with logo for needing to classify, establishes logo index database, to a large amount of on network The video not marked carries out TV station symbol recognition automatically, and classifies by logo.

In order to make it easy to understand, this method is applied to icon shown in FIG. 1 present applicant proposes a kind of map logo method for distinguishing Identifying system, referring to Fig. 1, Fig. 1 is a configuration diagram of icon-based programming system in the embodiment of the present application, as shown, TV station symbol recognition method provided by the present application can be used for server, can be used for terminal device, below in conjunction with Fig. 2, and with Applied to being illustrated for server.Referring to Fig. 2, Fig. 2 is an entirety of icon-based programming system in the embodiment of the present application Flow diagram, as shown, the process of entire TV station symbol recognition can be divided into two parts, one is to build library process offline, separately One detects and identifies for online logo.During building library offline, specifically comprise the following steps:

In step A1, server obtains the associated video artificially collected, carries logo in these videos, wherein these views Frequency is the video played on the client, it should be noted that client deployment is on terminal device, wherein terminal device packet Contain but be not limited only to tablet computer, laptop, palm PC, mobile phone, interactive voice equipment and PC (personal Computer, PC), herein without limitation；

In step A2, server detects the logo in video, that is, corresponding logo is extracted from video；

In step A3, server can automatically be cleaned these logos after extracting, and the purpose of cleaning is rejecting one The image for being not belonging to logo a bit in practical applications, can also manually reject the image of non-logo, anyway, purpose certainly It is provided to get logo image；

In step A4, after server gets video, logo image can also be obtained in a manual manner completely, i.e., Manually cut out each logo image；

In step A5, after obtaining logo image corresponding to video, local shape factor is carried out to logo image, than Such as Scale invariant features transform (Scale-Invariant Feature Transform, SIFT) feature extraction；

In step A6, quick indexing is established to the feature after extraction, such as to the quick indexing of SIFT feature, i.e., given one A SIFT feature quickly finds other SIFT features nearest with this SIFT feature distance by this quick indexing, until This builds the completion of library process offline.

In the detection of online logo with identification process, specifically comprise the following steps:

In step B1, video to be detected is obtained first；

In step B2, server is also required to for video to be detected and carries out logo detection, so that logo image is obtained, this When, logo is detected as detecting automatically, extracts logo image without artificial；

In step B3, server carries out the extraction of local feature to the logo image detected, it is to be understood that feature The characteristic extraction procedure type of extraction process and step A5, is not repeated herein；

In step B4, server is special by the local feature in video to be detected and the part in local feature quick indexing library Sign is slightly matched；

In step B5, server retrieves most like N number of logo in the local feature quick indexing library of above-mentioned foundation Image；

In step B6, the similarity that server does 1 to 1 to this N number of logo image is compared, and it is highest to select similarity score Logo image；

In step B7, if the similarity score is greater than given threshold value, the TV station symbol recognition result exported is A logo, , whereas if the similarity score is less than given threshold value, then illustrate the logo of the video to be detected not in logo library.

In conjunction with above-mentioned introduction, map logo method for distinguishing in the application will be introduced below, referring to Fig. 3, the application Map logo method for distinguishing one embodiment includes: in embodiment

101, P frame image to be detected is obtained at random from video to be detected, wherein video to be detected includes Q frame video figure Picture, Q are the integer greater than 1, and P is the integer more than or equal to 1, and less than or equal to Q；

In the present embodiment, icon-based programming device obtains video to be detected, which can be network video, or It is TV programme etc., herein without limitation.Include Q frame video image in video to be detected, is randomly choosed from Q frame video image P frame image to be detected can therefrom randomly choose 64 video images work for example, there is 1000 frame video images in video to be detected For image to be detected.

Wherein, there are many ways to obtaining P frame image to be detected at random, for example, it is to be detected using random function generation P frame Image specifically can be rand () function etc. in rand () function or MATLAB of C language.Random number is to use certainty Algorithm calculate from [0,1] equally distributed random number sequence, there is the statistical nature similar to random number, such as uniformity With independence etc..When calculating pseudo random number, if the initial value used is constant, then the number sequence of random number is also constant.Random number can Largely to be generated with computer.

A frame image is often extracted from video to be detected in the prior art, or extracts the figure of continuous a period of time Picture, it is this that frame method is taken to be possible to take the video image less than logo is had at all for cardon logo, this is because some are regarded Frequency is that logo just occur after playing a period of time, and can greatly drop if taking the full video image in video to be detected The efficiency of low system decodes all images video council and consumes a large amount of time for longer video.And the side of frame is taken at equal intervals Method may all obtain the video image at the logo disappearance moment for some cardon logos.Therefore, using random Frame is taken to will increase the variation of logo background, the superimposed logo of multiframe can reduce the influence of background.

102, edge detection is carried out to image to be detected in P frame image to be detected, obtains object edge detection set of graphs, It wherein, include M object edge detection figure in object edge detection set of graphs, each object edge detection figure is to P edge It is obtained after the fusion of detection figure, M is the integer more than or equal to 1；

In the present embodiment, icon-based programming device carries out edge detection to every frame image to be detected in P frame image to be detected, Wherein, every frame image to be detected includes edge detection graph, and edge detection graph is to obtain in image to be detected by edge detection Image.Then, edge detection graph in every frame image to be detected is merged, i.e., P frame border detection figure is merged, thus Obtain object edge detection figure.It is understood that in a frame image to be detected, there may be multiple edge detection graphs, for phase It is merged with P edge detection graph on position, object edge detection figure can be obtained.When to P side on each position After the completion of edge detection figure all merges, object edge detection set of graphs can be obtained.

The purpose of edge detection is the apparent point of brightness change in reference numbers image, and the edge detection of image can be substantially Degree ground reduces data volume, and eliminates it is considered that incoherent information, remains the important structure attribute of image.Have perhaps Multi-method is used for edge detection, including based on the one kind searching one kind and passing through based on zero.Wherein, the method based on lookup passes through The maximum and minimum value in image first derivative is found to detect boundary, usually by boundary alignment in the maximum direction of gradient. And the method passed through based on zero is passed through by searching image second order derivative zero to find boundary, usually Laplce (Laplacian) zero crossing that zero crossing or nonlinear difference indicate.

103, set of graphs is detected according to object edge and determines icon area；

In the present embodiment, icon-based programming device carries out each object edge detection figure in object edge detection set of graphs Detection, judges whether there is icon area, wherein the icon area specifically can be logo region.If detecting the presence of logo Region, then icon-based programming device continues to test in video to be detected according to P frame image to be detected and logo region and whether there is Logo image., whereas if being not detected there are logo region, it may be considered that not including logo image in video to be detected.

104, the icon in video to be detected is determined according to P frame image to be detected and icon area；

In the present embodiment, icon-based programming device detects video to be detected according to P frame image to be detected and icon area In icon, which specifically can be logo image.

105, icon is matched with preset icon set, obtains the icon-based programming result of video to be detected, wherein is pre- If icon set includes at least one preset icon.

In the present embodiment, according to P frame image to be detected and logo region, detect in logo region is icon-based programming device No there are logo images to be detected, if there is logo image, then can be by logo image and preset icon set Each preset icon is matched, and selects icon of the highest preset icon of matching degree as video to be detected according to matching result Recognition result.If determination does not match after logo image is matched with each preset icon a in preset icon set Preset icon, then can using TV station symbol recognition failure result as the icon-based programming result of the video to be detected.

Optionally, on the basis of above-mentioned Fig. 3 corresponding embodiment, map logo method for distinguishing provided by the embodiments of the present application In first alternative embodiment, edge detection is carried out to every frame image to be detected in P frame image to be detected, obtains object edge Before detecting set of graphs, can also include:

Every frame image to be detected in P frame image to be detected is divided, is obtained corresponding to every frame image to be detected Multiple images region；

M figure corresponding to every frame image to be detected is extracted from multiple images region corresponding to every frame image to be detected As region, wherein M image-region is for carrying out edge detection.

In the present embodiment, a kind of division mode of image to be detected will be introduced, for entire image to be detected, into Before row edge detection, icon-based programming device can be divided the image to be detected, for the ease of introducing, referring to Fig. 4, Fig. 4 is one embodiment schematic diagram of the M image-region of image to be detected in the embodiment of the present application, as shown in the figure, it is assumed that by P Every frame image to be detected in frame image to be detected is divided into 4 × 4 equal portions, and every a as image-region is to be detected Image includes 16 image-regions.It is understood that in practical applications, other division proportions can also be designed, such as 5 × 5 equal portions are divided into, or are divided into 4 × 5 equal portions, only one signal, is not construed as to the application's herein It limits.

Next, icon-based programming device from multiple images region corresponding to every frame image to be detected, extracts every respectively M image-region corresponding to frame image to be detected, please continue to refer to Fig. 4, it is assumed that image to be detected is divided into 16 images Region needs that several image-regions is selected to carry out subsequent processing from this 16 image-regions at this time.In view of logo image Several corners of video pictures are frequently found in, are upper left apex angle (No. 1 image-region i.e. as shown in Figure 4), upper right top respectively Angle (No. 2 image-regions i.e. as shown in Figure 4), lower-left apex angle (No. 3 image-regions i.e. as shown in Figure 4) and bottom right apex angle (No. 4 image-regions i.e. as shown in Figure 4).Therefore, the image-region of four apex angles can be only taken to carry out subsequent operation, i.e. M takes 4.It is understood that in practical applications, other M values can also be taken, and the M image-region selected can also be to be detected In the other positions of image.For example, referring to Fig. 5, Fig. 5 is the M image-region of image to be detected in the embodiment of the present application Another embodiment schematic diagram, as shown in the figure, it is assumed that select the image-region on M of image to be detected along part, that is, distinguish For No. 1 image-region shown in fig. 5, No. 2 image-regions shown in Fig. 4, No. 3 image-regions shown in Fig. 4 and shown in Fig. 44 Number image-region, M is set as 4 at this time.

For another example, referring to Fig. 6, Fig. 6 is another reality of the M image-region of image to be detected in the embodiment of the present application It applies illustration to be intended to, as shown in the figure, it is assumed that select the image-region of the M upper right of image to be detected, No. 1 figure as shown in fig. 6 As region, M is set as 1 at this time.

Icon-based programming device, can be to M image after extracting M image-region corresponding to every frame image to be detected Each image-region in region carries out edge detection.Assuming that extracting 4 in every frame image to be detected there are 64 frame image to be detected A image-region carries out edge detection, then needs altogether to carry out edge detection to 256 image-regions.

Secondly, a kind of division mode of image to be detected is provided in the embodiment of the present application, to P frame image to be detected In every frame image to be detected carry out edge detection, obtain object edge detection set of graphs before, can be to P frame image to be detected In every frame image to be detected divided, obtain multiple images region corresponding to every frame image to be detected, then wait for from every frame M image-region corresponding to every frame image to be detected is extracted in multiple images region corresponding to detection image.By above-mentioned Mode reasonably divides image to be detected, forms multiple operable regions, is convenient for subsequent operation, to be promoted The flexibility and operability of scheme.

Optionally, on the basis of above-mentioned Fig. 3 corresponding one embodiment, icon-based programming provided by the embodiments of the present application Second alternative embodiment of method in, in P frame image to be detected every frame image to be detected carry out edge detection, obtain mesh Edge detection set of graphs is marked, may include:

Edge detection is carried out to the object region in the M image-region of every frame image to be detected, obtains target figure The P edge detection graph as corresponding to region, wherein object region belongs to any one image in M image-region Region；

The P edge detection graph according to corresponding to object region determines target side corresponding to object region Edge detection figure；

When getting object edge detection figure corresponding to M image-region, object edge detection set of graphs is obtained.

In the present embodiment, it will introduce and how to generate object edge detection set of graphs, for ease of description, referring to Fig. 7, figure 7 is generate one embodiment schematic diagram that object edge detects figure in the embodiment of the present application, as shown, with P frame mapping to be checked It is introduced for wherein frame image to be detected as in, it is to be understood that the processing mode of other image to be detected It is similar, therefore be not repeated herein.Assuming that M is 4, i.e. M image-region be No. 1 image-region shown in Fig. 7, No. 2 image-regions, No. 3 image-regions and No. 4 image-regions.It is illustrated by taking any one image-region in M image-region as an example, the figure As region is object region, it is to be understood that other image-region processing modes and target in M image-region The processing mode of image-region is similar, therefore is not repeated herein.Assuming that object region is No. 1 image-region shown in Fig. 7, At this point, carrying out edge detection to No. 1 image-region, edge detection graph is obtained, due to sharing P frame image to be detected, every frame is to be checked Altimetric image all carries out edge detection to No. 1 image-region, therefore, available P edge detection graph, i.e., No. 1 figure shown in Fig. 7 The P edge detection graph as corresponding to region.P edge detection graph corresponding to No. 1 image-region is overlapped, obtains 1 The detection figure of object edge corresponding to number image-region, i.e., object edge detection figure A as shown in Figure 7.To No. 2 image-regions, 3 Number image-region and No. 4 image-regions are processed similarly, and the corresponding object edge detection figure of No. 2 image-regions is thus obtained B, No. 3 image-regions corresponding object edge detection figure C, the corresponding object edge detection figure D of No. 4 image-regions.When getting M When the detection figure of object edge corresponding to a image-region, object edge detection set of graphs is obtained.That is, when getting mesh When marking edge detection graph A, object edge detection figure B, object edge detection figure C and object edge detection figure D, that is, think to obtain mesh Mark edge detection set of graphs.

The task of edge detection is the set in order to find the pixel with Spline smoothing or roof variation.Wherein, Edge is the line of demarcation of different zones, is the set for the pixel that surrounding pixel has significant change, and has amplitude and two, direction attribute. It is generally acknowledged that profile is the description to the integral edge of object, marginal point connects composition profile one by one.Edge can be One Duan Bianyuan, and profile is usually complete.The application use edge detection method include but be not limited only to Canny operator, Roberts operator, Sobel operator, Prewitt operator, Kirsch operator and Robinson operator.For example it is calculated using Canny Son does edge detection, and available edge detection graph, specifically, the edge detection process based on Canny operator are as follows:

Color image is converted to gray level image by the first step；

Second step carries out Gaussian Blur to image；

Third step calculates image gradient, calculates image border amplitude and angle according to gradiometer, and the inspection of differential edge can be used Son is calculated to calculate gradient magnitude direction；

4th step carries out compression process using non-peak signal, i.e. progress edge thinning；

5th step, dual threshold edge connection processing；

6th step, binary image export result.

Again, in the embodiment of the present application, a kind of mode obtaining object edge detection set of graphs is provided, every frame is waited for first Object region in M image-region of detection image carries out edge detection, obtains P corresponding to object region Edge detection graph, then the P edge detection graph according to corresponding to object region, determines corresponding to object region Object edge inspection can be obtained when getting object edge detection figure corresponding to M image-region in object edge detection figure Mapping set.By the above-mentioned means, the image-region that part can be extracted from image to be detected carries out edge detection, without Edge detection is carried out for entire image to be detected, to reduce calculation amount, and improve detection efficiency.

Optionally, on the basis of above-mentioned Fig. 3 corresponding embodiment, map logo method for distinguishing provided by the embodiments of the present application In third alternative embodiment, set of graphs is detected according to object edge and determines icon area, may include:

Statistics with histogram is carried out to each object edge detection figure in object edge detection set of graphs, obtains M statistics As a result, wherein statistics with histogram is used to detect the statistics that figure carries out horizontal direction and vertical direction to object edge；

Judge whether each statistical result in M statistical result meets logo extracted region condition respectively；

If at least one statistical result in M statistical result meets logo extracted region condition, it is determined that object edge Detecting set of graphs, there are logo regions；

If meeting logo extracted region condition without statistical result in M statistical result, it is determined that object edge detection figure Logo region is not present in set.

In the present embodiment, it will introduce and a kind of detect whether that there are the methods in logo region, it is assumed that object edge detects atlas There are M object edge detection figures in conjunction, then detecting figure to each object edge carries out statistics with histogram, based on given system Threshold value is counted, the part that statistical result is less than statistical threshold is rejected along left and right side edge and lower edges, to obtain more compact Logo region.In practical applications, it needs to be carried out operation as above to each object edge detection figure, if M target side In the corresponding statistical result of edge detection figure, there are at least one statistical results to meet logo extracted region condition, it is determined that symbol Close the logo region of subsequent processing.

Specifically, for the ease of introducing, referring to Fig. 8, Fig. 8 is to detect figure to object edge in the embodiment of the present application to carry out One embodiment schematic diagram of statistics with histogram, as shown, by taking an object edge detects figure as an example, it is assumed that the object edge Detection figure is the image of 5 × 5 pixels, wherein black indicates 0, and canescence indicates 1, counts greyish white color pixel in the horizontal direction Number is followed successively by 1,4,3,5 and 2, and the number for vertically counting greyish white color pixel is followed successively by 2,4,2,5 and 2, it is assumed that statistics Threshold value is 3, then rejects the pixel less than 3 along left and right edges, i.e. the right and left value is undesirable for 2 two column, vertically side Edge rejects the region less than 3, i.e. outlier is 1 and 2 two rows, finally obtains the logo region of centre 3 × 3.Then, determining should Statistical result meets logo extracted region condition, i.e., there are logo images for the object edge detection figure.If statistical result is small In statistical threshold, then it is assumed that logo image is not present in object edge detection figure.

It is understood that although the third column in Fig. 8 are 2, because when encountering secondary series and the 4th column just It stopped, therefore, third column will not be removed.

Secondly, in the embodiment of the present application, provides and a kind of detect whether that there are the methods in logo region.It is to be detected detecting In video with the presence or absence of before logo image, object edge can also be detected each object edge in set of graphs detect figure into Column hisgram statistics, obtains M statistical result, then judges whether each statistical result in M statistical result meets respectively Logo extracted region condition, if at least one statistical result in M statistical result meets logo extracted region condition, it is determined that Object edge detects set of graphs, and there are logo regions, if meeting logo extracted region item without statistical result in M statistical result Part, it is determined that object edge detects set of graphs and logo region is not present.By the above-mentioned means, the case where giving statistical threshold Under, statistics with histogram is carried out to object edge detection figure both horizontally and vertically respectively, histogram is rejected and is less than statistics Compact logo region can be obtained in the part of threshold value, thus the feasibility of lifting scheme.

Optionally, on the basis of above-mentioned Fig. 3 corresponding embodiment, map logo method for distinguishing provided by the embodiments of the present application In 4th alternative embodiment, the logo image in video to be detected is determined according to P frame image to be detected and logo region, it can To include:

The image score value of image to be matched in logo region is determined according to P frame image to be detected；

If the image score value of image to be matched is greater than or equal to logo image threshold, it is determined that there are platforms in video to be detected Logo image.

In the present embodiment, a kind of method for detecting and whether there is logo image in video to be detected is described, based on The logo region detected, the averaged in P frame image to be detected.Specifically, it is assumed that detect that there is currently a platforms Region is marked, which is the a-quadrant of image to be detected, then, by every frame image to be detected institute in P frame image to be detected Corresponding a-quadrant is overlapped.Assuming that have 64 frame image to be detected, logo area detection result corresponding to every frame image to be detected It is 1 or 0, if there is 50 frame logo area detection results are 1, being left 14 frame logo area detection results is 0, then being calculated The image score value of the image to be matched is 0.83.According to the image score value of image to be matched, judge whether the image score value is greater than Or it is equal to logo image threshold, it is assumed that logo image threshold is 0.5, then the image score value of image to be matched is greater than logo image threshold Value, thereby determines that there are logo images in video to be detected.One logo image detected as shown in figure 9, referring to Fig. 9, Fig. 9 is a schematic diagram of logo image in the embodiment of the present application, wherein logo image may include pattern or text.

, whereas if the image score value of image to be matched is less than logo image threshold, thereby determine that in video to be detected not There are logo images.For a video to be detected, it is assumed that M image-region is all satisfied logo extracted region condition, then table Showing video to be detected, there are M logo regions, and therefore, the logo amount of images detected is up to M (i.e. each logo regions All there is logo image), the logo amount of images detected (the absolutely not logo figure in video i.e. to be detected that is at least 0 Picture).

Secondly, a kind of method for detecting and whether there is logo image in video to be detected is provided in the embodiment of the present application, The image score value for determining image to be matched in logo region according to P frame image to be detected first, if the image of image to be matched Score value is greater than or equal to logo image threshold, it is determined that there are logo images in video to be detected.By the above-mentioned means, for one For a video to be detected, the logo amount of images detected is at least absolutely not logo in 0, that is, video to be detected Image, the logo amount of images detected, which is up to each logo region, a logo image, image score value is greater than or Image to be matched equal to logo image threshold is determined as logo image, can be effectively reduced detection error rate, and obtain Relatively stable logo image.

Optionally, on the basis of above-mentioned Fig. 3 corresponding embodiment, map logo method for distinguishing provided by the embodiments of the present application In 5th alternative embodiment, logo image is matched with preset icon set, obtains the icon-based programming of video to be detected As a result, may include:

Obtain the local feature set to be matched of logo image, wherein local feature set to be matched includes at least one Local feature to be matched；

Obtain the local feature set of each preset icon in preset icon set, wherein local feature set includes extremely A few local feature；

According to local feature set to be matched and the local feature set of each preset icon, by k nearest neighbor algorithm from Candidate's logo image collection is determined in preset icon set, wherein candidate logo image collection includes N number of candidate logo image, N For the integer more than or equal to 1；

Each candidate logo image in logo image and candidate logo image collection is compared, each candidate is obtained The pairing point set of logo image, wherein pairing point set includes at least one match point, and match point indicates candidate logo image With the characteristic point of logo image successful matching；

According to the pairing point set of each candidate logo image and the local feature set of each candidate logo image, meter Calculation obtains N number of similarity score；

According to the maximum value of similarity score in N number of similarity score, determination is to be detected from candidate logo image collection The Target Station logo image of video.

In the present embodiment, a kind of mode for obtaining TV station symbol recognition result will be introduced.Firstly the need of extract logo image to Local feature set is matched, local feature set to be matched includes at least one local feature to be matched, it is to be understood that should Local feature to be matched specifically can be SIFT feature, be also possible to accelerate robust feature (Speeded Up Robust Features, SURF) or slewing and quickly rotation (oriented fast and rotated brief, ORB) Feature etc., this is illustrated for sentencing extraction SIFT feature, however this is not construed as the restriction to the application.

Similarly, it is also necessary to obtain the local feature set of each preset icon in preset icon set, wherein part is special It includes at least one local feature that collection, which is closed, it is to be understood that the local feature specifically can be SIFT feature, be also possible to SURF ORB feature etc., this is illustrated for sentencing extraction SIFT feature, however this is not construed as to the application Restriction.

Based on above-mentioned introduction, it is assumed that have S preset icon in preset icon set, extracted from each preset icon set The number of the SIFT feature arrived is n_i(i=1,2 ..., S), then preset icon set has in totalA SIFT Characteristic point needs to save two category informations to each SIFT feature, first is that the position coordinates (x of characteristic point_i,y_i), first is that feature The characteristic information F of point_i, wherein characteristic information F_iThe usually floating number of 128 dimensions, corresponds to the S in preset icon set_AA part Feature and local feature set to be matched can establish the quick indexing of local feature using k neighbour fast indexing method, In, the method for establishing k nearest neighbor quick indexing includes but is not limited only to fast nearest-neighbor search packet (Fast Library for Approximate Nearest Neighbors, FLANN) and similarity searching library (Fast Similarity Search, FAISS)。

Candidate's logo image collection is determined from preset icon set by k nearest neighbor algorithm, wherein candidate logo image Set includes N number of candidate logo image, and N is the integer more than or equal to 1.Next it needs to logo image and candidate logo Each candidate's logo image is compared in image collection, and specially 1 to 1 similarity compares, to obtain each candidate platform The pairing point set of logo image, wherein pairing point set includes at least one match point, match point indicate candidate logo image with The characteristic point of logo image successful matching.And then according to the pairing point set of each candidate logo image and each candidate logo The local feature set of image, is calculated N number of similarity score.Finally, similarity score is most from N number of similarity score Big to be worth, candidate logo image corresponding to the maximum value is the Target Station logo image of video to be detected.

Secondly, providing a kind of mode for obtaining TV station symbol recognition result, first acquisition logo image in the embodiment of the present application Local feature set to be matched, then obtain preset icon set in each preset icon local feature set, it is close by k Adjacent algorithm determines candidate's logo image collection from preset icon set, then can be to logo image and candidate logo image Each candidate's logo image is compared in set, obtains the pairing point set of each candidate logo image, and then according to each The local feature set of the pairing point set of candidate logo image and each candidate logo image, is calculated N number of similarity Score value finally determines the Target Station logo image of video to be detected according to N number of similarity score from candidate logo image collection.It is logical Aforesaid way is crossed, first using thick matching, then essence is used to match available TV station symbol recognition as a result, thus reducing matched calculating Amount, while accuracy in detection can also be effectively promoted, thus the reliability of lifting scheme.

Optionally, on the basis of above-mentioned Fig. 3 corresponding 5th embodiment, icon-based programming provided by the embodiments of the present application The 6th alternative embodiment of method in, according to the local feature collection of local feature set to be matched and each preset icon It closes, determines candidate's logo image collection from preset icon set by k nearest neighbor algorithm, may include:

1) local feature to be matched in local feature set to be matched is obtained；

2) it according to local feature to be matched, is obtained from the local feature set of each preset icon special with part to be matched Levy K nearest candidate feature, wherein K is the integer more than or equal to 1；

Step 1) is repeated to step 2), until getting each of local feature set to be matched part to be matched The candidate feature of feature；

According to the candidate feature of each of local feature set to be matched local feature to be matched, candidate logo figure is obtained Image set closes.

In the present embodiment, it will introduce and how to determine candidate's logo image set from preset icon set by k nearest neighbor algorithm It closes, for ease of description, will be illustrated by taking a local feature to be matched as an example, it is to be understood that part to be matched is special The processing mode of other local features to be matched is similar in collection conjunction, therefore is not repeated herein.

Specifically, a local feature to be matched in local feature set to be matched, the local feature to be matched are obtained It can be SIFT feature, it, can be in default figure for each of local feature set to be matched local feature to be matched In target local feature set (i.e. local feature quick indexing library), k neighbour retrieval is carried out in the range of given radius R, so Hit k nearest neighbor number is carried out according to logo in the local feature set of preset icon afterwards to arrange from more to less, takes hit-count Most N number of candidate logos are as slightly matching the candidate logo image collection chosen.

Wherein, k nearest neighbor number can take arbitrary integer, and R is an empirical value, will illustrate how to adopt with an example below Candidate logo image collection is obtained with k nearest neighbor algorithm.Assuming that K=5, and assume there are 50 preset icons in preset icon set, That is A1, A2 ..., A50, each preset icon extracts 10 local features, it follows that preset icon set shares 50 × 10= 500 local features.Assuming that the logo image extracted also has 10 local features to be matched, at this point, to logo image For first local feature to be matched, need to search 5 nearest candidate features in 500 local features, if this 5 Candidate feature is A1, A2, A3, A4 and A5, then A1, A2, A3, A4 and A5 have been hit 1 time by k neighbour respectively.Next, being directed to platform Second local feature to be matched of logo image, it is still necessary to 5 nearest candidate features are searched in 500 local features, it is false If current 5 candidate features are belonging respectively to A1, A36, A8, A10 and A25, then A1 has been hit 2 times, and so on, finally A candidate feature of the most N (such as N=3) of hit-count is chosen as candidate logo image collection.

Again, in the embodiment of the present application, a kind of method of determining candidate logo image collection is provided, that is, is obtained to be matched A local feature to be matched in local feature set, then according to local feature to be matched, from the office of each preset icon Acquisition and K nearest candidate feature of local feature to be matched, repeat above-mentioned steps in portion's characteristic set, until getting The candidate feature of each of local feature set to be matched local feature to be matched, finally according to local feature set to be matched Each of local feature to be matched candidate feature, obtain candidate logo image collection.By the above-mentioned means, utilizing k neighbour Algorithm obtains thick matched as a result, obtaining candidate logo image collection, can rapidly filter out time the most possible in this way Channel selection logo image avoids preset icon all in traversal preset icon set, to promote logo detection efficiency.

Optionally, on the basis of above-mentioned Fig. 3 corresponding 5th embodiment, icon-based programming provided by the embodiments of the present application The 7th alternative embodiment of method in, each candidate logo image in logo image and candidate logo image collection is carried out It compares, obtains the pairing point set of each candidate logo image, may include:

Each local feature of each of logo image local feature to be matched and each candidate logo image is carried out Matching, obtains projection matrix, wherein projection matrix indicates position coordinates of the logo image after projection；

According to logo image and projection matrix, the pairing point set of each candidate logo image is determined.

In the present embodiment, the pairing point set for how obtaining each candidate logo image will be introduced, for candidate logo figure The similarity that N number of candidate logo image in image set conjunction carries out 1 to 1 with currently detected logo image respectively compares, therefrom It selects similarity score highest and is greater than final identification knot of the candidate logo image as logo image of given score threshold Fruit.For the ease of introducing, referring to Fig. 10, Figure 10 is one embodiment schematic diagram that similarity compares in the embodiment of the present application, As shown, be illustrated by taking a candidate logo image in candidate logo image collection as an example, it first will be currently detected Logo image in the search in corresponding candidate logo image of each local feature (such as SIFT feature) to be matched obtain Europe Formula is thus used as match point apart from nearest local feature (such as SIFT feature).According to the position coordinates of all match points, so Corresponding projection matrix H is calculated using RANSAC algorithm (random sample consensus, ransac) afterwards. By the position coordinates of local feature to be matched in currently detected logo image, using projection matrix H by logo image projection Into the coordinate of candidate logo image.Next the Europe between the position coordinates of subpoint and the position coordinates of former corresponding points is calculated Formula distance, and reject the match point that Euclidean distance is greater than given threshold value, it is remaining be correct matched each match point to get To the pairing point set of candidate logo image.

Again, in the embodiment of the present application, a kind of method of pairing point set for determining each candidate logo image is provided, Each local feature of each of logo image local feature to be matched and each candidate logo image can be carried out first Matching, obtains projection matrix, then according to logo image and projection matrix, determines the pairing point set of each candidate logo image It closes.By the above-mentioned means, correctly pairing point set can be got in the less candidate logo image collection of relative data, Thus the similarity score between logo image and each candidate logo image is calculated, so that it is matched accurate to promote logo Degree.

Optionally, on the basis of above-mentioned Fig. 3 corresponding 7th embodiment, icon-based programming provided by the embodiments of the present application The 8th alternative embodiment of method in, according to the pairing point set of each candidate logo image and each candidate logo image Local feature set, N number of similarity score is calculated, may include:

Similarity score is calculated in the following way:

Wherein, score indicates that similarity score, A indicate area corresponding to the pairing point set of candidate logo image simultaneously Collection, B indicate area union corresponding to the local feature set of candidate logo image.

In the present embodiment, the concrete mode for calculating candidate logo image similarity score value will be introduced.Obtaining each candidate After the pairing point set of logo image, for ease of description, by the similarity point to calculate one of candidate logo image It being introduced for value, it is to be understood that the similarity score calculation of other candidate's logo images is similar, therefore herein It does not repeat them here.

Specifically, due to acquisition be candidate logo image local feature, each local feature can be enabled (i.e. characteristic point) represents radius as the territory of r, and r is empirical value, for example, can value be 9 pixels.It is using following formula Similarity score can be calculated:

A indicates area union corresponding to the pairing point set of candidate logo image, and B indicates the part of candidate logo image Area union corresponding to characteristic set.

There are area coincidence, need to subtract the part of area overlapping, that is to say, that all area overlappings Only calculate an area in part.For example, it is assumed that be all the characteristic point for having matched 10 local features, if this 10 parts are special The characteristic point of sign is all crowded together, that can only illustrate that a pocket matches, if the characteristic point of this 10 local features It is evenly distributed on candidate logo image, then the matched region is certainly bigger than what is be crowded together, therefore, the latter's is similar It is also just higher to spend score value.The logo that similarity score highest and the candidate logo image for being greater than given threshold value the most finally identify.

Further, in the embodiment of the present application, a kind of concrete mode for calculating similarity score is provided, above-mentioned side is passed through Formula provides a kind of reasonable and reliable implementation for the realization of scheme, thus the feasibility and operability of lifting scheme.

Optionally, above-mentioned Fig. 3 and Fig. 3 it is corresponding first to any one of the 8th embodiment on the basis of, this In the 9th alternative embodiment of map logo method for distinguishing that embodiment offer is provided, logo image and preset icon set are carried out It matches, before the icon-based programming result for obtaining video to be detected, can also include:

Obtain video collection to be processed, wherein include at least one video to be processed in video collection to be processed；

Each of video collection to be processed video to be processed is detected, logo image collection to be processed is obtained, In, logo image collection to be processed includes at least one logo image to be processed, and at least one in logo image collection to be processed A logo image to be processed corresponds to the same mark；

The logo image to be processed treated in treatment bench logo image set is handled, and preset icon set is obtained；

Feature extraction is carried out to each preset icon in preset icon set, obtains the local feature of each preset icon Set, wherein local feature set includes at least one local feature, and local feature includes characteristic point position coordinate and feature Information.

In the present embodiment, a kind of method for establishing preset icon set will be introduced, obtains video collection to be processed first, to Handling includes at least one video to be processed in video collection, and video to be processed can be the video manually extracted, be also possible to The randomly selected video from background data base.Next each video to be processed is detected, obtains logo figure to be processed Image set closes, and may include at least one logo image to be processed in a video to be processed, it is to be understood that in practical application In, at least one logo image to be processed corresponds to the same mark.In order to make it easy to understand, please referring to Figure 11, Figure 11 is the application The schematic diagram of the same logo difference display format in embodiment, as shown, the same logo can have different display shapes There is the performance of (a) in Figure 11 to realize for formula, such as the logo of one entitled " Sunny video ", it is possible to have in Figure 11 (b) The form of expression, therefore, no matter which kind of form of expression, mark be all mark corresponding to " Sunny video ".It collects as far as possible Multiple videos to be processed, making video to be processed includes different display format, can increase the covering of icon-based programming system in this way Rate.

After obtaining treatment bench logo image set, continue to each for the treatment of bench logo image set logo to be processed Image is handled, and processing mode includes but is not limited only to cutting out for logo image to be processed, the scaling of logo image to be processed, The cleaning etc. of logo image to be processed.Wherein, during building library offline, need manually to wash the non-of erroneous detection after logo detection Logo image, during online TV station symbol recognition, these erroneous detections can be excluded in subsequent characteristic matching process, therefore will not shadow Ring final recognition result.Logo cleaning refers to that rejecting is possible to erroneous detection to non-logo image, it would be desirable to ensure preset icon Preset icon in set is correct logo image, and static logo image can also be obtained by the way of manually cutting out Logo image is taken, then logo is obtained more preferably automatically to dynamic logo.

Preset icon set can be obtained in after treatment.Further, it is based on preset icon set, it can also be further The local feature for extracting each preset icon, obtains the local feature set of each preset icon, and local feature set includes extremely A few local feature, local feature includes characteristic point position coordinate and characteristic information, wherein characteristic point position coordinate is specific It is the abscissa and ordinate of pixel, characteristic information can be the floating number of 128 dimensions.

Further, in the embodiment of the present application, provide a kind of method for establishing preset icon set, i.e., firstly the need of Video collection to be processed is obtained, then each of video collection to be processed video to be processed is detected, is obtained to be processed Logo image collection, logo image collection to be processed include at least one logo image to be processed, logo image collection to be processed At least one of logo image to be processed correspond to the same mark, then treat to be processed in treatment bench logo image set Logo image is handled, and preset icon set is obtained, and is finally carried out feature to each preset icon in preset icon set and is mentioned It takes, obtains the local feature set of each preset icon.By the above-mentioned means, preset icon can be established in the state of offline Set, that is, establish logo index database, is convenient for subsequent matching operation as a result, and at least one logo image to be processed is corresponding The same mark, that is to say, that default figure of the same logo under different display formats can be also stored in preset icon set Mark, thus the reliability and flexibility of lifting scheme, and then promote successful match rate.

Optionally, on the basis of above-mentioned Fig. 3 corresponding 9th embodiment, icon-based programming provided by the embodiments of the present application The tenth alternative embodiment of method in, the logo image to be processed treated in treatment bench logo image set is handled, and is obtained Preset icon set may include:

When receiving the first process instruction, first is rejected from logo image collection to be processed according to the first process instruction Logo image to be processed, wherein the mark of the first logo image to be processed is carried in the first process instruction；

When receive second processing instruction when, according to second processing instruction treat in treatment bench logo image set second to Treatment bench logo image is adjusted, and obtains the preset icon in preset icon set, wherein carries second in second processing instruction The mark of logo image to be processed.

In the present embodiment, introduce that the logo image to be processed how treated in treatment bench logo image set is handled.It is main It to include two ways, a kind of mode is rejected to ineligible logo image to be processed, and another way is pair Qualified logo image to be processed is cut.

Specifically, it is introduced by taking the first logo image to be processed in logo image collection to be processed as an example, it is assumed that should First logo image to be processed does not meet logo condition, for example, the first logo image to be processed is not real logo image, or There are more noise or the first logo image to be processed to occur on the logo image to be processed of person first severely deformed etc..At this In a situation, user triggers the first process instruction, and the first process instruction carries the mark of the first logo image to be processed, map logo Other device rejects the first logo image to be processed according to the first process instruction from logo image collection to be processed.

It is introduced by taking the second logo image to be processed in logo image collection to be processed as an example, it is assumed that this is second wait locate Reason logo image meets logo condition, but the size of the second logo image to be processed is bigger than normal or less than normal.In that case, User triggers second processing instruction, and second processing instruction carries the mark of the second logo image to be processed, icon-based programming device root It is instructed according to second processing, the second logo image to be processed is adjusted, for example the second logo image to be processed is cut out Or amplification etc., to obtain preset icon.

Still further, providing a kind of method treating treatment bench logo image and being handled in the embodiment of the present application.When When receiving the first process instruction, the first logo to be processed is rejected from logo image collection to be processed according to the first process instruction Image.When receiving second processing instruction, according to second processing instruction treat in treatment bench logo image set second wait locate Reason logo image is adjusted, and obtains the preset icon in preset icon set.By the above-mentioned means, establishing preset icon collection During conjunction, treatment bench logo image can be treated according to the actual situation and is cut out, can also be rejected undesirable Logo image to be processed is conducive to subsequent matching thus, it is possible to obtain more regular preset icon, at the same time it can also add Add the corresponding logo image to be processed of the same mark, thus the diversity of lifting scheme.

It should be understood that can be metastable by dynamic logo present applicant proposes a kind of method merged by multiframe Static logo extracts local feature on obtained static logo and carries out knowledge method for distinguishing, can cope with dynamic logo, can also make The identification of static logo is more accurate.Based on aforesaid way, We conducted a series of experiments, and it is fast to collect 360 from network first Video, Baidu be good-looking and the associated video of 18 kinds of logos such as trill, is used as platform after 5 video extraction logos of every kind of video collect Index database is marked, in addition collecting correlation there are logo video totally 728, and no logo video 1000, as test set, test result is such as Shown in the following table 1.

Table 1

Logo	Total amount	True overlay capacity	Coverage rate	Hit total amount	Hit right amount	Accuracy rate
							No. 1 logo	41	40	98%	40	40	100%
No. 2 logos	44	44	100%	45	44	98%
							No. 3 logos	44	38	86%	38	38	100%
No. 4 logos	36	36	100%	36	36	100%
							No. 5 logos	31	31	100%	32	31	97%
No. 6 logos	54	51	94%	51	51	100%
							No. 7 logos	21	20	95%	20	20	100%
No. 8 logos	54	50	93%	50	50	100%
							No. 9 logos	32	28	88%	28	28	100%
No. 10 logos	14	13	93%	13	13	100%
							No. 11 logos	57	56	98%	56	56	100%
No. 12 logos	42	41	98%	41	41	100%
							No. 13 logos	47	43	91%	44	43	98%
No. 14 logos	52	52	100%	52	52	100%
							No. 15 logos	55	51	93%	52	51	98%
No. 16 logos	32	27	84%	27	27	100%
							No. 17 logos	18	16	89%	16	16	100%
No. 18 logos	54	52	96%	52	52	100%
							It amounts to	728	689	95%	693	689	99%

As it can be seen that the application has good coverage rate and accuracy rate to the dynamic logo of No. 5 logos, if only with the of video One frame is used as logo and extracts and identify, then the coverage rate of dynamic logo is then almost 0.?Core^TMI7-4790 centre Under the hardware test environment for managing device (Central Processing Unit, CPU)@3.6G hertz (Hertz, Hz), each video Average time-consuming about 1.5 seconds, if by the way of traversing logo library, each video average time-consuming about 3 seconds, and this Time-consuming can be linearly increasing with the picture number in logo library, and uses thick matching, then only does 1 to 1 phase to most like N number of logo Like the mode that degree compares, then in the increase of logo library picture number, time-consuming increase almost be can be ignored.

The icon-based programming device in the application is described in detail below, please refers to Figure 12, Figure 12 is the application implementation Icon-based programming device one embodiment schematic diagram in example, icon-based programming device 20 include:

Module 201 is obtained, for obtaining P frame image to be detected at random from video to be detected, wherein the view to be detected Frequency includes Q frame video image, and the Q is integer greater than 1, the P be more than or equal to 1, and it is whole less than or equal to the Q Number；

Detection module 202, the mapping to be checked in described P frame image to be detected for being obtained to the acquisition module 201 As carrying out edge detection, object edge detection set of graphs is obtained, wherein include M mesh in the object edge detection set of graphs Edge detection graph is marked, each object edge detection figure is obtained after merging to P edge detection graph, and the M is to be greater than or wait In 1 integer；

Determining module 203, the object edge detection set of graphs for being obtained according to the detection module 202 detection are true Determine icon area；

The determining module 203, be also used to be determined according to described P frame image to be detected and the icon area it is described to Detect the icon in video；

Identification module 204, the icon and the progress of preset icon set for determining the determining module 203 Match, obtain the icon-based programming result of the video to be detected, wherein the preset icon set includes at least one default figure Mark.

In the present embodiment, obtain module 201 obtain P frame image to be detected at random from video to be detected, wherein it is described to Detection video includes Q frame video image, and the Q is the integer greater than 1, and the P is and to be less than or equal to institute more than or equal to 1 State the integer of Q, image to be detected in described P frame image to be detected that detection module 202 obtains the acquisition module 201 into Row edge detection obtains object edge detection set of graphs, wherein includes M target side in the object edge detection set of graphs Edge detection figure, each object edge detection figure are obtained after merging to P edge detection graph, and the M is more than or equal to 1 Integer, the object edge detection set of graphs that determining module 203 is obtained according to the detection module 202 detection determine icon area Domain, the determining module 203 determine in the video to be detected according to described P frame image to be detected and the icon area Icon, identification module 204 match the icon that the determining module 203 determines with preset icon set, obtain institute State the icon-based programming result of video to be detected, wherein the preset icon set includes at least one preset icon.

In the embodiment of the present application, a kind of icon-based programming device is provided, the icon-based programming device is needed to be detected first P frame image to be detected is obtained in video at random, edge inspection then is carried out to every frame image to be detected in P frame image to be detected It surveys, obtains object edge detection set of graphs, it includes M object edge detection figure, each target that object edge, which detects in set of graphs, Edge detection graph is obtained after merging to P edge detection graph, if detecting set of graphs according to object edge determines that there are logos Region, then according in P frame image to be detected and logo region detection video to be detected whether there is logo image.If to be detected There are logo images in video, then by logo image matched with preset icon set, obtain the icon of video to be detected Recognition result, wherein preset icon set includes at least one preset icon.By the above-mentioned means, stochastical sampling multi-frame video Image is merged, and on the one hand can increase the diversity of logo background variation using stochastical sampling, reaches preferably sampling effect Fruit, on the other hand, multi-frame video image, which is carried out fusion, can become dynamic logo metastable static logo, and then right Static logo is identified, is thus suitable for the detection to static logo and dynamic logo, to promote recognition accuracy.

Optionally, on the basis of the embodiment corresponding to above-mentioned Figure 12, Figure 13 is please referred to, it is provided by the embodiments of the present application In another embodiment of icon-based programming device 20, the icon-based programming device 20 further includes division module 205 and extraction module 206；

The division module 205, in the detection module 202 to the mapping to be checked in described P frame image to be detected It is to be checked to every frame in described P frame image to be detected before obtaining object edge detection set of graphs as carrying out edge detection Altimetric image is divided, and multiple images region corresponding to described every frame image to be detected is obtained；

The extraction module 206, it is right for dividing obtained every frame image to be detected institute from the division module 205 M image-region corresponding to described every frame image to be detected is extracted in the multiple images region answered, wherein the M image Region is for carrying out edge detection.

Optionally, on the basis of the embodiment corresponding to above-mentioned Figure 13, icon-based programming dress provided by the embodiments of the present application It sets in 20 another embodiment,

The detection module 202, specifically for the target in the M image-region to described every frame image to be detected Image-region carries out edge detection, obtains P edge detection graph corresponding to the object region, wherein the target Image-region belongs to any one image-region in the M image-region；

Optionally, on the basis of the embodiment corresponding to above-mentioned Figure 12, icon-based programming dress provided by the embodiments of the present application It sets in 20 another embodiment,

The determining module 203, specifically for being detected to each object edge in object edge detection set of graphs Figure carries out statistics with histogram, obtains M statistical result, wherein the statistics with histogram is used to detect figure to object edge and carry out The statistics of horizontal direction and vertical direction；

If at least one statistical result in the M statistical result meets the statistical threshold, it is determined that the target Edge detection set of graphs deposits the logo region；

The determining module 203, it is to be matched in the logo region specifically for being determined according to described P frame image to be detected The image score value of image；

The identification module 204, specifically for obtaining the local feature set to be matched of the logo image, wherein institute Stating local feature set to be matched includes at least one local feature to be matched；

The identification module 204 is specifically used for 1) obtaining an office to be matched in the local feature set to be matched Portion's feature；

The identification module 204, specifically for by each of logo image local feature to be matched and described every Each local feature of a candidate's logo image is matched, and obtains projection matrix, wherein the projection matrix indicates described Position coordinates of the logo image after projection；

The identification module 204, specifically for calculating similarity score in the following way:

Optionally, on the basis of the embodiment corresponding to above-mentioned Figure 12, Figure 14 is please referred to, it is provided by the embodiments of the present application In another embodiment of icon-based programming device 20, the icon-based programming device 20 further includes processing module 207 and extraction module 206；

The acquisition module 201 is also used to carry out the icon and preset icon set in the identification module 204 Match, before the icon-based programming result for obtaining the video to be detected, obtains video collection to be processed, wherein the view to be processed It include at least one video to be processed in frequency set；

The detection module 202 is also used to every in the video collection to be processed to acquisition module 201 acquisition A video to be processed is detected, and logo image collection to be processed is obtained, wherein the logo image collection to be processed includes extremely A few logo image to be processed, at least one of described logo image collection to be processed logo image to be processed corresponds to same A mark；

The processing module 207, for detecting the obtained logo image collection to be processed to the detection module 202 In logo image to be processed handled, obtain the preset icon set；

The extraction module 206, it is every in the obtained preset icon set for handling the processing module 209 A preset icon carries out feature extraction, obtains the local feature set of each preset icon, wherein the local feature collection Closing includes at least one local feature, and the local feature includes characteristic point position coordinate and characteristic information.

Optionally, on the basis of the embodiment corresponding to above-mentioned Figure 14, icon-based programming dress provided by the embodiments of the present application It sets in 20 another embodiment,

The processing module 207, specifically for when receiving the first process instruction, according to first process instruction from The first logo image to be processed is rejected in the logo image collection to be processed, wherein carry institute in first process instruction State the mark of the first logo image to be processed；

Figure 15 is a kind of server architecture schematic diagram provided by the embodiments of the present application, which can be because of configuration or property Energy is different and generates bigger difference, may include one or more central processing units (central processing Units, CPU) 322 (for example, one or more processors) and memory 332, one or more storages apply journey The storage medium 330 (such as one or more mass memory units) of sequence 342 or data 344.Wherein, 332 He of memory Storage medium 330 can be of short duration storage or persistent storage.The program for being stored in storage medium 330 may include one or one With upper module (diagram does not mark), each module may include to the series of instructions operation in server.Further, in Central processor 322 can be set to communicate with storage medium 330, execute on server 300 a series of in storage medium 330 Instruction operation.

Server 300 can also include one or more power supplys 326, one or more wired or wireless networks Interface 350, one or more input/output interfaces 358, and/or, one or more operating systems 341, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM etc..

The step as performed by server can be based on the server architecture shown in figure 15 in above-described embodiment.

In the embodiment of the present application, CPU 322 included by the server is also with the following functions:

The embodiment of the present application also provides another icon-based programming devices, as shown in figure 16, for ease of description, only show Part relevant to the embodiment of the present application, it is disclosed by specific technical details, please refer to the embodiment of the present application method part.It should Terminal device can be include mobile phone, tablet computer, personal digital assistant (personal digital assistant, PDA), Any terminal device equipment such as point-of-sale terminal equipment (point of sales, POS), vehicle-mounted computer, using terminal device as mobile phone For:

Figure 16 shows the block diagram of the part-structure of mobile phone relevant to terminal device provided by the embodiments of the present application.Ginseng Figure 16 is examined, mobile phone includes: radio frequency (radio frequency, RF) circuit 410, memory 420, input unit 430, display list First 440, sensor 450, voicefrequency circuit 460, Wireless Fidelity (wireless fidelity, WiFi) module 470, processor The components such as 480 and power supply 490.It will be understood by those skilled in the art that handset structure shown in Figure 16 does not constitute opponent The restriction of machine may include perhaps combining certain components or different component layouts than illustrating more or fewer components.

It is specifically introduced below with reference to each component parts of the Figure 16 to mobile phone:

RF circuit 410 can be used for receiving and sending messages or communication process in, signal sends and receivees, particularly, by base station After downlink information receives, handled to processor 480；In addition, the data for designing uplink are sent to base station.In general, RF circuit 410 Including but not limited to antenna, at least one amplifier, transceiver, coupler, low-noise amplifier (low noise Amplifier, LNA), duplexer etc..In addition, RF circuit 410 can also be communicated with network and other equipment by wireless communication. Any communication standard or agreement, including but not limited to global system for mobile communications (global can be used in above-mentioned wireless communication System of mobile communication, GSM), general packet radio service (general packet radio Service, GPRS), CDMA (code division multiple access, CDMA), wideband code division multiple access (wideband code division multiple access, WCDMA), long term evolution (long term evolution, LTE), Email, short message service (short messaging service, SMS) etc..

Memory 420 can be used for storing software program and module, and processor 480 is stored in memory 420 by operation Software program and module, thereby executing the various function application and data processing of mobile phone.Memory 420 can mainly include Storing program area and storage data area, wherein storing program area can application journey needed for storage program area, at least one function Sequence (such as sound-playing function, image player function etc.) etc.；Storage data area can be stored to be created according to using for mobile phone Data (such as audio data, phone directory etc.) etc..It, can be in addition, memory 420 may include high-speed random access memory Including nonvolatile memory, for example, at least a disk memory, flush memory device or other volatile solid-states Part.

Input unit 430 can be used for receiving the number or character information of input, and generate with the user setting of mobile phone with And the related key signals input of function control.Specifically, input unit 430 may include that touch panel 431 and other inputs are set Standby 432.Touch panel 431, also referred to as touch screen, collect user on it or nearby touch operation (such as user use The operation of any suitable object or attachment such as finger, stylus on touch panel 431 or near touch panel 431), and root Corresponding attachment device is driven according to preset formula.Optionally, touch panel 431 may include touch detecting apparatus and touch Two parts of controller.Wherein, the touch orientation of touch detecting apparatus detection user, and touch operation bring signal is detected, Transmit a signal to touch controller；Touch controller receives touch information from touch detecting apparatus, and is converted into touching Point coordinate, then gives processor 480, and can receive order that processor 480 is sent and be executed.Furthermore, it is possible to using electricity The multiple types such as resistive, condenser type, infrared ray and surface acoustic wave realize touch panel 431.In addition to touch panel 431, input Unit 430 can also include other input equipments 432.Specifically, other input equipments 432 can include but is not limited to secondary or physical bond One of disk, function key (such as volume control button, switch key etc.), trace ball, mouse, operating stick etc. are a variety of.

Display unit 440 can be used for showing information input by user or be supplied to user information and mobile phone it is various Menu.Display unit 440 may include display panel 441, optionally, can use liquid crystal display (liquid crystal Display, LCD), the forms such as Organic Light Emitting Diode (organic light-emitting diode, OLED) it is aobvious to configure Show panel 441.Further, touch panel 431 can cover display panel 441, when touch panel 431 detect it is on it or attached After close touch operation, processor 480 is sent to determine the type of touch event, is followed by subsequent processing device 480 according to touch event Type corresponding visual output is provided on display panel 441.Although in Figure 16, touch panel 431 and display panel 441 It is that the input and input function of mobile phone are realized as two independent components, but in some embodiments it is possible to by touch-control Panel 431 and display panel 441 are integrated and that realizes mobile phone output and input function.

Mobile phone may also include at least one sensor 450, such as optical sensor, motion sensor and other sensors. Specifically, optical sensor may include ambient light sensor and proximity sensor, wherein ambient light sensor can be according to ambient light Light and shade adjust the brightness of display panel 441, proximity sensor can close display panel 441 when mobile phone is moved in one's ear And/or backlight.As a kind of motion sensor, accelerometer sensor can detect (generally three axis) acceleration in all directions Size, can detect that size and the direction of gravity when static, can be used to identify the application of mobile phone posture, (for example horizontal/vertical screen is cut Change, dependent game, magnetometer pose calibrating), Vibration identification correlation function (such as pedometer, tap) etc.；May be used also as mobile phone The other sensors such as gyroscope, barometer, hygrometer, thermometer, the infrared sensor of configuration, details are not described herein.

Voicefrequency circuit 460, loudspeaker 461, microphone 462 can provide the audio interface between user and mobile phone.Audio-frequency electric Electric signal after the audio data received conversion can be transferred to loudspeaker 461, be converted to sound by loudspeaker 461 by road 460 Signal output；On the other hand, the voice signal of collection is converted to electric signal by microphone 462, is turned after being received by voicefrequency circuit 460 It is changed to audio data, then by after the processing of audio data output processor 480, such as another mobile phone is sent to through RF circuit 410, Or audio data is exported to memory 420 to be further processed.

WiFi belongs to short range wireless transmission technology, and mobile phone can help user's transceiver electronics postal by WiFi module 470 Part, browsing webpage and access streaming video etc., it provides wireless broadband internet access for user.Although Figure 16 is shown WiFi module 470, but it is understood that, and it is not belonging to must be configured into for mobile phone, it can according to need do not changing completely Become in the range of the essence of invention and omits.

Processor 480 is the control centre of mobile phone, using the various pieces of various interfaces and connection whole mobile phone, is led to It crosses operation or executes the software program and/or module being stored in memory 420, and call and be stored in memory 420 Data execute the various functions and processing data of mobile phone, to carry out integral monitoring to mobile phone.Optionally, processor 480 can wrap Include one or more processing units；Optionally, processor 480 can integrate application processor and modem processor, wherein answer With the main processing operation system of processor, user interface and application program etc., modem processor mainly handles wireless communication. It is understood that above-mentioned modem processor can not also be integrated into processor 480.

Mobile phone further includes the power supply 490 (such as battery) powered to all parts, and optionally, power supply can pass through power supply pipe Reason system and processor 480 are logically contiguous, to realize management charging, electric discharge and power managed by power-supply management system Etc. functions.

Although being not shown, mobile phone can also include camera, bluetooth module etc., and details are not described herein.

In the embodiment of the present application, processor 480 included by the terminal device is also with the following functions:

It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description, The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.

In several embodiments provided herein, it should be understood that disclosed system, device and method can be with It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit It divides, only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components It can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown or The mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, the indirect coupling of device or unit It closes or communicates to connect, can be electrical property, mechanical or other forms.

The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme 's.

It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list Member both can take the form of hardware realization, can also realize in the form of software functional units.

If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product When, it can store in a computer readable storage medium.Based on this understanding, the technical solution of the application is substantially The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words It embodies, which is stored in a storage medium, including some instructions are used so that a computer Equipment (can be personal computer, server or the network equipment etc.) executes the complete of each embodiment the method for the application Portion or part steps.And storage medium above-mentioned include: USB flash disk, mobile hard disk, read-only memory (read-only memory, ROM), random access memory (random access memory, RAM), magnetic or disk etc. are various can store program The medium of code.

The above, above embodiments are only to illustrate the technical solution of the application, rather than its limitations；Although referring to before Embodiment is stated the application is described in detail, those skilled in the art should understand that: it still can be to preceding Technical solution documented by each embodiment is stated to modify or equivalent replacement of some of the technical features；And these It modifies or replaces, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution.

Claims

1. a kind of map logo method for distinguishing characterized by comprising

Obtain P frame image to be detected at random from video to be detected, wherein the video to be detected includes Q frame video image, institute Stating Q is the integer greater than 1, and the P is the integer more than or equal to 1, and less than or equal to the Q；

Edge detection is carried out to image to be detected in described P frame image to be detected, obtains object edge detection set of graphs, In, it include M object edge detection figure in the object edge detection set of graphs, each object edge detection figure is to P side It is obtained after the fusion of edge detection figure, the M is the integer more than or equal to 1；

The icon is matched with preset icon set, obtains the icon-based programming result of the video to be detected, wherein institute Stating preset icon set includes at least one preset icon.

2. the method according to claim 1, wherein the mapping to be checked in described P frame image to be detected As carrying out edge detection, before obtaining object edge detection set of graphs, the method also includes:

Described every frame image to be detected in described P frame image to be detected is divided, described every frame image to be detected is obtained Corresponding multiple images region；

M corresponding to described every frame image to be detected is extracted from multiple images region corresponding to described every frame image to be detected A image-region, wherein the M image-region is for carrying out edge detection.

3. according to the method described in claim 2, it is characterized in that, the mapping to be checked in described P frame image to be detected As carrying out edge detection, object edge detection set of graphs is obtained, comprising:

Edge detection is carried out to the object region in the M image-region of described every frame image to be detected, obtains institute State P edge detection graph corresponding to object region, wherein the object region belongs to the M image-region In any one image-region；

The P edge detection graph according to corresponding to the object region, determines mesh corresponding to the object region Mark edge detection graph；

When getting object edge detection figure corresponding to the M image-region, the object edge detection atlas is obtained It closes.

4. the method according to claim 1, wherein described detect the determining figure of set of graphs according to the object edge Mark region, comprising:

Statistics with histogram is carried out to each object edge detection figure in object edge detection set of graphs, obtains M statistics As a result, wherein the statistics with histogram is used to detect the statistics that figure carries out horizontal direction and vertical direction to object edge；

If at least one statistical result in the M statistical result meets the statistical threshold, it is determined that the object edge Detecting set of graphs, there are logo regions；

If meeting the statistical threshold without statistical result in the M statistical result, it is determined that the object edge detection figure Logo region is not present in set.

5. the method according to claim 1, wherein described according to described P frame image to be detected and the figure Mark region determines the icon in the video to be detected, comprising:

The image score value of image to be matched in logo region is determined according to described P frame image to be detected；

If the image score value of the image to be matched is greater than or equal to logo image threshold, it is determined that deposited in the video to be detected In logo image.

6. the method according to claim 1, wherein described by the icon and the progress of preset icon set Match, obtain the icon-based programming result of the video to be detected, comprising:

Obtain the local feature set to be matched of logo image, wherein the local feature set to be matched includes at least one Local feature to be matched；

Obtain the local feature set of each preset icon in the preset icon set, wherein the local feature set packet Include at least one local feature；

According to the local feature set to be matched and the local feature set of each preset icon, calculated by k neighbour Method determines candidate's logo image collection from the preset icon set, wherein candidate's logo image collection includes N number of time Channel selection logo image, the N are the integer more than or equal to 1；

Each candidate logo image in the logo image and the candidate logo image collection is compared, is obtained described The pairing point set of each candidate's logo image, wherein the pairing point set includes at least one match point, the match point Indicate the characteristic point of candidate logo image and the logo image successful matching；

According to the pairing point set of each candidate logo image and the local feature collection of each candidate logo image It closes, N number of similarity score is calculated；

According to the maximum value of similarity score in N number of similarity score, determined from the candidate logo image collection to Detect the Target Station logo image of video.

7. according to the method described in claim 6, it is characterized in that, described according to the local feature set to be matched and institute The local feature set for stating each preset icon determines candidate's logo figure by k nearest neighbor algorithm from the preset icon set Image set closes, comprising:

1) local feature to be matched in the local feature set to be matched is obtained；

2) according to the local feature to be matched, obtained from the local feature set of each preset icon with it is described to With K nearest candidate feature of local feature, wherein the K is the integer more than or equal to 1；

Step 1) is repeated to step 2), until getting each of the local feature set to be matched part to be matched The candidate feature of feature；

According to the candidate feature of each of the local feature set to be matched local feature to be matched, the candidate platform is obtained Logo image set.

8. according to the method described in claim 6, it is characterized in that, described to the logo image and the candidate logo figure Each candidate logo image is compared in image set conjunction, obtains the pairing point set of each candidate logo image, comprising:

By each local feature of each of logo image local feature to be matched and each candidate logo image It is matched, obtains projection matrix, wherein the projection matrix indicates position coordinates of the logo image after projection；

According to the logo image and the projection matrix, the pairing point set of each candidate logo image is determined.

9. according to the method described in claim 8, it is characterized in that, the match point according to each candidate logo image The local feature set of set and each candidate logo image, is calculated N number of similarity score, comprising:

Similarity score is calculated in the following way:

Wherein, the score indicates that the similarity score, the A indicate corresponding to the pairing point set of candidate logo image Area union, the B indicates area union corresponding to the local feature set of the candidate logo image.

10. method according to any one of claim 1 to 9, which is characterized in that described by the icon and preset icon Set is matched, before the icon-based programming result for obtaining the video to be detected, the method also includes:

Obtain video collection to be processed, wherein include at least one video to be processed in the video collection to be processed；

Each of the video collection to be processed video to be processed is detected, logo image collection to be processed is obtained, In, the logo image collection to be processed includes at least one logo image to be processed, in the logo image collection to be processed At least one logo image to be processed correspond to the same mark；

Logo image to be processed in the logo image collection to be processed is handled, the preset icon set is obtained；

Feature extraction is carried out to each preset icon in the preset icon set, obtains the part of each preset icon Characteristic set, wherein the local feature set includes at least one local feature, and the local feature includes characteristic point position Coordinate and characteristic information.

11. according to the method described in claim 10, it is characterized in that, it is described in the logo image collection to be processed to Treatment bench logo image is handled, and the preset icon set is obtained, comprising:

When receiving the first process instruction, rejected from the logo image collection to be processed according to first process instruction First logo image to be processed, wherein the mark of the described first logo image to be processed is carried in first process instruction；

When receiving second processing instruction, according to second processing instruction to the in the logo image collection to be processed Two logo images to be processed are adjusted, and obtain the preset icon in the preset icon set, wherein the second processing refers to The mark of the described second logo image to be processed is carried in order.

12. a kind of icon-based programming device characterized by comprising

Module is obtained, for obtaining P frame image to be detected at random from video to be detected, wherein the video to be detected includes Q Frame video image, the Q are the integer greater than 1, and the P is the integer more than or equal to 1, and less than or equal to the Q；

Detection module, image to be detected in described P frame image to be detected for obtaining to the acquisition module carry out edge Detection obtains object edge detection set of graphs, wherein includes M object edge detection in the object edge detection set of graphs Figure, each object edge detection figure are obtained after merging to P edge detection graph, and the M is the integer more than or equal to 1；

Determining module, the object edge detection set of graphs for being detected according to the detection module determine icon area Domain；

The determining module is also used to determine the view to be detected according to described P frame image to be detected and the icon area Icon in frequency；

Identification module, the icon for determining the determining module is matched with preset icon set, described in acquisition The icon-based programming result of video to be detected, wherein the preset icon set includes at least one preset icon.

13. a kind of server characterized by comprising memory, transceiver, processor and bus system；

Wherein, the memory is for storing program；

The icon is matched with preset icon set, obtains the icon-based programming result of the video to be detected, wherein institute Stating preset icon set includes at least one preset icon；

The bus system is for connecting the memory and the processor, so that the memory and the processor It is communicated.

14. a kind of terminal device characterized by comprising memory, transceiver, processor and bus system；

Wherein, the memory is for storing program；

15. a kind of computer readable storage medium, including instruction, when run on a computer, so that computer executes such as Method described in any one of claims 1 to 11.