WO2017088479A1 - Method of identifying digital on-screen graphic and device - Google Patents

Abstract

The invention relates to the technical field of information identification, and specifically, to a method of identifying a digital on-screen graphic and device. The method comprises: determining whether a digital on-screen graphic to be identified in a digital on-screen graphic field is a central television digital on-screen graphic (S101); if so, identifying, according to a first predefined strategy, the digital on-screen graphic to be identified (S102); and if not, identifying, according to a second predefined strategy, the digital on-screen graphic to be identified (S103). The method of identifying a digital on-screen graphic and the device do not identify different digital on-screen graphics using the same way, instead, it is first determined that whether the digital on-screen graphic to be identified is the central television digital on-screen graphic, and then different strategies are adopted to identify a central television digital on-screen graphic and a non-central television digital on-screen graphic, effectively identifying the channel to be identified with respect to the central television digital on-screen graphic and the non-central television digital on-screen graphic, and increasing identification efficiency and identification accuracy.

Description

Station identification method and device

cross reference

The present application claims the priority of the Chinese Patent Application, the entire disclosure of which is hereby incorporated by reference.

Technical field

The embodiments of the present invention relate to the field of information recognition technologies, and in particular, to a station logo identification method and apparatus.

Background technique

Smart TV is adapting to the trend of “high definition”, “networking” and “intelligence”. It has the ability to obtain program content from various channels such as the Internet, video equipment, computers, etc., and to meet the needs of consumers through information fusion. The ability to clearly show content on a large screen. Compared with traditional TV, smart TV provides users with faster, smarter and more humanized application services.

Smart TVs contain a large number of on-demand videos and live programs, and most TV programs retain the logo of the producers. The Taiwan Standard is an important symbol for distinguishing TV stations. It contains important semantic information such as TV station name, program source and program orientation. It is an important source of semantics for video analysis, understanding and retrieval. The realization of the logo recognition technology will effectively realize the function of the program list, and is of great significance for understanding user preferences and delivering value-added services.

In the prior art, when identifying the logo, the following two schemes are usually adopted:

The first scheme: a logo recognition scheme based on a single frame image. This method takes the edge distance transformation matrix as the feature, and adopts template matching method, including global edge matching (sliding window traverses the entire logo area), block matching (manual filtering non-edge), color shape matching (color and The shape features are matched at the same time, and the program search is sorted and sorted.

The second scheme: a logo recognition method based on multi-frame images. The scheme generally adopts the following three methods: 1. Segmenting an image by using a feature value change of a pixel of a sensitive multi-frame video frame sequence (ie, a logo area), and performing edge detection, using a sliding window and a block Square The method performs matching identification. Secondly, the color and shape features are comprehensively used, and the logo is divided in the frame sequence according to the time-space invariance. The spatial distribution histogram is combined with the HSV color histogram to describe the feature effectively. Finally, the SUV is used to complete the logo recognition. Method 3: Calculate the change of the adjacent frame image, extract the station mark and its Hu invariant moment, and obtain the recognition result according to the candidate set and the preset criterion.

However, in the prior art, the CCTV logo, the satellite logo, and the local logo are identified in the same manner, resulting in low recognition efficiency and low recognition accuracy.

Summary of the invention

The embodiment of the invention provides a method and a device for identifying a station logo, which are used to solve the defects of low recognition efficiency and low recognition accuracy in the prior art.

An embodiment of the present invention provides a station logo identification method, where the method includes:

Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo;

When the to-be-identified station is marked as a CCTV station logo, the to-be-identified station logo is identified according to a first preset policy;

When the to-be-identified station is marked as a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.

An embodiment of the present invention provides a station identification device, and the device includes:

The station label determining unit is configured to determine whether the station identifier to be identified in the station label area is a CCTV station label;

a first identifying unit, configured to: when the to-be-identified station is marked as a CCTV logo, identify the to-be-identified station logo according to a first preset policy;

The second identifying unit is configured to identify the to-be-identified station logo according to the second preset policy when the to-be-identified station is marked as a non-CCTV station logo.

An embodiment of the present invention provides a server, including:

a processor, a memory, a communication interface, and a bus; wherein

The communication interface is used for information transmission between the user equipment and the server;

The processor is configured to invoke logic instructions in the memory to perform the following method;

Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo; when the to-be-identified station logo is a CCTV station logo, identifying the to-be-identified station logo according to the first preset policy; When the station logo is a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.

An embodiment of the present invention provides a computer program, including program code, where the program code is used Do the following:

Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo;

When the to-be-identified station is marked as a CCTV station logo, the to-be-identified station logo is identified according to a first preset policy;

When the to-be-identified station is marked as a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.

Embodiments of the present invention provide a storage medium for storing the above computer program.

The station logo identification method and device provided by the embodiments of the present invention no longer identify each station label in the same manner, but first determine whether the station logo to be identified in the station logo area is a CCTV station logo, and then the CCTV station logo. Different non-CCTV logos are identified by different strategies, which can effectively identify the logos of CCTV logos and non-CCTV logos, which improves recognition efficiency and recognition accuracy.

DRAWINGS

1 is a flow chart of a method for identifying a station logo according to an embodiment of the present invention;

2 is a schematic diagram of a video frame image before area division in an embodiment of the present invention;

3 is a schematic diagram of a video frame image after region division in an embodiment of the present invention;

4 is an overall effect diagram of combining the edges of each video frame image by taking CCTV5 as an example;

FIG. 5 is a flowchart of a method for identifying a station logo according to an embodiment of the present invention; FIG.

6 is a block diagram showing the structure of a station identification device according to an embodiment of the present invention;

FIG. 7 is a schematic structural diagram of a server according to an embodiment of the present invention.

detailed description

The technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. It is a partial embodiment of the invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

It should be understood that although the background art is primarily directed to logo recognition of television programs in smart televisions, embodiments of the present disclosure are not limited thereto, and are also applicable to other objects (such as flat The identification of the logo of the television program in the tablet computer, the mobile phone, the PC, and the like, that is, the scene of the logo recognition involving the television program is included in the application range of the present embodiment.

1 is a flowchart of a method for identifying a logo of an embodiment of the present invention; referring to FIG. 1, the method includes:

S101: determining whether the to-be-identified station logo in the station logo area is a CCTV station logo;

It should be noted that the logo area is an area including only the logo to be identified.

It can be understood that the logo area can be extracted in a plurality of manners. In order to prevent the influence of noise such as random noise and picture noise on the logo recognition, in the embodiment, the station logo including the station identifier to be identified is obtained by the following steps. region:

(1) acquiring a video frame image sequence from a preset area of the video including the logo to be identified;

According to the prior knowledge, the TV station logo is basically located in the upper left corner of the video frame image (of course, if it is in other positions, it can be adjusted as needed), so when the logo is detected, only the fixed upper left corner needs to be extracted. The area (ie, the preset area) can be used as the station mark detection area. The existing logo recognition method generally acquires the logo area according to the optimal area rule (GSR). The difference between the present embodiment and the existing station logo recognition method is: (1) calculating all the station logos in each video frame image. The proportion position that is effectively recognized; (2) Calculate the maximum range of all proportional positions as the area divided by the station area. Taking the video of 1920*1080 as an example, the partition area of the station is - line start position 80 (1/24), column start position 40 (1/27), line width 450 (15/64), column width 180 (1) /6), the effect of the area division is as shown in FIG. 2 and FIG. 3, of course, the proportional position can be appropriately adjusted as needed, and the embodiment does not limit this.

In order to eliminate irrelevant information in the image, restore or enhance useful related information, improve the detectability of features, and minimize data to ensure the reliability of recognition, in this embodiment, each video frame image can be preprocessed. The pre-processing includes at least one of area segmentation, gradation, and image enhancement. Of course, other processes may be included, which is not limited in this embodiment.

The pre-processing can be grayed out by using the formula Gray=0.33R+0.59G+0.11B. Of course, it can also be replaced by a three-channel average method or a three-channel maximum method, wherein Gray is the gray level of the pixel. Value, R is the red component of the pixel, G is the green component of the pixel, and B is the blue component of the pixel.

The purpose of the image enhancement is to highlight the effective information of the logo area, such as icons, texts, numbers Etc., image enhancement uses grayscale stretching from 0 to 255 grayscale, and can also be replaced by histogram transformation.

(2) performing edge extraction on each video frame image;

Understandably, the edge is the sharp change of image gray scale. Edge extraction is the key to the identification of the logo. The integrity of the edge directly affects the logo recognition result. Of course, there are many methods for edge extraction, such as Canny, LOG, Sobel. , Laplacian operator, etc. Considering the requirements of denoising, edge integrity, edge positioning accuracy, etc., the Canny edge detection method is adopted in this embodiment.

In a specific implementation, the parameters of the Canny edge detection method are set to: a weak edge threshold of 50, a strong edge threshold of 200, and of course, may be appropriately floated as needed, for example, the threshold is floated within a range of ±10.

(3) synthesizing the edges of each video frame image;

In a specific implementation, the corresponding preset image threshold may be determined according to the number of the video frame images, and then according to whether the number of the video frame images is lower than the preset image threshold according to the edge points, whether to retain the Edge point.

That is, the correspondence between the number of video frame images and the preset image threshold is established in advance, and the corresponding relationship is searched according to the number of the video frame images to determine a corresponding preset image threshold, and the video has each edge point. When the number of frame images is lower than the preset image threshold, the edge point is not retained, and the edge point is retained when the number of video frame images is higher than or equal to the preset image threshold at each edge point.

In the following, a specific embodiment is used to describe the edges of each video frame image, but the scope of protection of the present invention is not limited: N is the number of video frame images, and X is a preset image threshold.

When N=6, correspondingly, X=4, that is, only the edge points are retained when there are more than 4 (including 4) video frame images, if the edge points are below 3 (including 3) Discard when there is a video frame image;

When 6>N>3, correspondingly, X=3, that is, only the edge points are retained when there are more than 3 (including 3) video frame images, if the edge points are below 2 (including 2 ) is discarded when there is a video frame image;

When N ≤ 3, correspondingly, X = N, that is, only the edge points are retained in the presence of all video frame images, and other cases are discarded.

Of course, the parameters in the corresponding relationship may be adjusted according to the resolution of the image, which is not limited in this embodiment.

Taking CCTV5 as an example, Figure 4 shows the overall effect of the synthesis.

The edge noise, the black border, and the non-essential characters all affect the recognition accuracy. To further improve the recognition accuracy, the synthesized edge is optimized. In this embodiment, the optimization process includes: edge noise deletion, At least one of black edge removal and unnecessary text deletion.

(4) obtaining a minimum circumscribed matrix of the synthesized edges;

(5) segmenting each video frame image according to the minimum outer matrix, and synthesizing the segmented images by weighted averaging to obtain a logo area including the station identifier to be identified.

S102: When the to-be-identified station is marked as a CCTV station logo, the to-be-identified station logo is identified according to a first preset policy;

In the specific implementation, since the difference between the CCTV logos is only that the numbers and characters are different, the logos to be identified may be identified according to the first preset policy for the features of the CCTV logo.

S103: When the to-be-identified station is marked as a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.

It can be understood that the to-be-identified station logo can be identified according to the second preset policy for the feature of the non-CCTV logo.

In this embodiment, the identification of each station label is not performed in the same manner, but it is first determined whether the station identifier to be identified in the station label area is a CCTV station label, and then different strategies are used for identifying the CCTV station logo and the non-CCTV station logo. It can effectively identify the logos of CCTV logos and non-CCTV logos, which improves the recognition efficiency and recognition accuracy.

FIG. 5 is a flowchart of a method for identifying a logo of an embodiment of the present invention; referring to FIG. 2, the method includes:

S501: determining whether the to-be-identified station logo in the logo area is a CCTV station logo;

It can be understood that, when judging whether the to-be-identified station logo in the station logo area is a CCTV station logo, a plurality of methods may be adopted, in order to ensure the accuracy of the judgment, in the present embodiment, according to the aspect ratio of the station logo area, gray The degree and color determine whether the station logo to be identified in the station logo area is a CCTV station logo.

The study found that the characteristics and differences of CCTV logos relative to satellite logos and local logos were found: (1) the aspect ratio difference, the aspect ratio of the CCTV logo (length: vertical length; width: horizontal width) is generally significantly smaller Other logos; (2) The color map of the CCTV logo has a wide range of white pixel features. Especially in the left 2/3; (3) after the CC image of the CCTV is divided into blocks, the sub-blocks meet the relevant constraints of grayscale proximity, such as mean, variance and so on.

Therefore, the logo area can be classified by the aspect ratio, the gradation, and the color of the logo area.

Because the aspect ratio is one of the most direct features of the logo. Therefore, the aspect ratio of the logo area can be first determined by the aspect ratio, that is, the aspect ratio of each station area is calculated first, and the aspect ratio is calculated by calculating the length H and the width of the station area. W, aspect ratio ratio = W / H.

The ratio of the CCTV logo is usually below 0.3. Therefore, the initial judgment condition that can be constructed is: ratio<0.3. However, the ratings of the satellites of Inner Mongolia Satellite TV, Chongqing Satellite TV, Beijing Satellite TV, etc. (including the local stations of these satellite TV stations) are all below 0.3, so they can be screened by the aspect ratio and then filtered again by grayscale and color.

When re-screening by gradation and color, it can be classified according to the following judgment conditions, that is, the following conditions for further judgment are set:

(1) The mean value of the red component in the first predetermined range in the upper left corner of the logo area and the red component in the second preset range in the lower right corner of the logo area is smaller than the preset red component.

In other words, the station standard area can be divided into 5*3 sub-blocks by row* column (of course, it can also be divided by 6*3 or 4*3, etc.), and the first sub-block area1 in the upper left corner is extracted. (ie the first preset range) and the first sub-block area2 of the lower right corner (ie the second preset range). The red distribution of the CCTV logo and the Chongqing Satellite TV logo and the Beijing Satellite TV logo are quite different in these two regions.

Combining the color characteristics of the interference station logo and the CCTV station logo, condition 1 (ie, Condition1) can be constructed—the red mean values of area1 and area2 are less than 150.

(2) The gray mean value in the third preset range on the left side of the logo area is smaller than the preset gray value.

Considering the fault tolerance of the Condition1 and the semi-transparent characteristics of the CCTV logo, the sub-area area3 is taken. 50 pixels in the width direction of the logo area can be used as the standard, and the leftmost 8 columns of pixels (that is, 4/25 in width) of the logo area are formed into area3 (ie, the third preset range).

Analyze the difference between the gray level of the interference station and the CCTV station in area3, and build condition 2 (Condition2)—the average gray level of area3 is less than 100.

It can be understood that the grayscale image can be obtained by the three-channel classical synthesis method Gray=0.33R+0.59G+0.11B, or the three-channel maximum value method, the three-channel average method, etc., and the present embodiment does not Limit it.

(3) dividing the logo area into at least four parts, and the absolute difference of the pixel mean values between the divided preset parts does not exceed a preset absolute difference;

Since Condition1 and Condition2 only contain the information of each sub-block, the constraint needs to be extended to the relationship between sub-blocks. The analysis found that the characters and digital pixels in the logo area are mainly located in the last 1/3 column of the logo area. To this end, the station standard area is divided into 2*3 sub-blocks according to the row* column (of course, it can also be divided by 3*3, etc.), and the first 2*2 sub-blocks (that is, the divided preset parts) are taken. Represented as area4, area5, area6, and area7, respectively.

The multi-sample rigorous calculation found that the CCTV station standard has the following conditions at these four sub-blocks:

Condition 3 (Condition3) - The absolute difference between the mean values of area4, area5, area6, and area7 does not exceed 100.

(4) The variance of the pixel mean values between the divided preset portions does not exceed the preset variance.

That is to say, the variances of the pixel mean values of the above four sub-blocks are different. Therefore, the CCTV station standard satisfies the following conditions at the four sub-blocks:

Condition 4 (Condition4) - The variance of the mean sequence of area4, area5, area6, and area7 does not exceed 1600.

When a certain sample sample satisfies the above Condition1~Condition4, it can be judged as the CCTV logo, otherwise it is judged as the non-CCTV logo. When the above conditions are judged at the same time, the accuracy rate is the highest. Experiments prove that the lack of any one condition will increase the error rate of multi-sample classification.

S502: When the to-be-identified station is marked as a CCTV logo, extract the digital area in the logo area, perform digit identification on the extracted digital area, and identify the number in the digital area according to the digit determination result. The combination between the CCTV logo and the identified number is used as the recognition result of the to-be-identified logo;

It should be noted that the CCTV logo includes a logo (ie, CCTV), characters, and numbers. The difference between the common CCTV logos is text and numbers. In the field of pattern recognition, digital recognition is easier, more stable, and faster than text recognition. At the same time, the CCTV logo number can describe the specific channel separately, so the embodiment will remove the text in the logo area (such as "comprehensive", "finance", etc.), and extract the numbers in the logo area (ie 1, 2, etc.) . Wherein, the text is below the station logo and has a clear pixel interval from the mark, and a preset pixel interval may be set, and whether the pixel interval between the extracted edge and the mark exceeds a preset pixel interval exceeds a preset pixel interval. In this case, it is confirmed as a character and is deleted. Therefore, the following logo area is an area including only numbers and flags.

It can be understood that the digital area is located in the logo area and has a certain positional relationship. Therefore, the positional relationship between the logo area and the digital area may be established in advance, and then according to the digital area. The position information divides the image to obtain the digital area.

For the logo area, the following relationship exists between the digital area and the logo area (ie, the area where the CCTV is located):

(1) The digital area is located on the right side of the marked area, and the occupied width is approximately equal to 1/4 of the marked area;

(2) The letters in the digital area and the marked area are equal in height, which is about 0.8 of the overall height of the CCTV logo.

Therefore, the digital area in the logo area can be extracted according to the positional relationship between the logo area and the digital area.

In addition, in order to facilitate the identification of the numbers in the digital area, the digital portion and the background portion in the digital area may be binarized, with the digital portion set to white and the background portion set to black.

Since the positions of the four corners of the digital area are likely to generate white pixel blocks/points, and there are still noise points in the digital area, the interference information may affect the digital recognition. In this embodiment, the binarization may be performed. The digital area performs interference information deletion.

In a specific implementation, the white pixel block/point of the four corners of the digital area may be deleted according to the following manner: the horizontal width of the digital area is W (equal to 0.25 W _A , and W _A is the width of the logo area) ), the vertical length is H (equal to H _A , H _A is the height of the logo area), each

For noise points in the digital region, noise filtering can be performed to further attenuate and reduce noise point effects.

In order to further improve the accuracy of digital recognition, in the present embodiment, the extracted digital region is first subjected to bit number discrimination, and the number in the digital region is identified based on the bit number discrimination result.

It can be understood that the method for determining the extracted digital area can be performed in various manners. In this embodiment, the gray value of each pixel in the digital area is projected in columns to form a horizontal width of the digital area. Projection vector with more than a preset number in a column of projection vectors When the pixel belongs to the digital part, the column projection vector is identified. If the minimum distance between two adjacent identified projection vectors is greater than the preset distance, the digital discrimination result is set to two bits, otherwise The numerical discrimination result is set to one bit.

In order to further improve the efficiency of the digital recognition, in the present embodiment, the numbers in the digital area can be identified according to the number of digits determination result by the following three steps:

(1) Acquiring the white pixel area A in the digital area, and calculating the horizontal width w and the vertical line width h of A. If h/w>2, the station is marked as the station logo of CCTV-1. Otherwise enter (2).

(2) Perform edge extraction on the digital area, if the block is divided into m*n sub-blocks by row * column, as shown in Table 1. The histogram of the edge point probability spatial distribution is constructed, and the matching probability is calculated with the histogram of the edge probability of the digital image edge point of the standard 0, 2-9.

Table 1 block parameters

(3) Identifying the number in the digital area based on the result of the digital discrimination. If the digital discrimination result is one bit, the probability that the station logo is CCTV-1, CCTV-10~CCTV15 is 0. If the digital discrimination result is two digits, the probability that the station logo is CCTV1~CCTV9 is 0, thus completing the Digital identification in the digital area.

Considering the problem of whether the identification is accurate, in the embodiment, the digital area is matched with the standard number according to the bit number discrimination result, and when the highest matching rate and the second highest matching rate are not equal, corresponding to the highest matching rate. The standard number is used as the number in the digital area.

Of course, when the highest matching rate and the second highest matching rate are equal, no unique identification is made.

S503: Calculate a matching ratio between an edge of the logo area and each standard station mark when the to-be-identified station is marked as a non-CCTV station logo;

Understandably, since the non-CCTV station logo usually includes: a TV station logo and a local station logo, in this embodiment, considering the characteristics and differences of the CCTV logo, it can be divided into no-text logo (such as Oriental TV), and Separate text logos (such as Southeast Satellite TV, which is the above-mentioned "non-essential text") and have text labels that cannot be completely separated (such as Hebei Satellite TV, which is the "necessary text" mentioned above). Therefore, it will be based on the standard library. Three different types of satellite station logos are used for corresponding types of standard logos. Construct.

In order to facilitate the calculation of the matching ratio between the edge of the logo area and each standard station mark, in the embodiment, the edge in the station label area is matched with the standard station logo by the spatial distribution histogram to calculate the station label. The matching rate between the edge of the area and each standard station mark.

Because Jiangsu Satellite TV, Dongfang Satellite TV and other logos only have round and elliptical logos (this type of station logo is defined as short station logo), the length and width are relatively small, and the TV stations of Inner Mongolia Satellite TV and Xinjiang Satellite TV (such stations) The aspect ratio defined as long platform is significantly larger than the aspect ratio of these satellite stations. Therefore, the aspect ratio can be used as a classification condition. The ratio of the long station logo is above 1.5, and the aspect ratio comparison condition of the recognition level is: ratio ≥ 1.5. After the classification, when matching is performed, the matching can be directly performed according to the standard logo of the same aspect ratio type, without All standard logos are matched to further reduce the matching duration.

In this embodiment, the edge Edge of the logo area is divided into the same number of sub-blocks A1, A2, A3, ... Am, where m is the number of sub-blocks. For each sub-block, the edge point probability is counted, that is, the spatial distribution histogram is obtained. For example, the Edge of the short station is divided into 5*5 sub-blocks, that is, 25 sub-blocks; the Edge of the long-station label is divided into 5*10 (rows*columns) sub-blocks, that is, 50 sub-blocks. Each standard station mark in the standard library is also calculated according to the short and long spatial distribution histogram, and the spatial distribution histogram of the station label area is matched with the standard station label sub-block. The number of sub-blocks to be matched is n (initial value is 0), and the total number of sub-blocks is N. If the probability of a sub-block is very close (for example, probability difference <0.05), then n=n+1; matching ends, calculation The matching ratio is p=n/N. This results in a matching rate array P that traverses all standard logos.

That is to say, the parameter Δp is used to determine whether pixel information matching is required. Δp is defined as: Δp=p _max -p _second , where p _max is the maximum value of the matching rate array P (ie, the highest matching rate), and p _second is the next largest value other than the maximum value in the matching rate array P (ie the second highest match rate).

S504: When the difference between the highest matching rate and the second highest matching rate is greater than the first preset difference, the standard station corresponding to the highest matching rate is used as the recognition result of the to-be-identified station target;

Taking into account the matching identify certain fault tolerance, △ p≥0.25 judgment condition setting (i.e., a first preset difference, of course, but also for other values), if the determination condition △ p≥0.25 established, then the corresponding p _max The standard station logo is used as the identification result of the logo to be identified.

S505: determining, when the difference between the highest matching rate and the second highest matching rate is less than or equal to the first preset difference, determining whether the standard logo corresponding to the highest matching rate or the second highest matching rate belongs to a local station label;

When the determination condition Δp ≥ 0.25 is not established, it is determined whether the to-be-identified station logo is a local station label. In this embodiment, it is required to determine whether the standard logo corresponding to the highest matching rate or the second highest matching rate belongs to the locality. Taiwan standard.

S506: If the standard station label corresponding to the highest matching rate or the second highest matching rate does not belong to the local station label, the standard station label corresponding to the highest matching rate and the second highest matching rate is used as a standard standard to be matched, and The color information of the logo area is matched with the standard station to be matched to update the matching rate of the standard station to be matched, and the to-be-matched standard with a high matching rate is used as the recognition result of the to-be-identified station logo.

Since the standard logo corresponding to the highest matching rate or the second highest matching rate does not belong to the local station logo, the to-be-identified station logo does not belong to the local station logo, and the possibility of belonging to the satellite station logo is very high.

In this embodiment, by matching the color information of the logo area with the standard station to be matched, to update the matching rate of the standard station to be matched, the following steps can be performed:

(1) Extract the RGB color three-channel gray matrix M _R , M _G , M _{B of the} logo area.

(2) extracting the three-channel gray-scale region matrix M _0.5R , M _0.5G , M _{0.5B in the logo area} , the width and height of the three are half of M _R , M _G , M _B , respectively The pixel origin coordinates are as shown in equation (1). Where gray represents R, G, B, x is the row coordinate, y is the column coordinate, and H is the moment.

For subsequent matching.

(3) Matrix partitioning. M _0.5R , M _0.5G , and M _0.5B were separately divided. As the size of the area is reduced, the block scheme is modified as follows: the edge of the short station is divided into 3*3 sub-blocks, that is, 9 areas; the edge of the long station is divided into 3*6 (rows*columns) sub-blocks, that is, 18 areas. .

(4) RGB three-channel spatial distribution histogram construction. According to the blocking scheme of step (3), the average red gray value m _Ri , the average green gray value m _Gi , and the average blue gray value m _{Bi of} each sub-block region are calculated, and the standard logo in the standard library is the same. The block average is calculated and a three-channel mean match is performed. The number of matching regions is n _rgb , and the total region number is N _rgb . If the matching difference of the three channels (the difference between the mean gray level of the sample and the standard gray mean in the same sub-block region) is less than 50, then n _rgb =n _Rgb +1. At the end of the match, the matching ratio is calculated as p _rgb =n _rgb /N _rgb . This results in an RGB matching rate array P _rgb traversing all standard station marks, where the matching rate of similar stations is p _rgb and the matching rate of non-similar stations is 0.

(5) Matching rate update. The results of the matching ratio arrays P and P _rgb are combined to update the matching rate array P _new =P+P _rgb .

S507: If the standard station label corresponding to the highest matching rate or the second highest matching rate belongs to the local station label, determining the area to which the to-be-identified station label belongs according to the matching ratio of the edge of the station label area and each standard station label Separating the text area in the logo area, separating the single word part in the text area, identifying the separated single word part, and matching the single word part recognition result with the standard station logo corresponding to the area, The matching standard station logo is used as the recognition result of the to-be-identified station logo.

It can be understood that the standard logo corresponding to the region can be understood as all standard logos of the region.

Since the standard station label corresponding to the highest matching rate or the second highest matching rate belongs to the local station label, at this time, the possibility that the to-be-identified station label belongs to the local station label is very high, and usually corresponds to a plurality of standard stations in the same area. Each has a high matching rate, and in this case, the area to which the to-be-identified station logo belongs can be determined according to the matching rates.

In the present embodiment, the area to which the to-be-identified station label belongs is determined according to the number of standard station labels in each area in the obtained standard station logo and the total number of standard station labels in each area, and the following criteria can be adopted:

Obtaining the standard station label (the standard station label corresponding to the highest matching rate) whose difference between the matching rate and the highest matching rate is smaller than the first preset difference, and determining the target of the most standard station label according to the region to which each standard station label is obtained. a region, determining whether a ratio between a number of standard logos of the target area in each of the obtained standard logos and a total number of standard logos corresponding to the target area exceeds a preset ratio, and if yes, using the target area as the Identify the area to which the logo belongs.

For example, the difference between the matching rate and the highest matching rate is less than the first preset difference of 10 standard stations, and 8 of the 10 standard stations belong to Beijing TV, 1 belongs to Sichuan TV, and 1 belongs to Xinjiang TV. There are 12 TV stations in Beijing, 6 TV stations in Sichuan, and 5 TV stations in Xinjiang. The preset ratio is 50%. The target area is Beijing, and the number of standard stations in Beijing is obtained. That is, 8) and the total number of standard logos corresponding to Beijing (ie 12) is 72.7%, which exceeds the preset ratio (50%). At this time, Beijing is used as the area to which the logo to be identified belongs. .

It should be noted that there are multiple ways to extract the text area in the logo area. The extraction efficiency and accuracy of the high text area, in the embodiment, determining the position information of the text area according to the positional relationship between the logo area and the text area and the position information of the logo area, and according to the text area The location information is extracted from the logo area. Of course, the method may be otherwise limited.

Taking the logo of the station to be identified as the Beijing TV station logo as an example, the text area is the same as the logo “BTV” in the Beijing TV station. Therefore, the right side of the cutable sign “BTV” is twice the height of the station logo area. As the text area, that is, the height of the text area is the same as the height of the logo "BTV", and the length of the text area is twice the height of the logo area;

In order to further improve the recognition accuracy, in the present embodiment, the text region is subjected to Ostu binarization processing, and the connected domain method is used to remove noise interference.

Understandably, for some local landmarks, such as the Beijing TV station logo, the difference lies in the text, such as "Beijing Satellite TV", "sports", "life" and so on. The difference can be uniquely described by the first character in the text area. In order to improve the recognition efficiency, in the embodiment, the separated single word part is the single word part where the first character in the text area is located, and can be separated by the following process. :

(1) removing the horizontal and vertical interference lines (segments) in the text area;

(2) acquiring a pixel interval between the word portions in the text area;

(3) extracting a complete single word part according to the pixel interval, such as: text (literary art), shadow (video);

(4) Perform threshold segmentation on the single word part. In the process of segmentation, take the 5*5 subfield in the upper left corner and calculate the black ratio r. If r is less than 0.5, exchange the black and white pixels of the single word part.

Of course, all the word parts can be separated and identified separately, which is not limited in this embodiment.

In the prior art, there are various methods for identifying the separated single-word portions, but in order to improve the recognition accuracy and stability, referring to FIG. 2, the identification of the separated single-word portions includes:

A1: refining the word part;

In this embodiment, the single word portion is refined to obtain a refined edge map of the single word portion, and the process of the refinement process may refer to a practical parallel refinement algorithm written by Lu Yue and Shi Pengfei. The implementation thereof is not repeated here.

In order to facilitate subsequent processing, the standard words in the single word set can also be refined to obtain a refined edge map of each standard word;

In order to keep the refinement edge map of the single-word portion and the refinement edge map of each standard word consistent, for example, the refined edge map of the single-word portion and the refined edge map of each standard word may be converted into 50 *50 pixel size.

A2: calculating an edge matching ratio between the single word portion and the standard word according to the refined edge of the refined word portion;

The edge features of the refined text include strokes depicting the text, which narrows the range of matching standard data, while the regional features are refined text content, enriching the area description, and correcting the matching text, because the refined data volume is smaller than In the present embodiment, the edge feature is matched first, and then the region feature is corrected, thereby reducing the amount of calculation.

In this embodiment, when calculating the edge matching ratio between the single word portion and the standard word according to the refined edge of the single word portion after the thinning process, the following process may be adopted:

(1) traversing the refined edge map si of the standard word, dividing the si into 100 sub-pictures of 10*10;

(2) Calculate the edge ratio pi_j of each subgraph to form a refined edge scale matrix Mi, and calculate the refined edge scale matrix M of the single word part by the same reason;

(3) comparing the approximation of the same position elements in M and Mi, and then calculating the edge matrix matching degree Pe_i;

(4) According to Ps_i+=Pe_i (Ps_i is the i-th element of Ps), form a new probability array Ps;

A3: determining whether the matching rate difference between the highest edge matching rate and the second highest edge matching rate is higher than the second preset difference;

That is to say, it is judged that the highest value Ps_m1 and the second highest value Ps_m2 of Ps satisfy the condition:

Ps_m1-Ps_m2≧0.25 (0.25 can be set to other values),

A4: when the matching rate difference is higher than the second preset difference, the standard word corresponding to the highest edge matching rate is used as the word recognition result of the single word area;

That is, when Ps_m1 - Ps_m2 ≧ 0.25 is satisfied, the character corresponding to Ps_m1 is recognized and the matching ratio is updated.

A5: when the matching ratio difference is not higher than the second preset difference, the edge matching rate is updated according to the regional feature of the refined word part and the standard word matching;

That is to say, when Ps_m1-Ps_m2≧0.25 is not satisfied, it can be corrected according to the regional characteristics by the following process:

Defining the pixel-by-region pixel of the refined word-by-region pixel to determine whether the pixel at the same position of the standard word after the thinning process is a region pixel, and calculating a whole map matching ratio pi_ds;

Defining the normalized single-word-by-region pixel to determine whether the pixel at the same position of the single-word portion after the thinning process is a region pixel, and calculating a whole map matching ratio pi_sd;

According to Ps_i+=(pi_ds+2*pi_sd)/3 (Ps_i is the i-th element of Ps), a new probability array Ps is formed;

A6: The standard word corresponding to the highest edge matching rate is used as the word recognition result of the single word area.

In other words, the character corresponding to the maximum value Ps_m of Ps is used as the recognized character, and the matching ratio is corrected.

Table 1 shows the average recognition time for several typical stations using the method of the present embodiment. The traversal recognition time is short and the difference is not large, both are around 2s.

Table 1 Average recognition time of typical TV station logos

Television station	Station identification time (s)
CCTV1	1.44
CCTV5	1.29
Phoenix Television	1.76
Southeast Satellite TV	1.93
Beijing life	2.17
Beijing Sports	2.25

Table 2 shows the average matching ratio and recognition rate of several typical stations using the method of the present embodiment. Table 2 demonstrates the effectiveness, stability, and reliability of the method of the present embodiment.

Table 2 Average matching rate and recognition rate of typical TV stations

The method of the present embodiment is suitable for matching identification of CCTV stations, satellite stations and local stations, achieving high recognition rate recognition of more than 95%, and controlling the traverse recognition time within 3s.

For the method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should understand that the embodiments of the present invention are not limited by the described action sequence, because the embodiment according to the present invention Some steps can be performed in other orders or at the same time. In the following, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

6 is a structural block diagram of a station identification device according to an embodiment of the present invention; referring to FIG. 6, the device includes:

The station label determining unit 601 is configured to determine whether the station label to be identified in the station label area is a CCTV station label;

The first identifying unit 602 is configured to identify the to-be-identified station logo according to the first preset policy when the to-be-identified station is marked as a CCTV station logo;

The second identifying unit 603 is configured to identify the to-be-identified station logo according to the second preset policy when the to-be-identified station is marked as a non-CCTV station logo.

In an optional embodiment of the present invention, the station identifier determining unit is further configured to determine, according to the aspect ratio, the gray scale, and the color of the logo area, whether the station logo to be identified in the logo area is a CCTV logo .

In an optional embodiment of the present invention, the first identifying unit is further configured to extract a digital area in the logo area, performing bit number discrimination on the extracted digital area, identifying a number in the digital area according to the bit number discrimination result, and using a combination between the CCTV logo mark and the recognized number as the The identification result of the station logo to be identified.

In an optional embodiment of the present invention, the first identifying unit is further configured to extract a digital area in the logo area according to a positional relationship between the logo area and the digital area, and The digital portion and the background portion in the digital area are binarized.

In an optional embodiment of the present invention, the first identifying unit is further configured to project a gray value of each pixel in the digital area by a column to form a horizontal width of the digital area. a projection vector, when there are more than a preset number of pixels belonging to the digital part in a column of projection vectors, the column projection vector is identified, if there is a minimum distance between two adjacent identified projection vectors greater than a preset distance, Then, the digital discrimination result is set to two bits, otherwise the digital discrimination result is set to one bit.

In an optional embodiment of the present invention, the second identifying unit is further configured to calculate a matching ratio between an edge of the logo area and each standard station label; the difference between the highest matching rate and the second highest matching rate is greater than When the first preset difference is used, the standard station corresponding to the highest matching rate is used as the recognition result of the to-be-identified station logo; when the difference between the highest matching rate and the second highest matching rate is less than or equal to the first preset difference, the determining unit Whether the standard logo corresponding to the highest matching rate or the second highest matching rate belongs to the local station label; if the standard label corresponding to the highest matching rate or the second highest matching rate does not belong to the local station label, the highest matching rate is The standard station label corresponding to the second highest matching rate is used as a standard station logo to be matched, and the color information of the station label area is matched with the standard station label to be matched to update the matching rate of the standard station label to be matched. The criterion to be matched with a high matching rate is used as the recognition result of the to-be-identified station logo.

In an optional embodiment of the present invention, the second identifying unit is further configured to match an edge in the logo area by a spatial distribution histogram with a standard logo to calculate the logo area. The matching rate between the edge of each standard and each standard.

In an optional embodiment of the present invention, the second identifying unit is further configured to: when the standard label corresponding to the highest matching rate or the second highest matching rate belongs to a local station logo, according to the station label area And determining a region to which the to-be-identified logo belongs, separating a text region in the logo region, separating a single-word portion in the text region, and performing the separated single-word portion Identification, the standard corresponding to the word recognition result of the word part The logo is matched by the logo, and the matching standard logo is used as the recognition result of the logo to be identified.

In an optional embodiment of the present invention, the second identifying unit is further configured to perform refinement processing on the single word portion; and calculate the single word portion according to the refined edge of the refined word portion The edge matching rate between the standard words; determining whether the matching rate difference between the highest edge matching rate and the second highest edge matching rate is higher than the second preset difference; wherein the matching rate difference is higher than the second preset difference And the standard single word corresponding to the highest edge matching rate is used as a single word recognition result of the single word area; when the matching rate difference is not higher than the second preset difference, according to the refined word part The area feature is matched with the standard word, and the edge matching rate is updated; the standard word corresponding to the highest edge matching rate is used as the word recognition result of the single word area.

In an optional embodiment of the present invention, the device further includes:

The area obtaining unit is configured to obtain a video frame image sequence from a preset area of the video that includes the to-be-identified station logo, perform edge extraction on each video frame image, and combine edges of each video frame image to obtain a minimum external connection of the synthesized edge. a matrix, respectively dividing each video frame image according to the minimum outer matrix, and synthesizing the segmented images by weighted averaging to obtain a logo area including a station identifier to be identified.

FIG. 7 is a block diagram showing the structure of a server according to another embodiment of the present application.

Referring to FIG. 7, the server includes:

a processor 701, a memory 702, a communication interface 703, and a bus 704; wherein

The processor 701, the memory 702, and the communication interface 703 complete communication with each other through the bus 704;

The communication interface 703 is used for information transmission between the server and the user equipment;

The processor 701 is configured to invoke logic instructions in the memory 702 to perform the following methods;

Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo; when the to-be-identified station logo is a CCTV station logo, identifying the to-be-identified station logo according to the first preset policy; When the station logo is a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.

Referring to FIG. 1, another embodiment of the present invention discloses a computer program, including program code, for performing the following operations:

Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo;

When the to-be-identified station is marked as a CCTV station logo, the to-be-identified station logo is identified according to a first preset policy;

When the to-be-identified station is marked as a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.

Another embodiment of the present invention discloses a storage medium for storing a computer program as described in the foregoing embodiments.

A person skilled in the art can understand that all or part of the steps of implementing the above method embodiments may be completed by using hardware related to the program instructions. The foregoing program may be stored in a computer readable storage medium, and the program is executed when executed. The foregoing steps include the steps of the foregoing method embodiments; and the foregoing storage medium includes: a medium that can store program codes, such as a ROM, a RAM, a magnetic disk, or an optical disk.

It should be noted that the above embodiments are only used to explain the technical solutions of the embodiments of the present invention, and are not limited thereto; although the embodiments of the present invention are described in detail with reference to the foregoing embodiments, those skilled in the art It should be understood that the technical solutions described in the foregoing embodiments may be modified, or some or all of the technical features may be equivalently replaced; and the modifications or substitutions do not deviate from the embodiments of the present invention. The scope of the technical solution.

Claims (23)

1. A station identification method, characterized in that the method comprises:

   Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo;

   When the to-be-identified station is marked as a CCTV station logo, the to-be-identified station logo is identified according to a first preset policy;

   When the to-be-identified station is marked as a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.
2. The method according to claim 1, wherein the determining whether the to-be-identified station logo in the logo area is a CCTV logo further comprises:

   It is judged whether the to-be-identified station logo in the station logo area is a CCTV station logo according to the aspect ratio, the gray scale and the color of the station logo area.
3. The method according to claim 1, wherein the identifying the to-be-identified station logo according to the first preset policy further comprises:

   Extracting a digital area in the logo area, performing bit number discrimination on the extracted digital area, identifying a number in the digital area according to the bit number discrimination result, and using a combination between the CCTV logo mark and the recognized number as a The recognition result of the identification logo is mentioned.
4. The method according to claim 3, wherein the extracting the digital area in the logo area further comprises:

   And extracting a digital area in the logo area according to a positional relationship between the logo area and the digital area, and performing binarization processing on the digital part and the background part in the digital area.
5. The method according to claim 4, wherein said determining the number of digits of the extracted digital region further comprises:

   Projecting a gray value of each pixel in the digital area in a column to form a projection vector having a horizontal width of the digital area, and having more than a preset number of pixels belonging to the digital part in a column of projection vectors The column projection vector is identified. If there is a minimum distance between two adjacent identified projection vectors that is greater than a preset distance, the digital discrimination result is set to two bits, otherwise the digital discrimination result is set to one. Bit.
6. The method according to claim 1, wherein the identifying the to-be-identified station logo according to the second preset policy further comprises:

   Calculating a matching ratio between an edge of the logo area and each standard station mark;

   When the difference between the highest matching rate and the second highest matching rate is greater than the first preset difference, the standard station corresponding to the highest matching rate is used as the recognition result of the to-be-identified station logo;

   When the difference between the highest matching rate and the second highest matching rate is less than or equal to the first preset difference, it is determined whether the standard logo corresponding to the highest matching rate or the second highest matching rate belongs to the local station label;

   If the standard station label corresponding to the highest matching rate or the second highest matching rate does not belong to the local station label, the standard station label corresponding to the highest matching rate and the second highest matching rate is used as the standard station logo to be matched, and the The color information of the logo area is matched with the standard standard to be matched to update the matching rate of the standard standard to be matched, and the to-be-matched standard with a high matching rate is used as the recognition result of the to-be-identified station logo.
7. The method according to claim 6, wherein the calculating the matching ratio of the edge of the logo area to each standard station mark further comprises:

   The edge in the logo area is matched with the standard logo by the spatial distribution histogram to calculate the matching ratio between the edge of the logo area and each standard station mark.
8. The method according to claim 6, wherein the method further comprises: after determining whether the standard logo corresponding to the highest matching rate or the second highest matching rate belongs to a local station logo, the method further comprises:

   If the standard station label corresponding to the highest matching rate or the second highest matching rate belongs to the local station logo, determining the area to which the to-be-identified station label belongs according to the matching ratio of the edge of the station label area and each standard station label, and dividing Extracting a text area in the logo area, separating a single word part in the text area, identifying the separated single word part, and matching the single word part recognition result with the standard station logo corresponding to the area, and matching The standard logo is used as the recognition result of the logo to be identified.
9. The method of claim 8 wherein said identifying said separated word portion further comprises:

   Refine the single word portion;

   Calculating an edge matching ratio between the single word portion and the standard word according to the refined edge of the refined word portion;

   Determining whether the edge matching rate difference between the highest edge matching rate and the second highest edge matching rate is higher than the second preset difference;

   When the edge matching rate difference is higher than the second preset difference, the standard word corresponding to the highest edge matching rate is used as the single word recognition result of the single word area;

   When the edge matching rate difference is not higher than the second preset difference, the edge matching rate is updated according to the regional feature of the refined word portion and the standard word matching;

   The standard word corresponding to the highest edge matching rate is used as the word recognition result of the single word area.
10. The method according to any one of claims 1 to 9, wherein the method further comprises: before determining whether the to-be-identified station logo in the logo area is a CCTV logo, the method further comprises:

    Obtaining a video frame image sequence from a preset area of the video including the to-be-identified station logo, performing edge extraction on each video frame image, synthesizing the edges of each video frame image, and obtaining a minimum outer matrix of the synthesized edge, according to the minimum The external matrix separately segments each video frame image, and combines the segmented images by weighted averaging to obtain a logo area including the station identifier to be identified.
11. A station identification device, characterized in that the device comprises:

    The station label determining unit is configured to determine whether the station identifier to be identified in the station label area is a CCTV station label;

    a first identifying unit, configured to: when the to-be-identified station is marked as a CCTV logo, identify the to-be-identified station logo according to a first preset policy;

    The second identifying unit is configured to identify the to-be-identified station logo according to the second preset policy when the to-be-identified station is marked as a non-CCTV station logo.
12. The device according to claim 11, wherein the station marking determining unit is further configured to determine, according to the aspect ratio, the gradation and the color of the logo area, whether the station logo to be identified in the logo area is a CCTV station Standard.
13. The apparatus according to claim 11, wherein the first identifying unit is further configured to extract a digital area in the logo area, perform bit number discrimination on the extracted digital area, and identify according to the number of digits The number in the digital area is a combination of the CCTV logo and the identified number as the recognition result of the to-be-identified station logo.
14. The apparatus according to claim 13, wherein the first identifying unit is further configured to extract a digital area in the logo area according to a positional relationship between the logo area and the digital area, and The digital portion and the background portion in the digital area are binarized.
15. The apparatus according to claim 14, wherein the first identifying unit is further configured to project a gray value of each pixel in the digital area by a column to form a horizontal width of the digital area. The projection vector, when there are more than a preset number of pixels belonging to the digital part in a column of projection vectors, the column projection vector is identified, if there is a minimum distance between adjacent two identified projection vectors is greater than a preset distance Then, the digital discrimination result is set to two bits, otherwise the digital discrimination result is set to one bit.
16. The apparatus according to claim 11, wherein the second identifying unit is further configured to calculate a matching ratio between an edge of the logo area and each standard station mark; a difference between a highest matching rate and a second highest matching rate When the difference is greater than the first preset difference, the standard station corresponding to the highest matching rate is used as the recognition result of the to-be-identified station logo; when the difference between the highest matching rate and the second highest matching rate is less than or equal to the first preset difference, the judgment is performed. Whether the standard logo corresponding to the highest matching rate or the second highest matching rate belongs to the local station label; if the standard label corresponding to the highest matching rate or the second highest matching rate does not belong to the local station label, the highest matching is performed. The standard station label corresponding to the rate and the second highest matching rate is used as a standard station logo to be matched, and the color information of the station label area is matched with the standard station label to be matched to update the matching rate of the standard station label to be matched. The criterion to be matched with a high matching rate is used as the recognition result of the to-be-identified station logo.
17. The apparatus according to claim 16, wherein the second identifying unit is further configured to match an edge in the logo area with a standard station logo by a spatial distribution histogram to calculate the station label The matching rate between the edge of the area and each standard station mark.
18. The device according to claim 16, wherein the second identifying unit is further configured to: when the standard station label corresponding to the highest matching rate or the second highest matching rate belongs to a local station logo, according to the station label Determining the area to which the to-be-identified station logo belongs by dividing the edge of the area with the matching rate of each standard station mark, separating the text area in the station label area, separating the single-word part in the text area, and separating the single-word part The identification is performed, and the single-part partial recognition result is matched with the standard standard logo corresponding to the region, and the matched standard logo is used as the recognition result of the to-be-identified logo.
19. The apparatus according to claim 18, wherein the second identifying unit is further configured to perform refinement processing on the single word portion; and calculate the single word portion according to the refined edge of the refined word portion An edge matching ratio between the standard word and the standard single word; determining whether the matching rate difference between the highest edge matching rate and the second highest edge matching rate is higher than the second preset difference; When the ratio difference is higher than the second preset difference, the standard word corresponding to the highest edge matching rate is used as the word recognition result of the single word area; and the matching rate difference is not higher than the second preset difference When the value is matched, the edge matching rate is updated according to the regional feature of the single word portion after the refinement processing, and the standard matching word corresponding to the highest edge matching rate is used as the single word recognition result of the single word region.
20. The device according to any one of claims 11 to 19, wherein the device further comprises:

    The area obtaining unit is configured to obtain a video frame image sequence from a preset area of the video that includes the to-be-identified station logo, perform edge extraction on each video frame image, and combine edges of each video frame image to obtain a minimum external connection of the synthesized edge. a matrix, respectively dividing each video frame image according to the minimum outer matrix, and synthesizing the segmented images by weighted averaging to obtain a logo area including a station identifier to be identified.
21. A server, comprising:

    a processor, a memory, a communication interface, and a bus; wherein

    The communication interface is used for information transmission between a user equipment and a server;

    The processor is configured to invoke logic instructions in the memory to perform the following method;

    Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo; when the to-be-identified station logo is a CCTV station logo, identifying the to-be-identified station logo according to the first preset policy; When the station logo is a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.
22. A computer program comprising program code for performing the following operations:

    Determining whether the to-be-identified station logo in the station logo area is a CCTV station logo;

    When the to-be-identified station is marked as a CCTV station logo, the to-be-identified station logo is identified according to a first preset policy;

    When the to-be-identified station is marked as a non-CCTV station logo, the to-be-identified station logo is identified according to a second preset policy.
23. A storage medium characterized by storing the computer program of claim 22.

Images