CN105868682A - Local channel logo identification method and apparatus - Google Patents
Local channel logo identification method and apparatus Download PDFInfo
- Publication number
- CN105868682A CN105868682A CN201510824087.9A CN201510824087A CN105868682A CN 105868682 A CN105868682 A CN 105868682A CN 201510824087 A CN201510824087 A CN 201510824087A CN 105868682 A CN105868682 A CN 105868682A
- Authority
- CN
- China
- Prior art keywords
- station symbol
- matching rate
- identified
- standard
- individual character
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/49—Segmenting video sequences, i.e. computational techniques such as parsing or cutting the sequence, low-level clustering or determining units such as shots or scenes
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/635—Overlay text, e.g. embedded captions in a TV program
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computing Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a local channel logo identification method and apparatus, and relates to the technical field of information identification. According to the method and apparatus disclosed by the present invention, global channel logo identification is not adopted any longer, but a characteristic of a local channel logo is fully considered. The method comprises: acquiring a channel logo region comprising a to-be-identified channel logo; when the to-be-identified channel logo is a local channel logo, determining a district that the to-be-identified channel logo belongs to according to a matching rate between the channel logo region and a standard channel logo; extracting a text region in the channel logo region; separating a single word part from the text region, and identifying the separated single word part; and performing text matching on a single word identification result and a standard channel logo corresponding to the district, and using the standard channel logo with the highest matching rate as an identification result. Therefore, time complexity is reduced, and local channel logos can be rapidly identified.
Description
Technical field
The present invention relates to information discriminating technology field, particularly to a kind of local TV station symbol recognition method and device.
Background technology
Intelligent television just complies with " high Qinghua ", " networking ", the trend of " intelligent " develops rapidly, possess from the Internet,
The content that the multiple channel such as video equipment, computer obtains programme content, consumer needed most by the way of information fusion
The function clearly represented on giant-screen.Compared with traditional tv, intelligent television has provided the user more quick, intelligent, people
The application service of property.
Intelligent television comprises a large amount of order video, programme televised live, and most of TV programme remain program product side
Station symbol.Station symbol is to discriminate between the important symbol of television station, contains television station's platform name, programming source, program orientation etc. important
Semantic information, is the important semantic source realizing video analysis, understanding and retrieve.The realization of TV station symbol recognition technology, will effectively realize
Programme function, and to understanding that user preferences, input value-added service tool are of great significance.
In prior art when station symbol is identified, generally employing following two scheme:
The first scheme: TV station symbol recognition scheme based on single-frame images.This kind of method using Edge Distance transformation matrix as
Feature, all uses the mode of template matching, including overall situation edge matching (sliding window travels through whole station symbol region), divided-fit surface
(CF feature is mated simultaneously, carries out point for program searching for (use manual type filter non-edge), color form fit
Class sorts) etc..
First scheme: TV station symbol recognition method based on multiple image.This kind of scheme generally can use following three kinds of methods:
One, image is split in the eigenvalue change utilizing sensitizing range (the i.e. station symbol region) pixel of continuous multiple frames sequence of frames of video,
And carry out rim detection, use sliding window and method of partition to carry out match cognization.Two, comprehensive utilization CF feature,
In frame sequence, split station symbol according to spatio-temporal invariant, utilize spatial distribution rectangular histogram to combine HSV color histogram and feature is entered
Row effectively describes, and finally utilizes SUV to complete TV station symbol recognition.Method three: calculate the change of consecutive frame image, extracts station symbol and Hu thereof
Not bending moment, and it is identified result etc. according to candidate collection and pre-set criteria.
But in prior art, the mode such as template matching all uses station symbol global recognition, do not take into full account that local broadcasting stations' target is special
Point, time complexity is higher, it is difficult to ensure that local broadcasting stations' target quickly identifies.
Summary of the invention
The embodiment of the present invention provides a kind of local TV station symbol recognition method and device, complicated in order to solve the time in prior art
Spend higher, it is difficult to ensure the defect that local broadcasting stations' target quickly identifies.
The embodiment of the present invention provides a kind of local TV station symbol recognition method, and described method includes:
Obtain the station symbol region including station symbol to be identified；
It is local broadcasting stations' timestamp at described station symbol to be identified, determines with standard unit target matching rate according to described station symbol region
Area belonging to described station symbol to be identified；
Extract the character area in described station symbol region；
From described character area, isolate individual character part, and isolated individual character part is identified；
The standard station symbol that individual character recognition result is corresponding with described area is carried out characters matching, by the highest for characters matching rate
Standard unit is denoted as recognition result.
The embodiment of the present invention provides a kind of local TV station symbol recognition device, and described device includes:
Area acquisition unit, for obtaining the station symbol region including station symbol to be identified；
Area determines unit, for being local broadcasting stations' timestamp at described station symbol to be identified, according to described station symbol region and standard
The matching rate of station symbol determines the area belonging to described station symbol to be identified；
Word Input unit, for extracting the character area in described station symbol region；
Separate recognition unit, for isolating individual character part from described character area, and to isolated individual character part
It is identified；
Result acquiring unit, for the standard station symbol that individual character recognition result is corresponding with described area is carried out characters matching,
Standard unit the highest for characters matching rate is denoted as recognition result.
The present invention no longer uses station symbol global recognition, but takes into full account local broadcasting stations' target feature, and acquisition includes to be identified
The station symbol region of station symbol, is local broadcasting stations' timestamp at described station symbol to be identified, mates with standard unit target according to described station symbol region
Rate determines the area belonging to described station symbol to be identified, extracts the character area in described station symbol region, from described character area
In isolate individual character part, and isolated individual character part is identified, individual character recognition result is corresponding with described area
Standard station symbol carries out characters matching, is denoted as by standard unit the highest for characters matching rate as recognition result, thus it is multiple to reduce the time
Miscellaneous degree, it is possible to local station symbol is quickly identified.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
In having technology to describe, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below is this
Some bright embodiments, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to root
Other accompanying drawing is obtained according to these accompanying drawings.
Fig. 1 is the flow chart of the local TV station symbol recognition method of one embodiment of the present invention；
Fig. 2 be one embodiment of the present invention local TV station symbol recognition method in isolated individual character part is identified
Flow chart；
Fig. 3 is the flow chart of the local TV station symbol recognition method of one embodiment of the present invention；
Fig. 4 is the structured flowchart of the local TV station symbol recognition device of one embodiment of the present invention.
Detailed description of the invention
For making the purpose of the embodiment of the present invention, technical scheme and advantage clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
The a part of embodiment of the present invention rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under not making creative work premise, broadly falls into the scope of protection of the invention.
Fig. 1 is the flow chart of the local TV station symbol recognition method of one embodiment of the present invention；With reference to Fig. 1, described method bag
Include:
S101: obtain the station symbol region including station symbol to be identified；
It should be noted that described station symbol region is only includes to be identified target area.
It will be appreciated that described station symbol region can be extracted in several ways, in order to prevent random noise, picture from making an uproar
The impact on TV station symbol recognition of the noises such as sound, in present embodiment, obtains the station symbol district including station symbol to be identified by following steps
Territory:
(1) in the predeterminable area of the video including station symbol to be identified, video frame images sequence is obtained；
According to priori, the substantially all upper left corner of video frame images that is positioned at of television station's station symbol is (certainly, if being in other
Position, it is possible to carry out accommodation as required), therefore during station symbol detection, only need to extract fixing upper left corner area
(i.e. predeterminable area) detects region as station symbol.Existing TV station symbol recognition method is generally according to optimal region rule (GSR)
Obtaining station symbol region, present embodiment is with existing TV station symbol recognition method difference: (1) calculates all station symbols and respectively regarding
Frequently the proportional positions effectively identified in two field picture；(2) maximum magnitude of all proportions position is calculated as station symbol region segmentation
Region.As a example by the video of 1920*1080, station symbol cut zone is row starting position 80 (1/24), arranges starting position 40
(1/27), line width 450 (15/64), col width 180 (1/6), region segmentation effect as shown in Figures 2 and 3, certainly, described ratio position
Putting and can the most suitably adjust, this is not any limitation as by present embodiment.
For information unrelated in elimination image, recover or strengthen useful relevant information, improve the detectability of feature,
Limits ground simplifies data, to guarantee the reliability identified, in present embodiment, each video frame images can be carried out pretreatment,
Described pretreatment includes: at least one in region segmentation, gray processing and image enhaucament, certainly, may also include other and processed
Journey, this is not any limitation as by present embodiment.
Described pretreatment can use formula Gray=0.33R+0.59G+0.11B to carry out gray processing, certainly, it is possible to by three
The modes such as passage mean value method or triple channel maximum value process substitute, and wherein, Gray is the gray value of pixel, and R is the redness of pixel
Component, G is the green component of pixel, and B is the blue component of pixel.
The purpose of described image enhaucament is prominent station symbol region effective information, such as icon, word, numeral etc., image enhaucament
Use the gray scale stretching of 0～255 gray levels, it is also possible to rectangular histogram converter technique substitutes.
(2) each video frame images is carried out edge extracting；
It will be appreciated that the edge violent part that is variation of image grayscale, edge extracting is the key of TV station symbol recognition, edge
Integrated degree directly affects TV station symbol recognition result, and certainly, the method for edge extracting has a lot, such as Canny, LOG, Sobel, La Pu
Laplacian operater method etc..Consider the requirements such as denoising, edge integrity, edge precision, present embodiment uses Canny
Edge detection method.
In implementing, the parameter of Canny edge detection method is set to: weak edge threshold 50, strong edge threshold
200, certainly, it is possible to the most suitably float, such as, threshold value is floated in the range of ± 10.
(3) edge of each video frame images is synthesized；
In implementing, the pre-set image threshold value of correspondence, then root can be determined according to the quantity of described video frame images
Whether judge whether to retain this edge less than described pre-set image threshold value in the quantity of video frame images according to described each marginal point
Point.
It is to say, the corresponding relation pre-build between the quantity of video frame images and pre-set image threshold value, according to institute
The quantity stating video frame images searches corresponding relation, to determine the pre-set image threshold value of correspondence, has each marginal point at video
When the quantity of two field picture is less than described pre-set image threshold value, do not retain this marginal point, at each marginal point at the number of video frame images
When amount is optionally greater than described pre-set image threshold value, retain this marginal point.
Illustrate to synthesize the edge of each video frame images with a specific embodiment below, but do not limit this
Bright protection domain: set the N quantity as video frame images, X is pre-set image threshold value.
As N=6, correspondingly, X=4, say, that the video frame images that only marginal point (includes 4) more than 4
In the presence of in just retain, if marginal point below 3, (include 3) video frame images in the presence of; give up；
When 6 > N > 3 time, correspondingly, X=3, say, that the video frame images that only marginal point (includes 3) more than 3
In the presence of in just retain, if marginal point below 2, (include 2) video frame images in the presence of; give up；
When N≤3, correspondingly, X=N, say, that just protect in the presence of only marginal point is in all video frame images
Staying, other situations are all given up.
Certainly, the parameter in described corresponding relation can be adjusted according to the resolution of image, and present embodiment is to this not
It is any limitation as.
All recognition accuracy can be impacted due to edge noise, black surround and inessential word etc., for improving further
Recognition accuracy, the edge that can synthesize is optimized process, and in present embodiment, described optimization processes and includes: edge noise is deleted
Remove, black surround remove and inessential word delete at least one.
S102: be local broadcasting stations' timestamp at described station symbol to be identified, according to described station symbol region and standard unit target matching rate
Determine the area belonging to described station symbol to be identified；
It will be appreciated that owing to being local broadcasting stations' timestamp at described station symbol to be identified, it will usually with the mark in multiple identical areas
Quasi-station symbol is respectively provided with higher matching rate, in the case, can determine described station symbol institute to be identified according to these matching rates
The area belonged to.
S103: extract the character area in described station symbol region；
It should be noted that the character area extracted in described station symbol region has various ways, for improving character area
Extraction efficiency and accuracy rate, present embodiment, according to the position relationship between station symbol region and character area and described
The positional information in mark region determines the positional information of described character area, and according to the positional information of described character area from described
Extracting character area in station symbol region, certainly, also by other modes, this is not any limitation as by present embodiment.
S104: isolate individual character part from described character area, and isolated individual character part is identified；
Owing to character area having between different literals certain pixel separation, so, according to pixel separation from described
Character area is isolated individual character part.
It addition, for the ease of isolated individual character part is identified, according to pixel separation from described character area
In isolate individual character part after, also described individual character part can be carried out binary conversion treatment.
S105: the standard station symbol that individual character recognition result is corresponding with described area is carried out characters matching, by characters matching rate
The highest standard unit is denoted as recognition result.
It will be appreciated that the standard station symbol of described area correspondence can be regarded as all standard station symbols in described area.
Present embodiment no longer uses station symbol global recognition, but takes into full account local broadcasting stations' target feature, obtains and includes treating
Identify the station symbol region of station symbol, be local broadcasting stations' timestamp at described station symbol to be identified, according to described station symbol region and standard unit target
Matching rate determines the area belonging to described station symbol to be identified, extracts the character area in described station symbol region, from described word
Region is isolated individual character part, and isolated individual character part is identified, by right with described area for individual character recognition result
The standard station symbol answered carries out characters matching, is denoted as by standard unit the highest for characters matching rate as recognition result, thus when reducing
Between complexity, it is possible to local station symbol is quickly identified.
There is the multiple method that isolated individual character part is identified in the prior art, but in order to improve identification standard
Really rate and stability, with reference to Fig. 2, described is identified including to isolated individual character part:
A1: described individual character part is carried out micronization processes；
In present embodiment, described individual character part is carried out micronization processes, to obtain the refinement edge of described individual character part
Figure, the process of micronization processes can refer to be executed, by Lv Yuehe, " a kind of practical parallel thinning algorithm and the realization thereof " that roc flies to write, this
Embodiment does not repeats them here.
For the ease of subsequent treatment, also the standard one-letter that individual character is concentrated can be also carried out micronization processes, to obtain each standard
The refinement edge graph of individual character；
Size for the refinement edge graph of the refinement edge graph and each standard one-letter that make described individual character part keeps consistent, example
: can by as described in the refinement edge graph of individual character part and the refinement edge graph of each standard one-letter be transformed to 50*50 pixel size.
A2: according to described in the refinement edge calculations of the individual character part after micronization processes between individual character part and standard one-letter
Edge matching rate；
The edge feature of refinement word contains the stroke featuring word, can reduce the scope of coupling normal data, and
Provincial characteristics is the content of refinement word, enriches region description, can revise and determine matching literal, owing to the data volume of refinement is little
In provincial characteristics, present embodiment, first pass through Edge Feature Matching, then by provincial characteristics correction, thus reduce operand.
In present embodiment, according to individual character part and standard described in the refinement edge calculations of the individual character part after micronization processes
During edge matching rate between individual character, below scheme can be passed through:
(1) travel through the refinement edge graph si of standard one-letter, si is divided into 100 subgraphs of 10*10；
(2) calculate each subgraph refinement edge ratio pi_j, composition refinement edge scaling matrices Mi, be in like manner calculated individual character
The refinement edge scaling matrices M of part；
(3) compare the degree of approximation of same position element in M and Mi, and then calculate matrix of edge matching degree Pe_i；
(4) according to Ps_i+=Pe_i the i-th element of Ps (Ps_i be), new probability array Ps is formed；
A3: judge whether the matching rate difference of high rim matching rate and second highest edge matching rate is preset higher than first poor
Value；
That is, it is judged that the peak Ps_m1 of Ps and time high level Ps_m2 meet condition:
Ps_m1-Ps_m2 0.25 (0.25 can be set to other value),
A4: when described matching rate difference is higher than the first preset difference value, by the described standard that high rim matching rate is corresponding
Individual character is as the individual character recognition result in described individual character region；
It is to say, when meeting Ps_m1-Ps_m2 0.25, be identified as word corresponding for Ps_m1 and update matching rate.
A5: when described matching rate difference is not higher than the first preset difference value, according to the district of the individual character part after micronization processes
Characteristic of field and standard one-letter mate, and are updated described edge matching rate；
It is to say, when being unsatisfactory for Ps_m1-Ps_m2 0.25, can be repaiied by below scheme according to provincial characteristics
Just:
Individual character part region-by-region pixel after micronization processes is determined the picture of the standard one-letter same position after micronization processes
Whether element is area pixel, calculates whole figure matching rate pi_ds；
Standard one-letter region-by-region pixel after micronization processes is determined the picture of the individual character part same position after micronization processes
Whether element is area pixel, calculates whole figure matching rate pi_sd；
According to Ps_i+=(pi_ds+2*pi_sd)/3 the i-th element of Ps (Ps_i be), form new probability array Ps；
A6: using standard one-letter corresponding for high rim matching rate as the individual character recognition result in described individual character region.
It is to say, using word corresponding for the maximum Ps_m of Ps as identifying word, and revise matching rate.
Present embodiment, in revising coupling, only considers area pixel, does not use blocked histogram matching method, can be effective
Ground reduces the area pixel quantitative error impact that binaryzation is brought；On the other hand, the result of bi-directional matching is more reliable and more stable.
Fig. 3 is the flow chart of the local TV station symbol recognition method of one embodiment of the present invention；With reference to Fig. 3, described method bag
Include:
S301: obtain the station symbol region including station symbol to be identified；
Step S301 is identical with step S101 of the embodiment shown in Fig. 1, does not repeats them here.
S302: judge that whether the station symbol to be identified in station symbol region is according to length-width ratio, gray scale and the color in station symbol region
CCTV's station symbol, when described station symbol to be identified is non-CCTV's station symbol, performs step S303；
Will be understood that, it is judged that when whether the station symbol to be identified in station symbol region is CCTV's station symbol, various ways can be used,
For ensureing the accuracy rate judged, in present embodiment, judge in station symbol region according to length-width ratio, gray scale and the color in station symbol region
Station symbol to be identified whether be CCTV's station symbol.
Research finds, CCTV's station symbol finds relative to satellite TV's station symbol, local broadcasting stations' target feature and difference: (1) length-width ratio difference,
The length-width ratio of CCTV's station symbol is (long: vertical direction length；Wide: horizontal direction width) generally it is significantly less than other station symbols；(2) CCTV
The cromogram of station symbol has white pixel feature widely, especially shows at left side 2/3；(3) the gray-scale map piecemeal of CCTV's station symbol
After, meet the related constraint that gray scale is close between sub-block, such as average, variance etc..
So, by the difference of length-width ratio, gray scale and the color in described station symbol region, described station symbol region can be carried out
Classification.
Owing to length-width ratio is one of the most direct feature of station symbol.So, can first described station symbol region be entered by length-width ratio
Row is preliminary to be judged, say, that first calculate the length-width ratio in each station symbol region, and the method calculating length-width ratio is: calculate station symbol region
Length H and width W, length-width ratio ratio=W/H.
The ratio of CCTV's station symbol generally below 0.3, so, the preliminary Rule of judgment that can build is: ratio < 0.3.But
It is that the ratio of satellite TV's platform (including the local broadcasting stations of these satellite TVs) station symbols such as Inner Mongol satellite TV, Chongqing satellite TV, BTV all exists
Less than 0.3, so, after can being screened by length-width ratio, more again screened by gray scale and color.
When again being screened by gray scale and color, can classify according to following Rule of judgment, say, that arrange
Below for the condition determined whether:
(1) red component in first preset range in the upper left corner, described station symbol region and the lower right corner, described station symbol region
The average of the red component in the second preset range is less than presetting red component.
It is to say, can be divided into 5*3 sub-block (certainly, it is possible to entered by modes such as 6*3 or 4*3 by row * row in station symbol region
Row piecemeal), extract first sub-block area1 (the i.e. first preset range) and first sub-block area2 in the lower right corner in the upper left corner
(the i.e. second preset range).CCTV's station symbol and Chongqing satellite TV station symbol, BTV's station symbol etc. are in the red distribution in the two region
Completely different.
Integrated interference station symbol and the color characteristic of CCTV's station symbol, can build condition 1 (i.e. Condition1) area1 and
The red average of area2 is less than 150.
(2) on the left of described station symbol region, the gray average in the 3rd preset range is less than and presets gray value.
Consider fault-tolerance and the translucent feature of CCTV's station symbol of Condition1, take sub-block area3.Can be with station symbol region
50 pixels of width be standard, take the most left 8 row pixels in station symbol region (the widest 4/25) constitute area3 (the i.e. the 3rd is pre-
If scope).
Analysis interference station symbol and CCTV station symbol, at the gray difference of area3, build condition 2 (Condition2)
The gray average of area3 is less than 100.
It will be appreciated that gray-scale map can be obtained by triple channel classics synthetic method Gray=0.33R+0.59G+0.11B
Picture, it is possible to by triple channel maximum value process, triple channel mean value method etc., this is not any limitation as by present embodiment.
(3) it is at least 4 parts by described station symbol region segmentation, the pixel average between the predetermined fraction after segmentation exhausted
To difference less than presetting absolute difference；
Owing to Condition1 and Condition2 contains only the self information of each sub-block, need to be by constraint expansion to sub-block
Between relation.Analyzing and find, the word in station symbol region, digital pixel are predominantly located at rear 1/3 row in station symbol region.To this end, will
Station symbol region is divided into 2*3 sub-block (certainly, it is possible to carry out piecemeal by modes such as 3*3) according to row * row, takes front 2*2 sub-block and (i.e. divides
Predetermined fraction after cutting) it is expressed as area4, area5, area6 and area7.
Multisample strictly calculates discovery, and CCTV station symbol is substantially following condition at these 4 sub-blocks:
The absolute difference of the average of condition 3 (Condition3) area4, area5, area6 and area7 is less than 100.
(4) variance of the pixel average between the predetermined fraction after segmentation is less than presetting variance.
It is to say, the variance of the pixel average of above 4 sub-blocks all has difference, so, CCTV station symbol is in these 4 sub-blocks
Place meets following condition:
The variance of the equal value sequence of condition 4 (Condition4) area4, area5, area6 and area7 is less than
1600。
When certain station symbol sample meets above Condition1～Condition4 simultaneously, CCTV's station symbol can be judged as,
Otherwise it is judged as non-CCTV station symbol.When above-mentioned condition judges simultaneously, accuracy rate is the highest, it is demonstrated experimentally that lack any one condition
All can improve the error rate of multisample classification.
S303: calculate the edge in described station symbol region and each standard unit target matching rate, the highest matching rate with second highest
When joining the difference of rate less than or equal to the second preset difference value, it is judged that the highest described matching rate or standard station symbol corresponding to second highest matching rate are
No belong to local station symbol, if the highest described matching rate or standard station symbol corresponding to second highest matching rate belong to local station symbol, then assert
Described station symbol to be identified is local station symbol；
Owing to the highest described matching rate or standard station symbol corresponding to second highest matching rate belong to local station symbol, now, treat described in
Identify that station symbol belongs to local broadcasting stations' target probability the highest, it will usually be respectively provided with higher with the standard station symbol in multiple identical areas
Matching rate, in the case, can determine the area belonging to described station symbol to be identified according to these matching rates.
S304: be local broadcasting stations' timestamp at described station symbol to be identified, according to described station symbol region and standard unit target matching rate
Determine the area belonging to described station symbol to be identified；
Owing to the highest described matching rate or standard station symbol corresponding to second highest matching rate belong to local station symbol, now, treat described in
Identify that station symbol belongs to local broadcasting stations' target probability the highest, it will usually be respectively provided with higher with the standard station symbol in multiple identical areas
Matching rate, in the case, can determine the area belonging to described station symbol to be identified according to these matching rates.
In present embodiment, according to the standard station symbol quantity of each department and the standard of each department in the standard station symbol obtained
Station symbol sum determines the area belonging to described station symbol to be identified, can be by following standard:
The difference obtaining matching rate and the highest matching rate (includes the highest matching rate pair less than the standard station symbol of the second preset difference value
The standard station symbol answered), determine the objective area belonging to most standard station symbol according to the area belonging to each standard station symbol obtained, sentence
The standard unit target sum that in the disconnected each standard station symbol obtained, the standard station symbol quantity of target area is corresponding with described target area
Between ratio whether exceed preset ratio, the most then using described target area as the area belonging to described station symbol to be identified.
Such as: matching rate is designated as 10, and 10 with the difference of the highest matching rate less than the standard unit of the second preset difference value
In standard station symbol, 8 belong to BTV, and 1 belongs to Sichuan television station, and 1 belongs to Xinjiang television station, if BTV
One has 12, and Sichuan television station one has 6, and Xinjiang television station one has 5, and preset ratio is 50%, it may be determined that target
Region is Beijing, the standard unit target that in each standard station symbol of acquisition, Pekinese's standard station symbol quantity (i.e. 8) is corresponding with Beijing
Ratio between sum (i.e. 12) is 72.7%, has exceeded preset ratio (50%), now, using Beijing as described to be identified
Area belonging to station symbol.
S305: extract the character area in described station symbol region；
As a example by described station symbol to be identified is for BTV's station symbol, character area and the mark in BTV
" BTV " is contour, so, the region of described station symbol region height can be doubled as character area, also in cut label " BTV " right side
That is, the height of character area is identical with the height of mark " BTV ", a length of described station symbol region height of character area
Twice；
For improving recognition accuracy further, in present embodiment, described character area is carried out Ostu binary conversion treatment,
And use connected domain mode to remove noise jamming.
S306: isolate individual character part from described character area, and isolated individual character part is identified；
It will be appreciated that for some local station symbol, such as BTV's station symbol, its difference is word, such as " Beijing
Satellite TV ", " physical culture ", " life " etc..This difference uniquely can be described by character area first word, in order to improve identification effect
Rate, in present embodiment, isolated individual character part is the individual character part at first word place in described character area, can
Realize separating by below scheme:
(1) interference straight line (section) horizontal, longitudinal in character area is removed；
(2) pixel separation between individual character part in described character area is obtained；
(3) complete individual character part is extracted according to described pixel separation, such as: literary composition (literature and art), shadow (video display)；
(4) individual character part is carried out Threshold segmentation, cutting procedure takes upper left corner 5*5 subdomain, calculate black ratio r, as
Really r is less than 0.5, the monochrome pixels of exchange individual character part.
Certainly, also whole individual character parts can all be separated, then be identified respectively, present embodiment is to this most in addition
Limit.
S307: the standard station symbol that individual character recognition result is corresponding with described area is carried out characters matching, by characters matching rate
The highest standard unit is denoted as recognition result.
Step S307 is identical with step S105 of the embodiment shown in Fig. 1, does not repeats them here.
Through Monte Carlo experiment, by the TV station symbol recognition result of the method for present embodiment channel common to BTV such as table 1 institute
Showing, table 1 gives the recognition time of each station symbol, all at about 2s.Watch 1 demonstrates the high accuracy of the method for present embodiment, fast
Speed and stability.
The recognition result of the station symbol of table 1 BTV channel
| BTV channel | Correct recognition rata | Recognition failures rate | Recognition time (s) |
| BTV (BTV1) | 100% | 0% | 1.76 |
| Beijing literature and art (BTV2) | 98.5% | 1.5% | 2.07 |
| Beijing science and education (BTV3) | 98.8% | 1.2% | 1.83 |
| Beijing video display (BTV4) | 98.6% | 1.4% | 2.03 |
| Beijing finance and economics (BTV5) | 99.0% | 1.0% | 1.72 |
| Beijing physical culture (BTV6) | 99.2% | 0.8% | 1.99 |
| Beijing life (BTV7) | 99.4% | 0.6% | 1.86 |
| Beijing is young (BTV8) | 99.6% | 0.4% | 1.85 |
| Beijing news (BTV9) | 97.9% | 2.1% | 1.93 |
| Beijing card is cruel (BTV10) | 97.7% | 2.3% | 1.82 |
| Beijing record (BTV11) | 98.3% | 1.7% | 2.01 |
For embodiment of the method, in order to be briefly described, therefore it is all expressed as a series of combination of actions, but this area
Technical staff should know, the embodiment of the present invention is not limited by described sequence of movement, because implementing according to the present invention
Example, some step can use other orders or carry out simultaneously.Secondly, those skilled in the art also should know, description
Described in embodiment belong to preferred embodiment, necessary to the involved action not necessarily embodiment of the present invention.
Fig. 4 is the structured flowchart of the local TV station symbol recognition device of one embodiment of the present invention；With reference to Fig. 4, described device
Including:
Area acquisition unit 401, for obtaining the station symbol region including station symbol to be identified；
Area determines unit 402, for being local broadcasting stations' timestamp at described station symbol to be identified, according to described station symbol region and mark
The matching rate of quasi-station symbol determines the area belonging to described station symbol to be identified；
Word Input unit 403, for extracting the character area in described station symbol region；
Separate recognition unit 404, for isolating individual character part from described character area, and to isolated individual character portion
Divide and be identified；
Result acquiring unit 405, for carrying out word by the standard station symbol that individual character recognition result is corresponding with described area
Join, standard unit the highest for characters matching rate is denoted as recognition result.
In a kind of alternative embodiment of the present invention, described separation recognition unit, it is further used for described individual character part
Carry out micronization processes；According to described in the refinement edge calculations of the individual character part after micronization processes between individual character part and standard one-letter
Edge matching rate；Judge that the matching rate difference of high rim matching rate and second highest edge matching rate is default the poorest higher than first
Value；Described matching rate difference higher than the first preset difference value time, using the described standard one-letter that high rim matching rate is corresponding as
The individual character recognition result in described individual character region；When described matching rate difference is not higher than the first preset difference value, according to micronization processes
After the provincial characteristics of individual character part and standard one-letter mate, described edge matching rate is updated；By high rim
Standard one-letter corresponding to matching rate is as the individual character recognition result in described individual character region.
In a kind of alternative embodiment of the present invention, described area acquisition unit, it is further used for from including to be identified
Obtain video frame images sequence in the predeterminable area of target video, each video frame images is carried out edge extracting, by each frame of video
The edge of image synthesizes, and obtains the minimum external matrix at the edge of synthesis, according to described minimum external matrix respectively to respectively
Video frame images is split, and is synthesized by average weighted mode by the image being partitioned into, and includes waiting to know to obtain
The station symbol region of other station symbol.
In a kind of alternative embodiment of the present invention, described device also includes:
Matching rate judging unit, for calculating the edge in described station symbol region and each standard unit target matching rate, the highest
When the difference of matching rate and second highest matching rate is less than or equal to the second preset difference value, it is judged that the highest described matching rate or second highest matching rate pair
Whether the standard station symbol answered belongs to local station symbol, if the highest described matching rate or standard station symbol corresponding to second highest matching rate belong to ground
Fang Taibiao, then assert that described station symbol to be identified is for local station symbol.
In a kind of alternative embodiment of the present invention, described device also includes:
According to length-width ratio, gray scale and the color in station symbol region, station symbol taxon, for judging that treating in station symbol region is known
Whether other station symbol is CCTV's station symbol, when described station symbol to be identified is non-CCTV's station symbol, calls described matching rate computing unit.
For device embodiment, due to itself and embodiment of the method basic simlarity, so describe is fairly simple, relevant
Part sees the part of embodiment of the method and illustrates.
Device embodiment described above is only schematically, and the wherein said unit illustrated as separating component can
To be or to may not be physically separate, the parts shown as unit can be or may not be physics list
Unit, i.e. may be located at a place, or can also be distributed on multiple NE.Can be selected it according to the actual needs
In some or all of module realize the purpose of the present embodiment scheme.Those of ordinary skill in the art are not paying creativeness
Work in the case of, be i.e. appreciated that and implement.
Through the above description of the embodiments, those skilled in the art it can be understood that to each embodiment can
The mode adding required general hardware platform by software realizes, naturally it is also possible to pass through hardware.Based on such understanding, on
State the part that prior art contributes by technical scheme the most in other words to embody with the form of software product, should
Computer software product can store in a computer-readable storage medium, such as ROM/RAM, magnetic disc, CD etc., including some fingers
Make with so that a computer equipment (can be personal computer, server, or the network equipment etc.) performs each and implements
The method described in some part of example or embodiment.
Last it is noted that above example is only in order to illustrate technical scheme, it is not intended to limit；Although
With reference to previous embodiment, the present invention is described in detail, it will be understood by those within the art that: it still may be used
So that the technical scheme described in foregoing embodiments to be modified, or wherein portion of techniques feature is carried out equivalent；
And these amendment or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit and
Scope.
Claims (10)
1. a local TV station symbol recognition method, it is characterised in that described method includes:
Obtain the station symbol region including station symbol to be identified；
It is local broadcasting stations' timestamp at described station symbol to be identified, determines according to described station symbol region and standard unit target matching rate described
Area belonging to station symbol to be identified；
Extract the character area in described station symbol region；
From described character area, isolate individual character part, and isolated individual character part is identified；
The standard station symbol that individual character recognition result is corresponding with described area is carried out characters matching, by standard the highest for characters matching rate
Station symbol is as recognition result.
2. the method for claim 1, it is characterised in that described isolated individual character part is identified, further
Including:
Described individual character part is carried out micronization processes；
According to edge between individual character part and standard one-letter described in the refinement edge calculations of the individual character part after micronization processes
Join rate；
Judge that whether the matching rate difference of high rim matching rate and second highest edge matching rate is higher than the first preset difference value；
Described matching rate difference higher than the first preset difference value time, using the described standard one-letter that high rim matching rate is corresponding as
The individual character recognition result in described individual character region；
When described matching rate difference is not higher than the first preset difference value, according to the provincial characteristics of the individual character part after micronization processes and
Standard one-letter mates, and is updated described edge matching rate；
Using standard one-letter corresponding for high rim matching rate as the individual character recognition result in described individual character region.
3. the method as according to any one of claim 1～2, it is characterised in that described acquisition includes the platform of station symbol to be identified
Mark region, farther includes:
In the predeterminable area of the video including station symbol to be identified, obtain video frame images sequence, each video frame images is carried out limit
Edge extracts, and is synthesized at the edge of each video frame images, obtains the minimum external matrix at the edge of synthesis, according to described minimum
Each video frame images is split by external matrix respectively, and is closed by average weighted mode by the image being partitioned into
Become, to obtain the station symbol region including station symbol to be identified.
4. the method as according to any one of claim 1～2, it is characterised in that described is local at described station symbol to be identified
During station symbol, before determining the area belonging to described station symbol to be identified according to described station symbol region and standard unit target matching rate,
Described method also includes:
Calculate the edge in described station symbol region and each standard unit target matching rate, little with the difference of second highest matching rate at the highest matching rate
When equal to the second preset difference value, it is judged that whether the highest described matching rate or standard station symbol corresponding to second highest matching rate belong to local
Station symbol, if the highest described matching rate or standard station symbol corresponding to second highest matching rate belong to local station symbol, then assert described to be identified
Station symbol is local station symbol.
5. method as claimed in claim 4, it is characterised in that the edge in described calculating described station symbol region and each standard station symbol
Matching rate before, described method also includes:
Length-width ratio, gray scale and color according to station symbol region judges whether the station symbol to be identified in station symbol region is CCTV's station symbol,
When described station symbol to be identified is non-CCTV's station symbol, the edge performing described calculation described station symbol region mates with each standard unit target
The step of rate.
6. a local TV station symbol recognition device, it is characterised in that described device includes:
Area acquisition unit, for obtaining the station symbol region including station symbol to be identified；
Area determines unit, for being local broadcasting stations' timestamp at described station symbol to be identified, according to described station symbol region and standard station symbol
Matching rate determine the area belonging to described station symbol to be identified；
Word Input unit, for extracting the character area in described station symbol region；
Separate recognition unit, for isolating individual character part from described character area, and isolated individual character part is carried out
Identify；
Result acquiring unit, for carrying out characters matching, by literary composition by the standard station symbol that individual character recognition result is corresponding with described area
The standard unit that word matching rate is the highest is denoted as recognition result.
7. device as claimed in claim 6, it is characterised in that described separation recognition unit, is further used for described individual character
Part carries out micronization processes；According to individual character part and standard one-letter described in the refinement edge calculations of the individual character part after micronization processes
Between edge matching rate；Judge that the matching rate difference of high rim matching rate and second highest edge matching rate is the most pre-higher than first
If difference；When described matching rate difference is higher than the first preset difference value, by the described standard one-letter that high rim matching rate is corresponding
Individual character recognition result as described individual character region；When described matching rate difference is not higher than the first preset difference value, according to refinement
Provincial characteristics and the standard one-letter of the individual character part after process mate, and are updated described edge matching rate；To be the highest
Standard one-letter corresponding to edge matching rate is as the individual character recognition result in described individual character region.
8. the device as according to any one of claim 6～7, it is characterised in that described area acquisition unit, is further used for
In the predeterminable area of the video including station symbol to be identified, obtain video frame images sequence, each video frame images is carried out edge and carries
Take, the edge of each video frame images is synthesized, obtain the minimum external matrix at the edge of synthesis, external according to described minimum
Each video frame images is split by matrix respectively, and is synthesized by average weighted mode by the image being partitioned into, with
Obtain the station symbol region including station symbol to be identified.
9. the device as according to any one of claim 6～7, it is characterised in that described device also includes:
Matching rate judging unit, for calculating the edge in described station symbol region and each standard unit target matching rate, in the highest coupling
When the difference of rate and second highest matching rate is less than or equal to the second preset difference value, it is judged that the highest described matching rate or second highest matching rate are corresponding
Whether standard station symbol belongs to local station symbol, if the highest described matching rate or standard station symbol corresponding to second highest matching rate belong to local broadcasting stations
Mark, then assert that described station symbol to be identified is for local station symbol.
10. device as claimed in claim 9, it is characterised in that described device also includes:
Station symbol taxon, for judging to be identified in station symbol region according to length-width ratio, gray scale and the color in station symbol region
Whether mark is CCTV's station symbol, when described station symbol to be identified is non-CCTV's station symbol, calls described matching rate computing unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510824087.9A CN105868682A (en) | 2015-11-24 | 2015-11-24 | Local channel logo identification method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510824087.9A CN105868682A (en) | 2015-11-24 | 2015-11-24 | Local channel logo identification method and apparatus |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105868682A true CN105868682A (en) | 2016-08-17 |
Family
ID=56623813
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510824087.9A Pending CN105868682A (en) | 2015-11-24 | 2015-11-24 | Local channel logo identification method and apparatus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105868682A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546424A (en) * | 2008-03-24 | 2009-09-30 | 富士通株式会社 | Method and device for processing image and watermark detection system |
| CN103020650A (en) * | 2012-11-23 | 2013-04-03 | Tcl集团股份有限公司 | Station caption identifying method and device |
| CN103336954A (en) * | 2013-07-08 | 2013-10-02 | 北京捷成世纪科技股份有限公司 | Identification method and device of station caption in video |
| US20150139535A1 (en) * | 2013-11-18 | 2015-05-21 | Nant Holdings Ip, Llc | Silhouette-based object and texture alignment, systems and methods |
-
2015
- 2015-11-24 CN CN201510824087.9A patent/CN105868682A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546424A (en) * | 2008-03-24 | 2009-09-30 | 富士通株式会社 | Method and device for processing image and watermark detection system |
| CN103020650A (en) * | 2012-11-23 | 2013-04-03 | Tcl集团股份有限公司 | Station caption identifying method and device |
| CN103336954A (en) * | 2013-07-08 | 2013-10-02 | 北京捷成世纪科技股份有限公司 | Identification method and device of station caption in video |
| US20150139535A1 (en) * | 2013-11-18 | 2015-05-21 | Nant Holdings Ip, Llc | Silhouette-based object and texture alignment, systems and methods |
Non-Patent Citations (2)
| Title |
|---|
| MEERA SAMBHAJI SAWALKAR 等: "Logo Detection and Recognition from the Images as Well as Videos", 《INTERNATIONAL JOURNAL OF SCIENCE AND RESEARCH (IJSR)》 * |
| 胡星火 等: "基于改进Chamfer匹配的台标识别", 《计算机工程》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112818862B (en) | Face tampering detection method and system based on multi-source clues and mixed attention | |
| US10896349B2 (en) | Text detection method and apparatus, and storage medium | |
| CN106254933B (en) | Subtitle extraction method and device | |
| Fu et al. | A salt and pepper noise image denoising method based on the generative classification | |
| CN103020650B (en) | Station caption identifying method and device | |
| WO2017088479A1 (en) | Method of identifying digital on-screen graphic and device | |
| EP2587826A1 (en) | Extraction and association method and system for objects of interest in video | |
| CN100562074C (en) | A method for extracting video subtitles | |
| Wei et al. | A robust video text detection approach using SVM | |
| CN102208023A (en) | Method for recognizing and designing video captions based on edge information and distribution entropy | |
| CN104462381A (en) | Trademark image retrieval method | |
| US8300937B2 (en) | Image segmentation method and system based on region features of pixels | |
| CN113505670A (en) | Remote sensing image weak supervision building extraction method based on multi-scale CAM and super-pixels | |
| CN104036280A (en) | Video fingerprinting method based on region of interest and cluster combination | |
| CN102682298A (en) | Video fingerprint method based on graph modeling | |
| CN106056600A (en) | Contourlet transform-based image splicing detection method | |
| CN108229247A (en) | A kind of mobile vehicle detection method | |
| CN110287888A (en) | A kind of TV station symbol recognition method and system | |
| US20080310685A1 (en) | Methods and Systems for Refining Text Segmentation Results | |
| CN104268845A (en) | Self-adaptive double local reinforcement method of extreme-value temperature difference short wave infrared image | |
| JP7144384B2 (en) | Object detection device, method and program | |
| Zhang et al. | A novel approach for binarization of overlay text | |
| Arai et al. | Text extraction from TV commercial using blob extraction method | |
| CN105868681A (en) | CCTV channel logo identification method and apparatus | |
| Sreedevi et al. | Ngfica based digitization of historic inscription images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160817 |
|
| WD01 | Invention patent application deemed withdrawn after publication |