CN106162330A - A kind of method extracting caption area in video pictures - Google Patents
A kind of method extracting caption area in video pictures Download PDFInfo
- Publication number
- CN106162330A CN106162330A CN201610597947.4A CN201610597947A CN106162330A CN 106162330 A CN106162330 A CN 106162330A CN 201610597947 A CN201610597947 A CN 201610597947A CN 106162330 A CN106162330 A CN 106162330A
- Authority
- CN
- China
- Prior art keywords
- point
- seed
- corroded
- value
- corrosion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
- H04N21/44—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs
- H04N21/44008—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics in the video stream
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
- H04N21/435—Processing of additional data, e.g. decrypting of additional data, reconstructing software from modules extracted from the transport stream
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/47—End-user applications
- H04N21/488—Data services, e.g. news ticker
- H04N21/4884—Data services, e.g. news ticker for displaying subtitles
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/222—Studio circuitry; Studio devices; Studio equipment
- H04N5/262—Studio circuits, e.g. for mixing, switching-over, change of character of image, other special effects ; Cameras specially adapted for the electronic generation of special effects
- H04N5/278—Subtitling
Abstract
The invention discloses a kind of method extracting caption area in video pictures, it comprises the following steps: S1: remove noise treatment: use maximum gradient method to obtain gradient image data according to the texture difference of captions and background frame: wherein, subtitle parts has neat data, and background frame then there will be irregular noise data；Remove irregular noise data part；S2: use the mode locking corrosion to position caption area；S3: determine headline.The present invention is to improve the efficiency that news demolition processes, determine suitable title quickly to the news split, a kind of method extracting caption area from video pictures according to video caption characteristic simultaneously for convenience taxonomic revision, the fast search of news video provided.Wherein, caption area is convenient and reliability is high to use the method locking corrosion to determine.
Description
Technical field
The present invention relates to a kind of method extracting caption area in video pictures.
Background technology
For television station, the news material of this TV station has complete storage management system, it is easy to regard a news
A piece of news in Pin carries out follow-up use；But, the news video in remaining source is accomplished by carrying out demolition process, will news
Video split into one by one individually news so that follow-up use.Captions in news video, especially main title can letters
The bright main points embodying this then news in a capsule.If can the caption recognition in video pictures out, then just can be quickly for tearing open
Being divided into single news one by one and determine suitable title, this has the biggest for visual classification arrangement, fast search etc.
Benefit.
In order to improve the efficiency that news demolition processes, determine suitable title, simultaneously for side quickly to the news split
Just the one invented taxonomic revision, the fast search of news video extracts word according to video caption characteristic from video pictures
The method in curtain region.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, it is provided that a kind of in video pictures, extract caption area
Method.
It is an object of the invention to be achieved through the following technical solutions: a kind of in video pictures, extract caption area
Method, it comprises the following steps:
S1: remove noise treatment: use maximum gradient method to obtain gradient map according to the texture difference of captions and background frame
As data: wherein, subtitle parts has neat data, and background frame then there will be irregular noise data；Remove nothing
The noise data part of rule；
S2: location caption area, including following sub-step:
S21: cloth seed, under original state, is set to background the value in whole region, finds suitably according to gradient image data
Place plants seed, to ensure that seed is on word；
S22: use maximum value process to calculate the value waiting to judge the point i.e. G point of corrosion, including following sub-step:
S221: calculate original graph respectively in vertical, level, the gray scale in three directions of diagonal, the maximum of tetra-passages of R, G, B
Value；
S222: calculating the maximum in four values, described maximum is designated as the value of G point；
S23: lock corrosion, including following sub-step:
S231: the point of the four direction left and right, upper and lower of seed in traversal drawing of seeds successively, it is judged that whether this point is corroded and i.e. sentences
Whether this point disconnected is G point: if this point has been corroded, and continues to judge next point；If this point is not corroded, then continue next
Step；
S232: compare value and the size of first threshold nGsensitive set of this point, if the value of this point is more than threshold value
Then can not corrode this point, be this point and added one layer of lock；If the value of this point is less than threshold value, then carry out next step, carry out rotten to the left
Erosion；
S233:G point waits to judge the point of corrosion for the seed left side, has 0 to N number of point being corroded on the right of seed, to the left corruption
Erosion, including following sub-step:
S2331: judge the quantity of the point being corroded on the right of seed:
(1) when the point not being corroded on the right of seed, then G point can be corroded；
(2) when there being the point that 1 to N is corroded on the right of seed, G point and the gray scale of rightmost point, R, G, B four-way are calculated
Maximum, when this maximum less than set Second Threshold nSensitive, then G point can corrode, and otherwise can not be corroded
I.e. add one layer of lock to this point；
S2332: the point being corroded becomes seed, and returns step S2331 until cannot corrosion sites again；
S234: whether monitoring captions marginal area reaches Corrosion standards, if reaching, terminating corrosion, otherwise strengthening corrosion
Second Threshold nSensitive repeats above-mentioned steps；
S235: extract caption data according to Corrosion results, then remove noise data further according to character features, finally obtain one
Subtitle position region accurately；
S3: determine headline, including following sub-step:
S31: screen and mistake according to the continuity Characteristics caption area identical to continuous print in video of captions in video properties
Filter, and choose effect preferably as the picture identifying captions；
S32: captions picture is converted to word, is the headline of this then news.
A kind of method extracting caption area in video pictures also includes that a threshold value arranges sub-step S0, in beginning
Front respectively first threshold nGsensitive and Second Threshold nSensitive are configured.
Described N value is 4.
Described step S1 also includes following sub-step: according to the feature of news caption profile consistency from top to bottom to video pictures
Carry out horizontal striping identification, and carry out noise removal process according to the feature of news caption.
The invention has the beneficial effects as follows:
The present invention is to improve the efficiency that news demolition processes, and determines suitable title quickly to the news split, is simultaneously
The one that taxonomic revision, the fast search of news video are provided by convenience carries from video pictures according to video caption characteristic
The method taking caption area.Wherein, caption area is convenient and reliability is high to use the method locking corrosion to determine.
Accompanying drawing explanation
Fig. 1 is the inventive method flow chart；
Fig. 2 is that original graph is vertical, level, three direction schematic diagrams of diagonal；
Fig. 3 is for locking corrosion flow chart.
Detailed description of the invention
Technical scheme is described in further detail below in conjunction with the accompanying drawings:
First, news video has a characteristic that
1) captions and the texture difference of background: typically in order to make captions understand easily resolution, captions and background have significantly segmentation
Line, so the texture difference of captions and background is the biggest, can determine the profile of captions with this.
2) feature of news caption:
A) profile consistency from top to bottom: from profile, background frame is usually irregular pattern-information, on news caption is then
Under neat pattern.
B) profile transformation density is big: from profile, and background frame seldom has the irregular conversion of profile a lot, and news word
The profile transformation density of curtain is the biggest.
C) can be with cutting: because there being interval between word and word, in captions, each word is that shape is similar in terms of profile
Independent individuality, for caption area, can be with cutting；And background frame is difficult to appearance and can be cut into shape by profile
Similar individuality.
D) generally there are frame characteristic: news caption the most all can be placed on special captions base plate, has intrinsic colour
Continuous gradation or constant, but with up and down or the feature of left and right significant change, the program mark on side can be removed according to base plate feature
Know data.
3) characteristic of video:
A) seriality: for captions, continuous a period of time (is probably several milliseconds for news content captions and is also likely to be several
Second, but may be up to a few minutes a few minutes even ten for headline captions), video pictures change time, captions without
Change.It is to say, in one section of continuous print frame, captions do not change, there is seriality.
B) invariant position: captions are to edit on the regulation region of captions base plate, and the position of captions base plate is usually
Constant, so the position of caption area is also changeless, and the background colour of captions base plate is gradual change or constant.
Therefore, according to These characteristics, as it is shown in figure 1, a kind of method extracting caption area in video pictures, it includes
Following steps:
S1: removal noise treatment:
(1) maximum gradient method is used to obtain gradient image data according to the texture difference of captions and background frame: wherein, word
Curtain part has neat data, and background frame then there will be irregular noise data；Remove irregular noise data
Part；
(2) (striping is known according to the feature of news caption profile consistency from top to bottom, video pictures to be carried out horizontal striping identification
Other: picture is carried out the identification one by one of one fixed width), and carry out noise removal process according to the feature of news caption.
S2: location caption area,
A) according to the region that may have captions on the feature location frame out of news caption；
B) lemma is all to edit on captions base plate, is gradual change or constant according to captions base plate invariant position and background colour
May determine that captions base plate size, and again orient caption area；
C) method locking corrosion is used to further determine that caption area: to include following sub-step:
S21: cloth seed, under original state, is set to background (255) the value in whole region, finds according to gradient image data
Suitably place plants seed, to ensure that seed is on word；Wherein, the value of seed is 0；
S22: use maximum value process to calculate the value waiting to judge the point i.e. G point of corrosion, including following sub-step:
S221: calculate original graph respectively at vertical, level, the gray scale of three directions (as shown in Figure 2) of diagonal, R, G, B tetra-
The maximum of individual passage:
GGray scale[x]=Max(abs(a1-a2), abs (b1-b2), abs (c1-c2), abs (d1-d2))；
GR[x]=Max(abs(a1-a2), abs (b1-b2), abs (c1-c2), abs (d1-d2))；
GG[x]=Max(abs(a1-a2), abs (b1-b2), abs (c1-c2), abs (d1-d2))；
GB[x]=Max(abs(a1-a2), abs (b1-b2), abs (c1-c2), abs (d1-d2))；
S222: calculate the maximum in four values, described maximum is designated as the value of G point:
G [x]=Max (G gray scale [x], GR[x], GG[x], GB[x])
S23: as it is shown on figure 3, lock corrosion, including following sub-step:
S231: the point of the four direction left and right, upper and lower of seed in traversal drawing of seeds successively, it is judged that whether this point is corroded and i.e. sentences
Whether this point disconnected is G point: if the value of this point is 0, be corroded, and continues to judge next point；If the value of this point is not 0,
Then continue next step；
S232: compare value and the size of first threshold nGsensitive set of this point, if the value of this point is more than threshold value
Then can not corrode this point, be this point and added one layer of lock；If the value of this point is less than threshold value, then carry out next step, carry out rotten to the left
Erosion；
S233:G point waits to judge the point of corrosion for the seed left side, has 0 to 4 points being corroded, corrode to the left on the right of seed,
Including following sub-step:
S2331: judge the quantity of the point being corroded on the right of seed:
(1) when the point not being corroded on the right of seed, then G point can be corroded；
(2) when there being the point that 1 to 4 are corroded on the right of seed, G point and the gray scale of rightmost point, R, G, B four-way are calculated
Maximum, when this maximum less than set Second Threshold nSensitive, then G point can corrode, and otherwise can not be corroded
I.e. add one layer of lock to this point；
S2332: the point being corroded becomes seed, and returns step S2331 until cannot corrosion sites again；
S234: whether monitoring captions marginal area reaches Corrosion standards, if reaching, terminating corrosion, otherwise strengthening corrosion
Second Threshold nSensitive repeats above-mentioned steps；
S235: extract caption data according to Corrosion results, then remove noise data further according to character features, finally obtain one
Subtitle position region accurately；
S3: determine headline, including following sub-step:
S31: screen and mistake according to the continuity Characteristics caption area identical to continuous print in video of captions in video properties
Filter, and choose effect preferably as the picture identifying captions；
S32: captions picture is converted to word, is the headline of this then news.
Claims (4)
1. the method extracting caption area in video pictures, it is characterised in that: it comprises the following steps:
S1: remove noise treatment: use maximum gradient method to obtain gradient map according to the texture difference of captions and background frame
As data: wherein, subtitle parts has neat data, and background frame then there will be irregular noise data；Remove nothing
The noise data part of rule；
S2: location caption area, including following sub-step:
S21: cloth seed, under original state, is set to background the value in whole region, finds suitably according to gradient image data
Place plants seed, to ensure that seed is on word；
S22: use maximum value process to calculate the value waiting to judge the point i.e. G point of corrosion, including following sub-step:
S221: calculate original graph respectively in vertical, level, the gray scale in three directions of diagonal, the maximum of tetra-passages of R, G, B
Value；
S222: calculating the maximum in four values, described maximum is designated as the value of G point；
S23: lock corrosion, including following sub-step:
S231: the point of the four direction left and right, upper and lower of seed in traversal drawing of seeds successively, it is judged that whether this point is corroded and i.e. sentences
Whether this point disconnected is G point: if this point has been corroded, and continues to judge next point；If this point is not corroded, then continue next
Step；
S232: compare value and the size of first threshold nGsensitive set of this point, if the value of this point is more than threshold value
Then can not corrode this point, be this point and added one layer of lock；If the value of this point is less than threshold value, then carry out next step, carry out rotten to the left
Erosion；
S233:G point waits to judge the point of corrosion for the seed left side, has 0 to N number of point being corroded on the right of seed, to the left corruption
Erosion, including following sub-step:
S2331: judge the quantity of the point being corroded on the right of seed:
(1) when the point not being corroded on the right of seed, then G point can be corroded；
(2) when there being the point that 1 to N is corroded on the right of seed, G point and the gray scale of rightmost point, R, G, B four-way are calculated
Maximum, when this maximum less than set Second Threshold nSensitive, then G point can corrode, and otherwise can not be corroded
I.e. add one layer of lock to this point；
S2332: the point being corroded becomes seed, and returns step S2331 until cannot corrosion sites again；
S234: whether monitoring captions marginal area reaches Corrosion standards, if reaching, terminating corrosion, otherwise strengthening corrosion
Second Threshold nSensitive repeats above-mentioned steps；
S235: extract caption data according to Corrosion results, then remove noise data further according to character features, finally obtain one
Subtitle position region accurately；
S3: determine headline, including following sub-step:
S31: screen and mistake according to the continuity Characteristics caption area identical to continuous print in video of captions in video properties
Filter, and choose effect preferably as the picture identifying captions；
S32: captions picture is converted to word, is the headline of this then news.
A kind of method extracting caption area in video pictures the most according to claim 1, it is characterised in that: also include
One threshold value arranges sub-step S0, the most respectively to first threshold nGsensitive and Second Threshold nSensitive
It is configured.
A kind of method extracting caption area in video pictures the most according to claim 1, it is characterised in that: described
N value is 4.
A kind of method extracting caption area in video pictures the most according to claim 1, it is characterised in that: described
Step S1 also includes following sub-step: according to the feature of news caption profile consistency from top to bottom, video pictures is carried out horizontal band
Change and identify, and carry out noise removal process according to the feature of news caption.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610597947.4A CN106162330B (en) | 2016-07-27 | 2016-07-27 | A method of extracting caption area in video pictures |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610597947.4A CN106162330B (en) | 2016-07-27 | 2016-07-27 | A method of extracting caption area in video pictures |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106162330A true CN106162330A (en) | 2016-11-23 |
| CN106162330B CN106162330B (en) | 2019-04-12 |
Family
ID=58059925
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610597947.4A Active CN106162330B (en) | 2016-07-27 | 2016-07-27 | A method of extracting caption area in video pictures |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106162330B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108769776A (en) * | 2018-05-31 | 2018-11-06 | 北京奇艺世纪科技有限公司 | Main title detection method, device and electronic equipment |
| CN109348289A (en) * | 2018-11-15 | 2019-02-15 | 北京奇艺世纪科技有限公司 | The title extracting method and device of news program |
| CN111401368A (en) * | 2020-03-24 | 2020-07-10 | 武汉大学 | News video title extraction method based on deep learning |
| CN111931775A (en) * | 2020-09-28 | 2020-11-13 | 成都索贝数码科技股份有限公司 | Method, system, computer device and storage medium for automatically acquiring news headlines |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070019108A1 (en) * | 2005-07-25 | 2007-01-25 | Samsung Electronics Co., Ltd. | Broadcast receiving device for displaying closed caption data and method thereof |
| US20070222895A1 (en) * | 2006-03-24 | 2007-09-27 | Toshiba America Information Systems, Inc. | Subtitle detection apparatus, subtitle detection method and pull-down signal detection apparatus |
| CN101115151A (en) * | 2007-07-10 | 2008-01-30 | 北京大学 | Method for extracting video subtitling |
| CN101453575A (en) * | 2007-12-05 | 2009-06-10 | 中国科学院计算技术研究所 | Video subtitle information extracting method |
| CN102202224A (en) * | 2011-06-22 | 2011-09-28 | 清华大学 | Caption flutter-free method and apparatus used for plane video stereo transition |
-
2016
- 2016-07-27 CN CN201610597947.4A patent/CN106162330B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070019108A1 (en) * | 2005-07-25 | 2007-01-25 | Samsung Electronics Co., Ltd. | Broadcast receiving device for displaying closed caption data and method thereof |
| US20070222895A1 (en) * | 2006-03-24 | 2007-09-27 | Toshiba America Information Systems, Inc. | Subtitle detection apparatus, subtitle detection method and pull-down signal detection apparatus |
| CN101115151A (en) * | 2007-07-10 | 2008-01-30 | 北京大学 | Method for extracting video subtitling |
| CN101453575A (en) * | 2007-12-05 | 2009-06-10 | 中国科学院计算技术研究所 | Video subtitle information extracting method |
| CN102202224A (en) * | 2011-06-22 | 2011-09-28 | 清华大学 | Caption flutter-free method and apparatus used for plane video stereo transition |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108769776A (en) * | 2018-05-31 | 2018-11-06 | 北京奇艺世纪科技有限公司 | Main title detection method, device and electronic equipment |
| CN108769776B (en) * | 2018-05-31 | 2021-03-19 | 北京奇艺世纪科技有限公司 | Title subtitle detection method and device and electronic equipment |
| CN109348289A (en) * | 2018-11-15 | 2019-02-15 | 北京奇艺世纪科技有限公司 | The title extracting method and device of news program |
| CN111401368A (en) * | 2020-03-24 | 2020-07-10 | 武汉大学 | News video title extraction method based on deep learning |
| CN111931775A (en) * | 2020-09-28 | 2020-11-13 | 成都索贝数码科技股份有限公司 | Method, system, computer device and storage medium for automatically acquiring news headlines |
| CN111931775B (en) * | 2020-09-28 | 2021-01-08 | 成都索贝数码科技股份有限公司 | Method, system, computer device and storage medium for automatically acquiring news headlines |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106162330B (en) | 2019-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101615252B (en) | Method for extracting text information from adaptive images | |
| CN106162330A (en) | A kind of method extracting caption area in video pictures | |
| CN101533474B (en) | Character and image recognition system based on video image and method thereof | |
| CN102663382B (en) | Video image character recognition method based on submesh characteristic adaptive weighting | |
| CN105205488B (en) | Word area detection method based on Harris angle points and stroke width | |
| CN101102419B (en) | A method for caption area of positioning video | |
| CN108009591A (en) | A kind of contact network key component identification method based on deep learning | |
| CN106228157B (en) | Coloured image word paragraph segmentation and recognition methods based on image recognition technology | |
| CN109874313A (en) | Text line detection method and line of text detection device | |
| CN106228545A (en) | A kind of automatic division method of figure of buddha class Tangka's concordance damaged area | |
| CN106651872A (en) | Prewitt operator-based pavement crack recognition method and system | |
| CN102254159B (en) | Interpretation method for digital readout instrument | |
| CN106157323B (en) | A kind of insulator division and extracting method of dynamic division threshold value and block search combination | |
| CN105516802A (en) | Multi-feature fusion video news abstract extraction method | |
| CN103473551A (en) | Station logo recognition method and system based on SIFT operators | |
| CN103729856B (en) | A kind of Fabric Defects Inspection detection method utilizing S-transformation signal extraction | |
| CN103020618A (en) | Detection method and detection system for video image text | |
| CN104952066A (en) | Method for identifying phase signboards of power transmission lines on basis of HSV (hue, saturation and value) color spaces | |
| Wang et al. | A novel video caption detection approach using multi-frame integration | |
| CN103336961A (en) | Interactive natural scene text detection method | |
| CN104598907A (en) | Stroke width figure based method for extracting Chinese character data from image | |
| CN101470802B (en) | Object detection apparatus and method thereof | |
| CN109271988A (en) | A kind of subtitle extraction method based on image segmentation and dynamic threshold | |
| CN104463134A (en) | License plate detection method and system | |
| CN109766892A (en) | A kind of ray detection image tagged information character dividing method based on edge detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |