CN104869403A - Shot segmentation method based on X264 compressed video - Google Patents
Shot segmentation method based on X264 compressed video Download PDFInfo
- Publication number
- CN104869403A CN104869403A CN201510253354.1A CN201510253354A CN104869403A CN 104869403 A CN104869403 A CN 104869403A CN 201510253354 A CN201510253354 A CN 201510253354A CN 104869403 A CN104869403 A CN 104869403A
- Authority
- CN
- China
- Prior art keywords
- frame
- gop
- shot
- frames
- shot segmentation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
The invention relates to a shot segmentation method based on an X264 compressed video. The method comprises following steps: extracting frame numbers, frame types and intra-frame prediction macro-block proportions through a frame-by-frame decoding manner, wherein a group of pictures (GOP) of the X264 compressed video is employed as a unit; performing shot detection on I-frames of the GOP to be detected; determining whether the current I-frames generate an abrupt shot change according to the length of the GOP to be detected; taking the intra-frame prediction macro-block proportions of all P-frames of the GOP to be detected as sample data; determining whether shot segmentation is needed according to fluctuation degrees of the sample data; and performing shot segmentation based on the prediction macro-block proportions. By employing the method, the shot segmentation can be carried out on the compressed video whose GOP is unfixed in length and structure. When the shot segmentation occurs at the I-frames, in the prior art, the change of prediction macro-block proportions of adjacent frames of the I-frames are generally employed for detection, and the video is required to be constant is GOP length and structure. By employing the method, shot transition detection on the I-frames can be carried out only by the use of parameters of the I-frames.
Description
Technical field
The invention belongs to video lens segmentation field, be specifically related to a kind of shot segmentation method based on X264 compressed video.
Background technology
Along with the development of multimedia, Internet technology, on the one hand, people are home-confined just can enjoy the various service such as video request program, online viewing easily, more and more higher to the requirement of Video service.On the other hand, media industry brings people's information and amusement for a long time, and the storage of magnanimity also makes the importance of media asset management become increasingly conspicuous.Shot segmentation technique all has great importance for the solution of above two aspect problems.First, shot segmentation technique is also one of most important step based on the first step of Video content retrieval, and its accuracy will directly have influence on the effect of subsequent treatment.Secondly, in media article cataloguing, shot segmentation technique finds out Shot change point automatically according to content deltas, selects, substantially increase the automaticity of cataloging syytem for Category-makers '.
Camera lens is the elementary cell of composition video, is also the elementary cell of content-based Video processing simultaneously.Video data is made up of a series of image of continuous print in time.In the inside of a camera lens, between image, correlation is in terms of content stronger, and between image in different camera lens, correlation is in terms of content more weak.Shot segmentation is then detect the shot boundary in video, finds out Shot change point, and Video segmentation is become more easy-to-handle part.Under normal circumstances, according to the speed of camera lens conversion, sudden change conversion and gradual change conversion can be divided into.
Shot segmentation method can be divided into two large classes: the method based on uncompressed domain and the method based on compression domain.Shot segmentation method based on uncompressed domain needs first compressed video to be carried out decompress(ion), and then processes, and this greatly reduces the efficiency of shot segmentation.Compared with the method based on uncompressed domain, based on the method for compression domain without the need to complete decompress(ion), there is the advantages such as deal with data amount is few, speed is fast, real-time is high, feature extraction is convenient.
Based on the shot segmentation method of compression domain mainly for two class compressed videos: mpeg compressed video and H.264/AVC compressed video.Shot segmentation algorithm based on mpeg compressed video mainly comprises based on the method for DCT coefficient, the method based on DC figure and the method etc. based on motion vector.Compared with MPEG compression and coding standard, H.264/AVC create some differences, such as support the macroblock partition of different size and shape, support multiple reference frame prediction estimation etc.These new features, while bringing excellent compression performance and network applicability, also make these shot segmentation algorithms based on mpeg compressed video above-mentioned no longer be applicable to H.264/AVC compressed video.The method etc. that shot segmentation algorithm based on H.264/AVC compressed video mainly comprises the method based on macro block absolute quantity, the method based on macro block ratio and distributes based on macro block bit.Based on the method for the macro block absolute quantity predictive coding pattern of each macro block in extracting directly two field picture on compressed bit stream.Corresponding intraframe predictive coding macroblock number, forward predictive coded macroblock number and back forecast coded macroblocks number are added up respectively for dissimilar frame, is judged the existence of shot segmentation point by absolute quantity and the relation of the threshold value preset of predictive coding mode macro different between adjacent two frames.There is an obvious drawback in the method based on macro block absolute quantity, for the video of different resolution, great changes will take place in the absolute quantity meeting of its macro block.Method based on macro block ratio improves based on the one of macro block absolute quantity method, and it by the percentage of all kinds of for the employing of statistics predictive coding mode macro and the threshold value preset being compared, and then determines the existence of shot segmentation point.Owing to there occurs Shot change, infra-frame prediction macro block and the inter prediction macro block bit distribution condition in code stream is different.The method of distributing based on macro block bit asks ratio to obtain bit distribution condition bit number used for infra-frame prediction macroblock coding in one two field picture and its inter prediction macroblock coding bit number used, compare with the threshold value preset again, determine the existence of shot segmentation point.
For the shot segmentation of H.264/AVC compressed video, the method adopted based on macro block ratio more.Fig. 1 gives the schematic diagram of two kinds of shot segmentation situations.
(1) shot segmentation occurs in I frame or P frame (as shown in the c point in Fig. 1, d point)
Due to camera lens conversion, will there is larger difference with the picture material of B frame above in this I frame or P frame.Now, majority is adopted forward prediction mode by the macro block in B frame.If Shot change occurs in P frame, majority is adopted intra prediction mode by the macro block so in this P frame.
(2) shot segmentation occurs in B frame
If camera lens conversion occurs in first B frame (as Suo Shi a point in Fig. 1), due to camera lens conversion, will there is larger difference with the picture material of I frame above or P frame in this B frame.In like manner, a frame below of this B frame is similarly B frame, and these two B frames belong to same camera lens.So majority is adopted back forecast pattern by the macro block in these two B frames.
If camera lens conversion occurs in second B frame (as Suo Shi the b point in Fig. 1), the frame is so B frame equally.Due to camera lens conversion, will there is larger difference in the picture material of these two B frames.Majority is adopted forward prediction mode by the macro block therefore in first B frame, and majority is adopted back forecast pattern by the macro block in second B frame.
The structure of known video GOP (Group of Pictures, image sets) before the method requirement shot segmentation.If the structure of GOP changes during Video coding, then the method will be no longer applicable.In addition, if there is gradual shot, above-mentioned rule change is also not obvious.
Owing to being subject to the impact of high complexity, be H.264/AVC sometimes difficult to apply in real time.And X264 is as a kind of simple implementation H.264/AVC, gives up some new features H.264/AVC to reduce computation complexity, be optimized under the prerequisite significantly not reducing compression performance, to improve efficiency of algorithm.X264 is widely used in practice.
X264 divides GOP according to video content, as shown in Figure 2.Wherein keyint represents the maximum length of GOP, and min_keyint represents the minimum length of GOP, the Coding cost that visual_difference represents aspects such as considering scene difference and computation complexity and obtains, the threshold value of scenecut presentation code cost.First frame of video is every IDR frame, even if get final product coded frame, backward (along the direction that the DISPLAY ORDER time increases progressively) mobile min_keyint distance, calculates visual_difference to present frame.If visual_difference is greater than scenecut, then judge that video scene there occurs change, namely inserts an every IDR frame.If visual_difference is less than scenecut, then need the relation of more current displacement and keyint.If displacement reaches keyint, then the I frame that insertion one is common.If displacement is less than keyint, then moves a frame backward, continue the magnitude relationship comparing visual_difference and scenecut.In like manner, whenever insertion every IDR frame or I frame, then need to move min_keyint distance backward.
The length that the GOP of X264 compressed video does not fix and structure, conventional compression territory shot segmentation method is also inapplicable.In addition, X264 provides I frame, P frame and B frame three kinds of frame types.Find through great many of experiments, acting as of B frame improves compression performance to greatest extent, and lens mutation can not occur in B frame.Thus, X264 compressed video needs a kind of shot segmentation method of system.
Summary of the invention
The object of the invention is to, according to the coding principle of X264, by extracting the code stream information such as frame type, macro-block coding pattern of X264 compressed video, a kind of shot segmentation method based on compression domain being provided, the lens mutation in video and gradual shot are detected.On the basis ensureing gratifying recall ratio and precision ratio, realize shot segmentation fast.
Based on a shot segmentation method for X264 compressed video, comprise the following steps:
Step 1, in units of the GOP of X264 compressed video, frame number, frame type and infra-frame prediction macro block ratio are extracted in decoding frame by frame.Described frame type comprises I frame, P frame and B frame; Described infra-frame prediction macro block ratio refers to the ratio of all number of macroblocks in the macroblock number and this two field picture adopting intra prediction mode to encode in a two field picture.
Step 2, carries out Shot Detection to I frame, judges whether current I frame lens mutation occurs according to the length of GOP to be measured.
Step 3, with the infra-frame prediction macro block ratio of P frames all in described GOP to be measured for sample data, judges whether to need shot segmentation according to the degree of fluctuation of sample data.
Step 4, carries out shot segmentation based on predicted macroblock ratio.
Further, to judge whether current I frame the method for lens mutation occurs as follows for described step 2:
The sequence number difference of adjacent two I frames is asked to obtain the length of GOP to be measured.The maximum length keyint of described GOP length and GOP is compared.If described length is less than keyint, judge current I frame generation lens mutation.
Further, according to the degree of fluctuation of sample data, described step 3 judges whether that the method needing shot segmentation is as follows:
Obtain the standard deviation of described sample data.If described standard deviation is less than the threshold value of setting, illustrate that the degree of fluctuation of sample data is less, represent that the picture change in described GOP is less, there is not camera lens conversion in described GOP, do not need shot segmentation; Otherwise, illustrate that the degree of fluctuation of sample data is comparatively large, represent that the picture in described GOP changes greatly, there is camera lens conversion in described GOP, need to carry out shot segmentation.
Further, described step 4 method of carrying out shot segmentation based on predicted macroblock ratio is as follows:
The threshold value T2 of lens mutation threshold value T1 and gradual shot is set;
Frame in described GOP is divided into suspicion frame and normal frame: the P frame that infra-frame prediction macro block ratio is greater than T2 is suspicion frame; Infra-frame prediction macro block ratio is less than the P frame of T2 and B frame is normal frame.Continued presence 2 normal frame are allowed at most between every 2 suspicion frames.Travel through all frames in GOP to be measured, be divided into some fragments.Described fragment is divided into two classes: a class is not containing the fragment of normal frame, and another kind of is the fragment containing normal frame;
The length of described two class different fragments is arranged respectively to the threshold value Tw of detection of gradual transitions camera lens.Fragment length and corresponding threshold value Tw are compared, if fragment length is greater than Tw, then judges that this fragment is as a gradual shot; Otherwise, if the infra-frame prediction macro block ratio of P frame is greater than T1 in described fragment, then judge that this P frame there occurs lens mutation.
Further, the method to set up of the threshold value T2 of described lens mutation threshold value T1 and gradual shot is: if the desired value of sample data adds 0.5 be greater than 0.7, then T1 is set to 0.7; Otherwise, T1 is set to 0.6.The value that T2 equals T1 deducts 0.2.
Further, the method to set up of described threshold value Tw is: not containing the fragment of normal frame, Tw is set to 5 frames; Fragment containing normal frame, Tw is set to 10 frames.
Compared with prior art, the present invention has the following advantages:
When carrying out shot cut detection, the detection method for I frame and P frame that the present invention proposes is separate, is independent of each other to testing result each other.
The method of the invention carries out cycle detection in units of GOP, carries out statistical analysis before detecting to GOP correlated samples data, does not carry out shot segmentation, reduce operand to the less GOP of picture change.
The method of the invention can carry out shot segmentation to GOP length and all unfixed compressed video of structure.Especially when shot segmentation occurs in I frame, prior art generally utilizes the change of predicted macroblock ratio in I frame consecutive frame to detect, and this just requires that video has fixing GOP length and structure.The method of the invention, then not by the restriction of GOP length and structure, only utilizes the shot transition detection that the parameter of I frame self can realize I frame.Experiment shows, adopt the method for the invention to carry out shot segmentation to compressed video, have good recall ratio and precision ratio, recall level average and precision ratio can reach 96.1% and 98.6% respectively.
Accompanying drawing explanation
Fig. 1 is the principle schematic of carrying out shot segmentation in prior art based on macro block ratio;
Fig. 2 is method flow diagram I frame being carried out to shot segmentation;
Fig. 3 is the method flow diagram of embodiment.
Embodiment
Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.Following examples only for illustration of the present invention, but are not used for limiting the scope of the invention.
Based on a shot segmentation method for X264 compressed video, the sudden change conversion of camera lens in video and gradual change conversion are detected.Detection for abrupt shot is divided into two aspects: occur in the detection of I frame to sudden change and sudden change occurred in the detection of P frame.If sudden change occurs in I frame, the GOP according to X264 divides flow process, is detected by the length comparing GOP; If sudden change occurs in P frame, then the infra-frame prediction macro block ratio P_intra investigating P frame is needed whether to exceed the threshold value of lens mutation.The P_intra investigating P frame is then needed whether to exceed the threshold value of gradual shot to the detection of gradual shot.The present invention's extraction code stream information in units of GOP carries out cycle detection.
As shown in Figure 3, in concrete enforcement, following steps are completed successively in a computer:
The first step: using FFMPEG as experiment porch, input X264 compressed video is as cycle tests.The memory space vector of initialization code stream information and maximum GOP length keyint.
Second step: extract the information such as frame number, frame type and macro-block skip mode and calculate the macro block ratio P_intra adopting infra-frame prediction in the process that a GOP decodes frame by frame.These code stream information are stored in vector.
3rd step: when decoding to next I frame, represents that a GOP terminates.Suspend decoding.Shot segmentation is carried out to previous GOP.Step is as follows:
1) the I frame number of GOP to be measured and the I frame number of previous GOP are done difference, thus obtain the length of previous GOP.This GOP length and keyint are compared.If GOP length is less than keyint, then can judge that the I frame of GOP to be measured there occurs lens mutation; Otherwise, the value of keyint is upgraded by this GOP length.
2) in GOP to be measured the P_intra of all P frames as sample data, calculation expectation and standard deviation.If standard deviation is less than 0.1, represent that sample data fluctuation is less, the change of this GOP image content is less, judges to occur without camera lens conversion, skips to the 4th step.If standard deviation is more than or equal to 0.1, represent that sample data fluctuation is comparatively large, this GOP image content changes greatly, and there is camera lens conversion, needs to detect.
3) each threshold value of shot segmentation is set.For the threshold value T1 of lens mutation, adopt the method process of segmentation.The desired value of sample data is added 0.5 imparting T1.If T1 is greater than 0.7, then T1 is set to 0.7; If T1 is less than 0.7, then T1 is set to 0.6.For the threshold value T2 of gradual shot, directly the value of T1 is deducted 0.2 and obtain.
4) frame in GOP to be measured is divided into two classes: a class is suspicion frame, namely P_intra is greater than the P frame of T2; Another kind of is normal frame, and namely P_intra is less than P frame and the B frame of T2.Specify to allow continued presence 2 normal frame at most between every two suspicion frames.Travel through all frames in GOP to be measured, be divided into some fragments be made up of continuous suspicion frame, and calculate fragment length.
5) fragment length and corresponding threshold value Tw are compared, if fragment length is greater than corresponding threshold value Tw, then judge that this fragment is as a gradual shot.Here need fragment to be divided into two classes: a class is that its length threshold Tw is set to 5 frames not containing the fragment of normal frame; Another kind of is the fragment containing normal frame, and its length threshold Tw is set to 10.
6) length is less than to the fragment of respective threshold, then needs the detection carrying out lens mutation.If wherein the P_intra of P frame is greater than T1, then judge that this P frame there occurs lens mutation.
4th step: empty the variablees such as vector, T1, T2, skip to second step, until all frames having detected video to be measured.
In order to check the validity of the method for the invention, the X264 compressed video of 4 groups of different contents is adopted to test.Usual use recall ratio (Recall) and precision ratio (Precision) this two indices evaluate the effect of shot segmentation algorithm.Test result is as shown in table 1.As shown in Table 1, recall ratio and the precision ratio of the 1st cycle tests all reach 100%, and the recall level average of 4 groups of sequences and precision ratio are respectively 96.1% and 98.6%.
The recall ratio of table 1 cycle tests and precision ratio
| Cycle tests | Recall ratio | Precision ratio |
| 1 | 100％ | 100％ |
| 2 | 98.1％ | 96.2％ |
| 3 | 93.2％ | 98.2％ |
| 4 | 93.2％ | 100％ |
| Average | 96.1％ | 98.6％ |
The present embodiment shows: the method that the present invention proposes can carry out effective shot segmentation to X264 compressed video, can obtain good recall ratio and precision ratio.Based on compression domain shot segmentation algorithm operationally between on also possess some superiority compared with the algorithm based on uncompressed domain.
Claims (6)
1., based on a shot segmentation method for X264 compressed video, it is characterized in that comprising the following steps:
Step 1, in units of the image sets GOP of X264 compressed video, frame number, frame type and infra-frame prediction macro block ratio are extracted in decoding frame by frame;
Described frame type comprises I frame, P frame and B frame; Described infra-frame prediction macro block ratio refers to the ratio of all macroblock number in the macroblock number and this two field picture adopting intra prediction mode to encode in a two field picture;
Step 2, carries out Shot Detection to the I frame in GOP to be measured, judges whether current I frame lens mutation occurs according to the length of GOP to be measured;
Step 3, with the infra-frame prediction macro block ratio P_intra of P frames all in described GOP to be measured for sample data, judges whether to need shot segmentation according to the degree of fluctuation of sample data;
Step 4, carries out shot segmentation based on predicted macroblock ratio.
2. the shot segmentation method based on X264 compressed video according to claim 1, is characterized in that, it is as follows that described step 2 judges whether current I frame the method for lens mutation occurs:
The sequence number difference of adjacent two I frames is asked to obtain the length of GOP to be measured; The maximum length keyint of described GOP length and GOP is compared; If described length is less than keyint, judge current I frame generation lens mutation.
3. the shot segmentation method based on X264 compressed video according to claim 1, is characterized in that, according to the degree of fluctuation of sample data, described step 3 judges whether that the method needing shot segmentation is as follows:
Obtain the standard deviation of described sample data; If described standard deviation is less than the threshold value of setting, illustrate that the degree of fluctuation of sample data is less, represent that the picture change in described GOP is less, there is not camera lens conversion in described GOP, do not need shot segmentation; Otherwise, illustrate that the degree of fluctuation of sample data is comparatively large, represent that the picture in described GOP changes greatly, there is camera lens conversion in described GOP, need to carry out shot segmentation.
4. according to the shot segmentation method based on X264 compressed video in claims 1 to 3 described in any one, it is characterized in that, the method that described step 4 carries out shot segmentation based on predicted macroblock ratio is as follows:
The lens mutation threshold value T1 of shot segmentation and the threshold value T2 of gradual shot is set;
Frame in described GOP is divided into suspicion frame and normal frame: the P frame that P_intra is greater than T2 is suspicion frame; P_intra is less than the P frame of T2 and B frame is normal frame; Continued presence 2 normal frame are allowed at most between every 2 suspicion frames; Travel through all frames in GOP to be measured, be divided into some fragments; Described fragment is divided into two classes: a class is not containing the fragment of normal frame, and another kind of is the fragment containing normal frame;
The length of described two class different fragments is arranged respectively to the threshold value Tw of detection of gradual transitions camera lens; Fragment length and corresponding threshold value Tw are compared, if fragment length is greater than corresponding threshold value Tw, then judges that this fragment is as a gradual shot; Otherwise, if the P_intra of P frame is greater than T1 in described fragment, then judge that this P frame there occurs lens mutation.
5. the shot segmentation method based on X264 compressed video according to claim 4, it is characterized in that, the method to set up of the threshold value T2 of described lens mutation threshold value T1 and gradual shot is: if the desired value of sample data adds 0.5 be greater than 0.7, then T1 is set to 0.7; Otherwise, T1 is set to 0.6; The value that T2 equals T1 deducts 0.2.
6. the shot segmentation method based on X264 compressed video according to claim 4, is characterized in that, the method to set up of described threshold value Tw is: not containing the fragment of normal frame, threshold value Tw is set to 5 frames; Fragment containing normal frame, threshold value Tw is set to 10 frames.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510253354.1A CN104869403B (en) | 2015-05-18 | 2015-05-18 | A kind of shot segmentation method based on X264 compression videos |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510253354.1A CN104869403B (en) | 2015-05-18 | 2015-05-18 | A kind of shot segmentation method based on X264 compression videos |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104869403A true CN104869403A (en) | 2015-08-26 |
| CN104869403B CN104869403B (en) | 2018-06-12 |
Family
ID=53914862
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510253354.1A Expired - Fee Related CN104869403B (en) | 2015-05-18 | 2015-05-18 | A kind of shot segmentation method based on X264 compression videos |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104869403B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106231348A (en) * | 2016-09-19 | 2016-12-14 | 浙江宇视科技有限公司 | Back method, device and the system of a kind of GOP data |
| CN107437076A (en) * | 2017-08-02 | 2017-12-05 | 陈雷 | The method and system that scape based on video analysis does not divide |
| CN109561310A (en) * | 2017-09-26 | 2019-04-02 | 腾讯科技（深圳）有限公司 | Video coding processing method, device, equipment and storage medium |
| CN110119735A (en) * | 2018-02-06 | 2019-08-13 | 上海全土豆文化传播有限公司 | The character detecting method and device of video |
| CN110677657A (en) * | 2019-11-01 | 2020-01-10 | 杭州当虹科技股份有限公司 | Scene segmentation method for content adaptive coding |
| CN111327945A (en) * | 2018-12-14 | 2020-06-23 | 北京沃东天骏信息技术有限公司 | Method and apparatus for segmenting video |
| CN111966859A (en) * | 2020-08-27 | 2020-11-20 | 司马大大（北京）智能系统有限公司 | Video data processing method and device and readable storage medium |
| CN112714336A (en) * | 2020-12-01 | 2021-04-27 | 万兴科技集团股份有限公司 | Video segmentation method and device, electronic equipment and computer readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101184226A (en) * | 2006-11-14 | 2008-05-21 | 腾讯科技（深圳）有限公司 | Method and device of inserting video key frame |
| CN102630013A (en) * | 2012-04-01 | 2012-08-08 | 北京捷成世纪科技股份有限公司 | Bit rate control video compression method and device on basis of scene switching |
| CN103826121A (en) * | 2013-12-20 | 2014-05-28 | 电子科技大学 | Scene switching detection based code rate control method in low-delay video coding |
-
2015
- 2015-05-18 CN CN201510253354.1A patent/CN104869403B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101184226A (en) * | 2006-11-14 | 2008-05-21 | 腾讯科技（深圳）有限公司 | Method and device of inserting video key frame |
| CN102630013A (en) * | 2012-04-01 | 2012-08-08 | 北京捷成世纪科技股份有限公司 | Bit rate control video compression method and device on basis of scene switching |
| CN103826121A (en) * | 2013-12-20 | 2014-05-28 | 电子科技大学 | Scene switching detection based code rate control method in low-delay video coding |
Non-Patent Citations (2)
| Title |
|---|
| 朱耀麟等: "《视频检索常用的镜头分割方法的研究》", 《视频应用与工程》 * |
| 洪夏俊等: "《基于H.264/AVC压缩域的实时视频镜头分割算法》", 《电脑知识与技术》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106231348B (en) * | 2016-09-19 | 2019-06-11 | 浙江宇视科技有限公司 | A kind of back method of GOP data, device and system |
| CN106231348A (en) * | 2016-09-19 | 2016-12-14 | 浙江宇视科技有限公司 | Back method, device and the system of a kind of GOP data |
| CN107437076A (en) * | 2017-08-02 | 2017-12-05 | 陈雷 | The method and system that scape based on video analysis does not divide |
| CN107437076B (en) * | 2017-08-02 | 2019-08-20 | 逄泽沐风 | The method and system that scape based on video analysis does not divide |
| WO2019062226A1 (en) * | 2017-09-26 | 2019-04-04 | 腾讯科技（深圳）有限公司 | Video encoding method, computer device, and storage medium |
| CN109561310A (en) * | 2017-09-26 | 2019-04-02 | 腾讯科技（深圳）有限公司 | Video coding processing method, device, equipment and storage medium |
| US10827182B2 (en) | 2017-09-26 | 2020-11-03 | Tencent Technology (Shenzhen) Company Limited | Video encoding processing method, computer device and storage medium |
| CN110119735A (en) * | 2018-02-06 | 2019-08-13 | 上海全土豆文化传播有限公司 | The character detecting method and device of video |
| CN111327945A (en) * | 2018-12-14 | 2020-06-23 | 北京沃东天骏信息技术有限公司 | Method and apparatus for segmenting video |
| CN111327945B (en) * | 2018-12-14 | 2021-03-30 | 北京沃东天骏信息技术有限公司 | Method and apparatus for segmenting video |
| US11275950B2 (en) | 2018-12-14 | 2022-03-15 | Beijing Wodong Tianjun Information Technology Co., Ltd. | Method and apparatus for segmenting video |
| CN110677657A (en) * | 2019-11-01 | 2020-01-10 | 杭州当虹科技股份有限公司 | Scene segmentation method for content adaptive coding |
| CN110677657B (en) * | 2019-11-01 | 2022-02-18 | 杭州当虹科技股份有限公司 | Scene segmentation method for content adaptive coding |
| CN111966859A (en) * | 2020-08-27 | 2020-11-20 | 司马大大（北京）智能系统有限公司 | Video data processing method and device and readable storage medium |
| CN112714336A (en) * | 2020-12-01 | 2021-04-27 | 万兴科技集团股份有限公司 | Video segmentation method and device, electronic equipment and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104869403B (en) | 2018-06-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104869403A (en) | Shot segmentation method based on X264 compressed video | |
| US10728564B2 (en) | Systems and methods of encoding multiple video streams for adaptive bitrate streaming | |
| CN111010495B (en) | Video denoising processing method and device | |
| CN104427345B (en) | Acquisition methods, acquisition device, Video Codec and its method of motion vector | |
| JP2000224590A (en) | Method for extracting characteristics of video sequence | |
| US9167260B2 (en) | Apparatus and method for video processing | |
| US9883200B2 (en) | Method of acquiring neighboring disparity vectors for multi-texture and multi-depth video | |
| CN109194955B (en) | Scene switching detection method and system | |
| EP3005689A1 (en) | Systems and methods of encoding multiple video streams with adaptive quantization for adaptive bitrate streaming | |
| US7302000B2 (en) | Method and system for two-pass video encoding using sliding windows | |
| CN101056408B (en) | Method and apparatus for determining a threshold for skip mode in video coding | |
| CN107360431A (en) | A kind of determination methods and device of frame type | |
| EP3175621B1 (en) | Video-segment identification systems and methods | |
| US10412391B1 (en) | Minimize number of encoded video stream frames for content recognition | |
| Ren et al. | Key frame extraction based on information entropy and edge matching rate | |
| KR20170090868A (en) | Scene cut frame detecting apparatus and method | |
| KR101777776B1 (en) | Apparatus and method for sampled profile based motion estimation | |
| CN112714336B (en) | Video segmentation method and device, electronic equipment and computer readable storage medium | |
| CN1713733A (en) | Rapid kinematics estimation of video frequency object based on MPEG-4 | |
| JP4154459B2 (en) | Digital video processing method and apparatus | |
| US20090296823A1 (en) | Video characterisation identification and search system | |
| US8265155B2 (en) | Method of block partition for H.264 inter prediction | |
| Hong et al. | Scene change detection using edge direction based on intra prediction mode in H. 264/AVC compression domain | |
| Guo et al. | Quick extracting keyframes from compressed video | |
| KR101787004B1 (en) | Apparatus and method for sampled pixel based motion estimation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180612 Termination date: 20190518 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |