CN110267040A - A kind of method for compressing image based on video flow detection - Google Patents
A kind of method for compressing image based on video flow detection Download PDFInfo
- Publication number
- CN110267040A CN110267040A CN201910572004.XA CN201910572004A CN110267040A CN 110267040 A CN110267040 A CN 110267040A CN 201910572004 A CN201910572004 A CN 201910572004A CN 110267040 A CN110267040 A CN 110267040A
- Authority
- CN
- China
- Prior art keywords
- picture
- compression
- frame
- detection
- image based
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/146—Data rate or code amount at the encoder output
- H04N19/149—Data rate or code amount at the encoder output by estimating the code amount by means of a model, e.g. mathematical model or statistical model
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/172—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a picture, frame or field
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
Abstract
The invention discloses a kind of method for compressing image based on video flow detection, comprising the following steps: video file parsing: video file is disassembled as the continuous frame picture of time shaft；The detection of frame picture: being detected to frame picture and marked the content of picture, is formed detection log and is marked image；Compression of images: frame picture is merged.The present invention simplifies the quantity of figure by the similarity between reduced time window adjacent image, rejects the similar adjacent image of high similarity to achieve the purpose that compression.
Description
Technical field
The present invention relates to a kind of method for compressing image based on video flow detection, belong to the deep learning and figure of artificial intelligence
Shape technical field of image processing.
Background technique
Compression of images includes lossy compression and lossless compression, expresses original picture element matrix with less BIT, therefore scheme
As compression is also referred to as image coding.General lossy compression has higher compression ratio than lossless compression, has become the pressure of mainstream
It draws back one's hand section.For conventional pattern, image compression rate is generally capable of up to 80% or so.
Ignore the content similarity between image and image since current compression of images excessively sticks to processes pixel, leads
Cause has the following problems: first is that compression ratio is not high enough, for the mass data of real service generation, using Image Compression come
Saving storage is almost an utterly inadequate amount, is difficult in find effects, excessive lossy compression often sacrifices the authenticity of pixel, therefore simply
It is no longer the way recommended that compression ratio is pursued on ground；Second is that compression is low with reduction efficiency, pushed in the requirement of scan picture
Contracting technology overhead is excessive.
Therefore, it is necessary to research and develop a kind of to compress the higher compress technique of more efficient and compression ratio than traditional pixel.
Summary of the invention
In view of the deficienciess of the prior art, the invention proposes a kind of method for compressing image based on video flow detection,
Its quantity that can simplify figure, achievees the purpose that compression, and it is higher to compress more efficient and compression ratio than traditional pixel.
The present invention solves its technical problem and adopts the technical scheme that:
A kind of method for compressing image based on video flow detection provided in an embodiment of the present invention, comprising the following steps:
Video file parsing: video file is disassembled as the continuous frame picture of time shaft；
The detection of frame picture: being detected to frame picture and marked the content of picture, is formed detection log and is marked image；
Compression of images: frame picture is merged.
It is combined as a kind of possible implementation of the present embodiment, in video file resolving, is split by N frame per second
Video file has stringent time sequencing and identical size between frame picture, and N is positive integer.
It is combined as a kind of possible implementation of the present embodiment, in frame picture detection process, using mature target
The content of detection model mark frame picture and the detection log for recording every frame picture.
It is combined as a kind of possible implementation of the present embodiment, the content of the detection log of picture includes picture file
Name, detection moment (yyyy-mm-dd hh:mm:ss), target position [x1, y1, x2, y2], target type, lasting frame number, frame are adopted
Time slot and detection confidence level.
A kind of possible implementation of the present embodiment is combined as to classify to frame picture in image compression process,
" core " picture of prototypical member as such for extracting every class frame picture, using " core " picture as the compressed object of every class.
It is combined as a kind of possible implementation of the present embodiment, in image compression process, more temporally adjacent two
Frame " core " picture merges if the detection log content height of two frames " core " picture is similar.It is highly similar to refer to two frames
The target type of " core " picture is identical, and confidence level is close.
It is combined as a kind of possible implementation of the present embodiment, in image compression process, the first retention time value is most
Big " core " picture, deletes its neighbour, and " core " picture of second pair of reservation modifies its log.
It is combined as a kind of possible implementation of the present embodiment, a kind of method for compressing image based on video flow detection is also
The following steps are included:
As a result it exports: exporting compressed image information.
Be combined as a kind of possible implementation of the present embodiment, the compressed image information include: " core " picture,
Detect log and compression ratio.
It is combined as a kind of possible implementation of the present embodiment, the compression ratio
What the technical solution of the embodiment of the present invention can have has the beneficial effect that:
The technical solution of the embodiment of the present invention carries out video file parsing first, then disassembles video file for time shaft
Continuous frame picture；Secondly it carries out the detection of frame picture: the content of picture is detected and marked to frame picture, form detection log
And mark image；Compression of images is carried out again: frame picture is merged.The technical solution of the embodiment of the present invention passes through the reduced time
Similarity between window adjacent image simplifies the quantity of figure, rejects the similar adjacent image of high similarity to reach compression
Purpose.Compared with prior art, the present invention has the advantage that
1) lossless compression does not change the Pixel Information of image, but rejects the image of duplicate contents after compression, only retain
" core " member.The quantity of image is not only reduced in this way, but also ensures the quality of image.
2) can be realized Real Time Compression in efficiency, the strategy of compression is only the similarity system design according to target type, and
And this comparison is based on time shaft, for only temporally adjacent member just it is necessary to compare, this strategy not only conforms with target
The business need of detection can more substantially reduce the number compared.
Detailed description of the invention:
Fig. 1 is a kind of process of method for compressing image based on video flow detection shown according to an exemplary embodiment
Figure；
Fig. 2 is the process of another method for compressing image based on video flow detection shown according to an exemplary embodiment
Figure.
Fig. 3 be the compression of images based on video flow detection one have applicating flow chart；
Fig. 4 is the schematic diagram of frame picture and mark in frame picture detection process；
Fig. 5 is the frame assembly diagram during frame picture classification
Fig. 6 is engine stack architecture schematic diagram；
Fig. 7 is the schematic diagram that 24 frame set in Fig. 5 are carried out with complete compression process；
Fig. 8 is the schematic diagram restored to compression image；
Fig. 9 is compression of images-reduction process schematic.
Specific embodiment
The present invention will be further described with embodiment with reference to the accompanying drawing:
In order to clarify the technical characteristics of the invention, below by specific embodiment, and its attached drawing is combined, to this hair
It is bright to be described in detail.Following disclosure provides many different embodiments or example is used to realize different knots of the invention
Structure.In order to simplify disclosure of the invention, hereinafter the component of specific examples and setting are described.In addition, the present invention can be with
Repeat reference numerals and/or letter in different examples.This repetition is that for purposes of simplicity and clarity, itself is not indicated
Relationship between various embodiments and/or setting is discussed.It should be noted that illustrated component is not necessarily to scale in the accompanying drawings
It draws.Present invention omits the descriptions to known assemblies and treatment technology and process to avoid the present invention is unnecessarily limiting.
Fig. 1 is a kind of process of method for compressing image based on video flow detection shown according to an exemplary embodiment
Figure.As described in Figure 1, a kind of method for compressing image based on video flow detection provided in an embodiment of the present invention, comprising the following steps:
Video file parsing: video file is disassembled as the continuous frame picture of time shaft；
The detection of frame picture: being detected to frame picture and marked the content of picture, is formed detection log and is marked image；
Compression of images: frame picture is merged.
It is combined as a kind of possible implementation of the present embodiment, in video file resolving, is split by N frame per second
Video file has stringent time sequencing and identical size between frame picture, and N is positive integer.
It is combined as a kind of possible implementation of the present embodiment, in frame picture detection process, using mature target
The content of detection model mark frame picture and the detection log for recording every frame picture.
It is combined as a kind of possible implementation of the present embodiment, the content of the detection log of picture includes picture file
Name, detection moment (yyyy-mm-dd hh:mm:ss), target position [x1, y1, x2, y2], target type, lasting frame number, frame are adopted
Time slot and detection confidence level.
A kind of possible implementation of the present embodiment is combined as to classify to frame picture in image compression process,
" core " picture of prototypical member as such for extracting every class frame picture, using " core " picture as the compressed object of every class.
It is combined as a kind of possible implementation of the present embodiment, in image compression process, more temporally adjacent two
Frame " core " picture merges if the detection log content height of two frames " core " picture is similar.It is highly similar to refer to two frames
The target type of " core " picture is identical, and confidence level is close.
It is combined as a kind of possible implementation of the present embodiment, in image compression process, the first retention time value is most
Big " core " picture, deletes its neighbour, and " core " picture of second pair of reservation modifies its log.
Fig. 2 is the process of another method for compressing image based on video flow detection shown according to an exemplary embodiment
Figure.As described in Figure 2, another method for compressing image based on video flow detection provided in an embodiment of the present invention, including following step
It is rapid:
Video file parsing: video file is disassembled as the continuous frame picture of time shaft；
The detection of frame picture: being detected to frame picture and marked the content of picture, is formed detection log and is marked image；
Compression of images: frame picture is merged；
As a result it exports: exporting compressed image information.
Be combined as a kind of possible implementation of the present embodiment, the compressed image information include: " core " picture,
Detect log and compression ratio.
It is combined as a kind of possible implementation of the present embodiment, it is described
Fig. 3 be the compression of images based on video flow detection one have applicating flow chart.As shown in Fig. 2, being based on video flowing
The compression of images process of detection is divided into four big steps, is video file parsing respectively；The detection of frame picture, mark, record log；Frame
Merge and modifies log；Detection output, including mark picture, log, compression ratio.
One, image compression process
The parsing of first step video.Video file is split by 25 frame per second, there is stringent time sequencing and phase between picture
Same size (resolution ratio).
Second step image detection.The content of picture is marked using mature target detection model and records the inspection of every picture
Survey log.The content of log include picture file name, detection moment (yyyy-mm-dd hh:mm:ss), target position [x1, y1,
X2, y2], target type, continue frame number, frame and adopt time slot, detection confidence level.For example, as shown in figure 4, Fig. 4 is as a frame figure
Piece is detected and marks out " dress " and " arm0_beishou " two targets, and generates detection log two:
[" beishoub.jpg ", " 2019/6/19 21:47:35 ", [218,57,624,471], " dress ", 1,100,
0.6000]；
[" beishoub.jpg ", " 2019/6/19 21:47:35 ", [218,37,624,465], " arm0_beishou ",
1,100,0.8725].
Third step compression of images.Two more temporally adjacent frame pictures, if two figures detection log content height is similar
Merge two frames.The highly similar target type for referring to two frames is identical, and confidence level is close.The movement of compression includes two: the
One retention time is worth maximum picture, deletes its neighbour, and the picture of second pair of reservation modifies its log.It is specific as follows: (1)
Continue frame number from increasing 1；(2) confidence level takes same target average；(3) it repeats, until all frames are disposed.
The output of 4th step result.The content of output includes: first is that " core " picture；Second is that detection log；Third is that compression ratio, pressure
The formula of contracting ratio are as follows:
Two, specific example
The present invention is used to realize the merging of similar frame member and the preservation of " core " member.Here pressure is specifically described for image
Compression process illustrates using for 24 frame figure shown in fig. 5.
As shown in fig. 6, the initial stack of engine stack architecture is set as 100 deeply.The stack detects whole works of compression after undertaking
Make.
Firstly, initial frame f000 stacking.Due to being initial frame, so without comparing.It need to be with f000 pairs when f001 stacking
Than judging similitude to determine to merge (compression) or stacking, target type that this example f000 is detected " dress ",
" arm1_baoxiong " } it is dissimilar with f001 target type { " arm0_beishou " }, therefore, the direct stacking of foo1.Similarly
The target type of f002 is { " dress ", " arm0_beishou " } and its neighbour f001 is dissimilar, therefore also can only stacking.
But when f003 ({ " dress ", " arm0_beishou " }) arrives, it is similar to foo2, it is therefore desirable to compress (f002 pops,
Foo3 stacking, and modify the detection log of f003).And so on, until last frame f024 operation terminate, engine it is worked
Journey is as shown in Figure 7.It can be seen that storehouse is used to save " core " member, squeeze operation is only carried out in stack top.Since stack is deeply limited, so working as
" core " member just answers batch signatures when taking storehouse.
By compression processing, compressed original image can not be directly obtained, but their summary figure (only retains in detection
The key messages such as target type, position coordinates, the confidence level of appearance), the conversion from summary figure to original graph is referred to as with mapping process
For compression recovery.As shown in figure 8,6 " core " member's summary figures are respectively mapped to their original graph, i.e. f000, f001,
f004,f010,f021,f023.Final step can be carried out after mapping --- rear compression output.It can be seen that the rear inspection that this patent proposes
Measured compressed technology will not lose the pixel of original image, theoretically should belong to lossless compression.Interior bulk density has only been simplified after compression
Multiple, redundancy useless frame picture, as shown in Figure 9.
Rationally setting stack can effectively reduce I/O number of figure and journal file deeply and achieve the purpose that improve throughput.
According to this example, the compression ratio of this final second compression is 6/24=0.25.
The invention has the following advantages that
(1) lossless compression does not change the Pixel Information of image, but rejects the image of duplicate contents after compression, only retain
" core " member.The quantity of image is not only reduced in this way, but also ensures the quality of image.In order to test compression effectiveness of the invention,
We select two groups of business hall monitoring port at random and acquire five video images, and each 1 minute every time.Business hall monitor video is adopted
With the MP4 format of standard, supports 25 frame broadcasting speed per second and resolution ratio is unified for 1024*768.For this group of video file into
Compression is detected after row, 10 groups of compression results is obtained, compression result is as shown in table 1 after 10 groups.
Compression effectiveness statistical form after table 1:
| Video acquisition port | Frame analysis diagram the piece number | Compressed picture number (opening) afterwards | Compression ratio | Remarks |
| 1 | 60 seconds * 25 frame=1500 | 132 | 8.80% | |
| 2 | 60 seconds * 25 frame=1500 | 107 | 7.13% | |
| 1 | 60 seconds * 25 frame=1500 | 119 | 7.93% | |
| 2 | 60 seconds * 25 frame=1500 | 142 | 9.47% | |
| 1 | 60 seconds * 25 frame=1500 | 101 | 6.73% | |
| 2 | 60 seconds * 25 frame=1500 | 97 | 6.47% | |
| 1 | 60 seconds * 25 frame=1500 | 111 | 7.40% | |
| 2 | 60 seconds * 25 frame=1500 | 127 | 8.47% | |
| 1 | 60 seconds * 25 frame=1500 | 117 | 7.80% | |
| 2 | 60 seconds * 25 frame=1500 | 109 | 7.27% | |
| It is total | 15000 | 1162 | 7.75% |
Seen from table 1, the rear compression engine significant effect based on video flow detection, average compression ratio are lower than 8%.
Can be realized Real Time Compression in efficiency, the strategy of compression is only the similarity system design according to target type, and
This comparison is based on time shaft, and for only temporally adjacent member just it is necessary to compare, this strategy not only conforms with target inspection
The business need of survey can more substantially reduce the number compared.
It facts have proved one thousandth of the cost not as good as target detection of compression of images, therefore can approximately ignore.
The present invention parses video file in compression of images first, is by the video file dismantling that playing duration is one second
The continuous 25 frame picture of time shaft (format is unlimited).The content of picture is detected and marked by depth learning technology, forms detection
Log and mark image, carry out picture classification for content, and the prototypical member for extracting every class becomes such " core ", finally with
The compression result of " core " as every class, it is assumed that the scale of every class is N, then such compression ratio is 1/N.It can be seen that N is bigger, compression ratio
Higher, compression effectiveness is better.Meanwhile the classification of picture is based on time shaft, therefore the calculation amount of similarity system design is much small
In N*N (do not compare two-by-two, but adjacent comparison).The present invention by the similarity between reduced time window adjacent image come
The quantity for simplifying figure rejects the similar adjacent image of high similarity to achieve the purpose that compression, more than traditional pixel compression
Efficiently and compression ratio is higher.
The above is the preferred embodiment of the present invention, for those skilled in the art,
Without departing from the principles of the invention, several improvements and modifications can also be made, these improvements and modifications are also regarded as this
The protection scope of invention.
Claims (10)
1. a kind of method for compressing image based on video flow detection, characterized in that the following steps are included:
Video file parsing: video file is disassembled as the continuous frame picture of time shaft；
The detection of frame picture: being detected to frame picture and marked the content of picture, is formed detection log and is marked image；
Compression of images: frame picture is merged.
2. a kind of method for compressing image based on video flow detection according to claim 1, characterized in that in video file
In resolving, by N frame per second fractionation video file, it is with stringent time sequencing and identical size, N between frame picture
Positive integer.
3. a kind of method for compressing image based on video flow detection according to claim 1, characterized in that examined in frame picture
During survey, the content of frame picture is marked using mature target detection model and records the detection log of every frame picture.
4. a kind of method for compressing image based on video flow detection according to claim 3, characterized in that the detection of picture
The content of log includes picture file name, detection moment, target position, target type, lasting frame number, frame adopts time slot and detection is set
Reliability.
5. a kind of method for compressing image based on video flow detection according to claim 1, characterized in that in compression of images
In the process, classify to frame picture, " core " picture of prototypical member as such of every class frame picture is extracted, with " core " picture
Compressed object as every class.
6. a kind of method for compressing image based on video flow detection according to claim 5, characterized in that in compression of images
In the process, more temporally adjacent two frames " core " picture, if the detection log content of two frames " core " picture height is similar into
Row merges.
7. a kind of method for compressing image based on video flow detection according to claim 6, characterized in that in compression of images
In the process, the first retention time is worth maximum " core " picture, deletes its neighbour, and " core " picture of second pair of reservation modifies its
Log.
8. a kind of method for compressing image based on video flow detection described in -7 any one according to claim 1, characterized in that
It is further comprising the steps of:
As a result it exports: exporting compressed image information.
9. a kind of method for compressing image based on video flow detection according to claim 8, characterized in that after the compression
Image information include: " core " picture, detection log and compression ratio.
10. a kind of method for compressing image based on video flow detection according to claim 9, characterized in that the compression
Than
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910572004.XA CN110267040A (en) | 2019-06-27 | 2019-06-27 | A kind of method for compressing image based on video flow detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910572004.XA CN110267040A (en) | 2019-06-27 | 2019-06-27 | A kind of method for compressing image based on video flow detection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110267040A true CN110267040A (en) | 2019-09-20 |
Family
ID=67922772
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910572004.XA Pending CN110267040A (en) | 2019-06-27 | 2019-06-27 | A kind of method for compressing image based on video flow detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110267040A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120099793A1 (en) * | 2010-10-20 | 2012-04-26 | Mrityunjay Kumar | Video summarization using sparse basis function combination |
| CN102819528A (en) * | 2011-06-10 | 2012-12-12 | 中国电信股份有限公司 | Method and device for generating video abstraction |
| CN104104952A (en) * | 2014-08-01 | 2014-10-15 | 武汉高德信息产业有限公司 | Audio/video processing method and system adapted to storage and play of mobile device |
| CN105354231A (en) * | 2015-09-30 | 2016-02-24 | 腾讯科技（深圳）有限公司 | Image selection method and apparatus, and image processing method and apparatus |
| CN105678322A (en) * | 2015-12-31 | 2016-06-15 | 百度在线网络技术（北京）有限公司 | Sample labeling method and apparatus |
| CN109195026A (en) * | 2018-10-15 | 2019-01-11 | 合肥师范学院 | Video abstraction generating method and system |
-
2019
- 2019-06-27 CN CN201910572004.XA patent/CN110267040A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120099793A1 (en) * | 2010-10-20 | 2012-04-26 | Mrityunjay Kumar | Video summarization using sparse basis function combination |
| CN102819528A (en) * | 2011-06-10 | 2012-12-12 | 中国电信股份有限公司 | Method and device for generating video abstraction |
| CN104104952A (en) * | 2014-08-01 | 2014-10-15 | 武汉高德信息产业有限公司 | Audio/video processing method and system adapted to storage and play of mobile device |
| CN105354231A (en) * | 2015-09-30 | 2016-02-24 | 腾讯科技（深圳）有限公司 | Image selection method and apparatus, and image processing method and apparatus |
| CN105678322A (en) * | 2015-12-31 | 2016-06-15 | 百度在线网络技术（北京）有限公司 | Sample labeling method and apparatus |
| CN109195026A (en) * | 2018-10-15 | 2019-01-11 | 合肥师范学院 | Video abstraction generating method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110189253A (en) | A kind of image super-resolution rebuilding method generating confrontation network based on improvement | |
| CN110111256B (en) | Image super-resolution reconstruction method based on residual distillation network | |
| CN104484643B (en) | The intelligent identification Method and system of a kind of handwriting table | |
| CN106851437A (en) | A kind of method for extracting video frequency abstract | |
| CN105721869B (en) | The collection of compression tensor and reconfiguration system based on structural sparse | |
| CN108364281A (en) | A kind of ribbon edge hair defect defect inspection method based on convolutional neural networks | |
| CN112580502A (en) | SICNN-based low-quality video face recognition method | |
| CN110503063A (en) | Fall detection method based on hourglass convolution autocoding neural network | |
| CN113487576A (en) | Insect pest image detection method based on channel attention mechanism | |
| CN115223219A (en) | Goat face identification method based on improved YOLOV4 | |
| Kesiman et al. | ICFHR 2018 competition on document image analysis tasks for southeast asian palm leaf manuscripts | |
| CN109299702B (en) | Human behavior recognition method and system based on depth space-time diagram | |
| CN104680189A (en) | Pornographic image detection method based on improved bag-of-words model | |
| CN110267040A (en) | A kind of method for compressing image based on video flow detection | |
| CN113011399A (en) | Video abnormal event detection method and system based on generation cooperative judgment network | |
| Bai et al. | A hierarchical face recognition method based on local binary pattern | |
| Megahed et al. | Face2face manipulation detection based on histogram of oriented gradients | |
| CN111611962A (en) | Face image super-resolution identification method based on fractional order multi-set partial least square | |
| CN101276476A (en) | Process for the separating prospect background of 2D cartoon animation | |
| CN115578774A (en) | Non-local channel attention-based face motion unit detection method | |
| Zhang et al. | Global contrast enhancement detection via deep multi-path network | |
| CN110400250A (en) | Intelligent image preprocess method and system based on AI | |
| CN111813996B (en) | Video searching method based on sampling parallelism of single frame and continuous multi-frame | |
| Liao et al. | Facial image quality assessment based on support vector machines | |
| CN110446067B (en) | Tensor decomposition-based video concentration method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190920 |