CN110309829A - The on-line monitoring method, apparatus and readable storage medium storing program for executing of the resources of movie & TV undercarriage - Google Patents
The on-line monitoring method, apparatus and readable storage medium storing program for executing of the resources of movie & TV undercarriage Download PDFInfo
- Publication number
- CN110309829A CN110309829A CN201910584893.1A CN201910584893A CN110309829A CN 110309829 A CN110309829 A CN 110309829A CN 201910584893 A CN201910584893 A CN 201910584893A CN 110309829 A CN110309829 A CN 110309829A
- Authority
- CN
- China
- Prior art keywords
- movie
- resources
- screenshot
- undercarriage
- picture
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Television Signal Processing For Recording (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
Abstract
The present invention relates to image identification technical field, embodiment specifically discloses the on-line monitoring method, apparatus and readable storage medium storing program for executing of a kind of the resources of movie & TV undercarriage.The identification of the resources of movie & TV undercarriage is carried out by the way that the broadcasting screenshot of the resources of movie & TV is inputted picture recognition model, realize the effect that notice undercarriage or direct undercarriage in time are unable to the resources of movie & TV of normal play, while promoting undercarriage the resources of movie & TV speed, manpower and time cost are saved, and user can be greatly avoided to open the resources of movie & TV of undercarriage, to promote the video display search of user and play experience.
Description
Technical field
The present invention relates to image identification technical fields, and in particular to a kind of on-line monitoring method of the resources of movie & TV undercarriage, dress
It sets and readable storage medium storing program for executing.
Background technique
The resources of movie & TV that current most of smart television producers play all is that third party provides, and TV producer passes through own
The resources of movie & TV state that searches out of search channel it is asynchronous with third-party the resources of movie & TV state, cause user by searching for and broadcasting
The resources of movie & TV an of undercarriage is put.If marking the resources of movie & TV of undercarriage by the way of manual review, take time and effort.
Summary of the invention
In view of this, the application current smart television producer the resources of movie & TV undercarriage not in time aiming at the problem that, provide one
Kind realizes the resources of movie & TV for notifying to be unable to normal play in time with undercarriage according to the method for undercarriage picture recognition the resources of movie & TV undercarriage
Method and device.
In order to solve the above technical problems, technical solution provided by the invention is a kind of on-line monitoring side of the resources of movie & TV undercarriage
Method, comprising:
It obtains the source-information of the resources of movie & TV to be played and plays screenshot；
It is pre-processed to screenshot is played；
By treated, broadcasting screenshot input picture recognition model carries out the identification of the resources of movie & TV undercarriage；
If picture recognition model determines the resources of movie & TV undercarriage, undercarriage is issued according to the source-information of the resources of movie & TV to be played
Notice or the undercarriage the resources of movie & TV.
Preferably, the source-information for obtaining the resources of movie & TV to be played and the method for playing screenshot, comprising:
The operational order for playing the resources of movie & TV is received, the source-information of the resources of movie & TV to be played is obtained；
Within a preset time, if receiving the operational order of full frame the resources of movie & TV, refer in the operation for executing full frame the resources of movie & TV
Broadcasting screenshot is obtained after order to obtain before executing other operational orders if receiving other operational orders other than full frame the resources of movie & TV
Broadcasting screenshot is taken, if not receiving any one operational order, obtains and plays screenshot.
Preferably, described pair of broadcasting screenshot carries out pretreated method, comprising:
Judge to play whether screenshot is played in full screen screenshot；
If it is not, then obtaining picture in the broadcasting frame in currently playing screenshot, will play picture compression in frame is specified pixel
Broadcasting screenshot；
If so, by the broadcasting screenshot of played in full screen screenshot boil down to specified pixel.
Preferably, the method for playing picture in frame obtained in currently playing screenshot, including；
Currently playing screenshot is subjected to gray processing, obtains gray scale picture；
The high-frequency noise being superimposed in gray scale picture is filtered off, denoising picture is obtained；
Limb recognition is carried out to denoising picture using Canny algorithm, obtains identification picture；
Identification picture is subjected to binary conversion treatment, obtains binaryzation picture；
The framing mask for extracting binaryzation picture carries out rectangle fitting to image outline point；
The boundary point that scanning patter is selected in rectangle is cut, and obtains playing picture in frame.
Preferably, described that limb recognition is carried out to denoising picture using Canny algorithm, the method for obtaining identification picture, packet
It includes:
Amplitude and the direction of denoising picture gradient are calculated with the finite difference method of single order local derviation；
Using the point of preset rules filtering non-edge, the width for denoising image edge is made to be close to 1 pixel；
According to preset two threshold values maxVal and minVal detect denoising image edge, wherein greater than maxVal all by
It is detected as edge, is all detected as non-edge lower than minVal.
Preferably, it before the step of source-information for obtaining the resources of movie & TV to be played and broadcasting screenshot, further comprises the steps of:
Establish picture recognition model；The method for establishing picture recognition model, comprising:
The undercarriage picture for obtaining the resources of movie & TV in each source is pre-processed；
Load pretreated undercarriage picture；
Construct first convolutional layer；
Construct first pond layer；
Construct second convolutional layer；
Construct second pond layer；
Construct full articulamentum；
Use Dropout training deep neural network.
The present invention also provides a kind of on-line monitoring devices of the resources of movie & TV undercarriage, comprising:
Screenshot obtains module, for obtaining the source-information of the resources of movie & TV to be played and playing screenshot；
Screenshot processing module, for being pre-processed to broadcasting screenshot；
Screenshot identification module, for broadcasting screenshot input picture recognition model to carry out the knowledge of the resources of movie & TV undercarriage by treated
Not；
Off-frame treatment module, if the resources of movie & TV undercarriage is determined for picture recognition model, according to the resources of movie & TV to be played
Source-information issue undercarriage notice or the undercarriage the resources of movie & TV.
Preferably, the screenshot acquisition module includes:
Source acquiring unit obtains coming for the resources of movie & TV to be played for receiving the operational order for playing the resources of movie & TV
Source information；
Screenshot acquiring unit is used within a preset time, full frame executing if receiving the operational order of full frame the resources of movie & TV
Broadcasting screenshot is obtained after the operational order of the resources of movie & TV to execute if receiving other operational orders other than full frame the resources of movie & TV
It is obtained before other operational orders and plays screenshot, if not receiving any one operational order, obtained and play screenshot.
Preferably, the screenshot processing module carries out pretreated method to screenshot is played, comprising:
Judge to play whether screenshot is played in full screen screenshot；
If it is not, then obtaining picture in the broadcasting frame in currently playing screenshot, will play picture compression in frame is specified pixel
Broadcasting screenshot；
If so, by the broadcasting screenshot of played in full screen screenshot boil down to specified pixel.
Preferably, the method for playing picture in frame obtained in currently playing screenshot, including；
Currently playing screenshot is subjected to gray processing, obtains gray scale picture；
The high-frequency noise being superimposed in gray scale picture is filtered off, denoising picture is obtained；
Limb recognition is carried out to denoising picture using Canny algorithm, obtains identification picture；
Identification picture is subjected to binary conversion treatment, obtains binaryzation picture；
The framing mask for extracting binaryzation picture carries out rectangle fitting to image outline point；
The boundary point that scanning patter is selected in rectangle is cut, and obtains playing picture in frame.
Preferably, described that limb recognition is carried out to denoising picture using Canny algorithm, the method for obtaining identification picture, packet
It includes:
Amplitude and the direction of denoising picture gradient are calculated with the finite difference method of single order local derviation；
Using the point of preset rules filtering non-edge, the width for denoising image edge is made to be close to 1 pixel；
According to preset two threshold values maxVal and minVal detect denoising image edge, wherein greater than maxVal all by
It is detected as edge, is all detected as non-edge lower than minVal.
Preferably, the on-line monitoring device of the resources of movie & TV undercarriage further includes model generation module, and the model generates
The method that module establishes picture recognition model, comprising:
The undercarriage picture for obtaining the resources of movie & TV in each source is pre-processed；
Load pretreated undercarriage picture；
Construct first convolutional layer；
Construct first pond layer；
Construct second convolutional layer；
Construct second pond layer；
Construct full articulamentum；
Use Dropout training deep neural network.
The present invention also provides a kind of on-line monitoring devices of the resources of movie & TV undercarriage, comprising:
Memory, for storing computer program；
Processor, for executing the computer program to realize such as the on-line monitoring method of above-mentioned the resources of movie & TV undercarriage
Step.
The present invention also provides a kind of computer readable storage medium, the computer-readable recording medium storage has computer
Program, when the computer program is executed by processor the step of the realization such as on-line monitoring method of above-mentioned the resources of movie & TV undercarriage.
Compared with prior art, detailed description are as follows for its advantages by the application: the broadcasting screenshot of the resources of movie & TV is inputted
Picture recognition model carries out the identification of the resources of movie & TV undercarriage, realizes the resources of movie & TV for notifying to be unable to normal play in time with undercarriage
Effect has saved manpower and time cost, and can greatly avoid user while promoting undercarriage the resources of movie & TV speed
The resources of movie & TV of undercarriage is opened, to promote the video display search of user and play experience.
Detailed description of the invention
In order to illustrate the embodiments of the present invention more clearly, attached drawing needed in the embodiment will be done simply below
It introduces, it should be apparent that, drawings in the following description are only some embodiments of the invention, for ordinary skill people
For member, without creative efforts, it is also possible to obtain other drawings based on these drawings.
Fig. 1 is a kind of flow chart of the on-line monitoring method of the resources of movie & TV undercarriage provided in an embodiment of the present invention；
Fig. 2 is a kind of source-information for obtaining the resources of movie & TV to be played provided in an embodiment of the present invention and the side for playing screenshot
The flow chart of method；
Fig. 3 is a kind of flow chart that pretreated method is carried out to broadcasting screenshot provided in an embodiment of the present invention；
Fig. 4 is a kind of stream for playing the method for picture in frame obtained in currently playing screenshot provided in an embodiment of the present invention
Cheng Tu；
Fig. 5 is a kind of flow chart for the method for establishing picture recognition model provided in an embodiment of the present invention；
Fig. 6 is a kind of structure chart of the on-line monitoring device of the resources of movie & TV undercarriage provided in an embodiment of the present invention；
Fig. 7 is the structure chart of the on-line monitoring device of another the resources of movie & TV undercarriage provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, rather than whole embodiments.Based on this
Embodiment in invention, those of ordinary skill in the art are without making creative work, obtained every other
Embodiment belongs to the scope of the present invention.
It is with reference to the accompanying drawing and specific real in order to make those skilled in the art more fully understand technical solution of the present invention
Applying example, the present invention is described in further detail.
The resources of movie & TV that current most of smart television producers play all is that third party provides, and TV producer passes through own
The resources of movie & TV state that searches out of search channel it is asynchronous with third-party the resources of movie & TV state, cause user by searching for and broadcasting
The resources of movie & TV an of undercarriage is put.If marking the resources of movie & TV of undercarriage by the way of manual review, take time and effort.
The embodiment of the invention provides a kind of online monitoring systems of the resources of movie & TV undercarriage, include client and service
Device, multiple client can share 1 server, and 1 client represents 1 Intelligent television terminal.
As shown in Figure 1, the embodiment of the present invention provides a kind of on-line monitoring method of the resources of movie & TV undercarriage, visitor can be applied to
Family end and/or server, specific method include:
S11: obtaining the source-information of the resources of movie & TV to be played and plays screenshot；
S12: it is pre-processed to screenshot is played；
S13: by treated, broadcasting screenshot input picture recognition model carries out the identification of the resources of movie & TV undercarriage；
S14: it if picture recognition model determines the resources of movie & TV undercarriage, is issued according to the source-information of the resources of movie & TV to be played
Undercarriage notice or the undercarriage the resources of movie & TV.
It should be noted that as shown in Fig. 2, step S11 obtains the source-information of the resources of movie & TV to be played and plays screenshot
Method, can apply in client, when user opens a video display, message of film and TV and video display be played into screenshot and uploaded
To server, specific method includes: to carry out before next movement that (movement here does not include complete after playing when the user clicks
Screen operation) or broadcast page jump to other interfaces automatically before carry out screenshot.Client can direct compressed picture, will compress
Picture and corresponding video display source-information afterwards is uploaded to server, does not pre-process.Client can also direct compressed picture,
By compressed picture and corresponding video display source-information, continuation is pre-processed in client, and specific method includes:
S111: the operational order for playing the resources of movie & TV is received, the source-information of the resources of movie & TV to be played is obtained；
S112: within a preset time, if receiving the operational order of full frame the resources of movie & TV, in the behaviour for executing full frame the resources of movie & TV
It is obtained after instructing and plays screenshot, if receiving other operational orders other than full frame the resources of movie & TV, executing other operational orders
Preceding acquisition plays screenshot, if not receiving any one operational order, obtains and plays screenshot.
If directly carrying out screenshot for example, client does not carry out any operation after receiving play instruction in 10 seconds,
In 10 seconds (empirical value), user issues next instruction (non-full frame instruction), then executes preceding screenshot immediately in next instruction, if
User issues full frame instruction in 10 seconds, i.e. user is full frame by currently playing video display, then Shi Bujin occurs in the full frame operation
Row screenshot carries out screenshot again after the completion of full frame operation.The broadcasting screenshot of acquisition is uploaded onto the server after being compressed carry out under
The pretreatment of one step.
It should be noted that as shown in figure 3, step S12 can be applied to service to the pretreated method of screenshot progress is played
Device or client, by taking server as an example, server is pre-processed first after receiving broadcasting screenshot, comprising:
S121: judge to play whether screenshot is played in full screen screenshot；
S122: if it is not, then obtaining picture in the broadcasting frame in currently playing screenshot, it is specified for playing picture compression in frame
The broadcasting screenshot of pixel；
S123: if so, by the broadcasting screenshot of played in full screen screenshot boil down to specified pixel.
It should be noted that as shown in figure 4, since position of the video display playing pictures in broadcasting screenshot has with video display source
It closes, so needing to be cut according to different broadcasting sources, extracts the picture played in frame, the broadcasting screenshot after full frame operation is not
Need to do cutting processing, specifically, the method for playing picture in frame in currently playing screenshot is obtained in step S122, including；
S1221: currently playing screenshot is subjected to gray processing, obtains gray scale picture；
Specifically, playing screenshot is color image, needs that screenshot gray processing will be played, i.e., color image is switched into grayscale image
Piece.Gray processing is carried out to a width color image, is exactly weighted and averaged according to the sampled value in each channel of image.Weighted average
Formula are as follows: Gray=0.299R+0.587G+0.114B.
S1222: filtering off the high-frequency noise being superimposed in gray scale picture, obtains denoising picture；
Specifically, since the core that Gaussian function determines can provide between anti-noise jamming and edge detection are accurately positioned
Good trade-off scheme.Added according to the gray value of pixel and its neighborhood point to be filtered according to certain parameter rule
Weight average can effectively filter off the high-frequency noise being superimposed in ideal image in this way.By following two-dimensional Gaussian function, parameter is determined
Obtain two-dimensional nucleus vector (wherein Sigma square for variance):
This is gaussian kernel function formula, wherein π is pi, and m, n are
Image coordinate location；
One Gaussian matrix takes the average value of its Weight as last gray scale multiplied by each pixel and its neighborhood
Value.I.e. for the pixel of a position (m, n), gray value (only considering binary map here) is f (m, n).So through excessively high
This filtered gray value will become:
S1223: limb recognition is carried out to denoising picture using Canny algorithm, obtains identification picture；
Specifically, limb recognition is carried out to denoising picture using Canny algorithm, the method for obtaining identification picture, comprising:
(1) amplitude and the direction of denoising picture gradient are calculated with the finite difference method of single order local derviation；
Specifically, calculating amplitude and the direction of gradient with the finite difference of single order local derviation, common gradient operator has:
Roberts operator, Sobel operator, Prewitt operator etc..Here using method used by Canny algorithm, using such as last volume
Integrating, finds out the gradient value g of different directionsx(m, n), gy(m, n), wherein gx、gyFor the gradient value in the direction x and y；
Wherein, Sx、SyFor x and y Directional partial derivative calculation template；
Gradient value and gradient direction calculation formula are as follows:
(2) using the point of preset rules filtering non-edge, the width for denoising image edge is made to be close to 1 pixel；
Specifically, edge is possible to be exaggerated during gaussian filtering.The step is filtered using a rule
It is not the point at edge, make the width at edge is 1 pixel as far as possible: if a pixel belongs to edge, this picture
Gradient value of the vegetarian refreshments on gradient direction is the largest.Otherwise it is not edge, gray value is set as 0.
(3) denoising image edge is detected according to preset two threshold values maxVal and minVal, wherein greater than maxVal's
It is all detected as edge, is all detected as non-edge lower than minVal；
Specifically, two threshold values (threshold) of setting, respectively maxVal and minVal.Wherein greater than maxVal's
Pixel is all detected as edge, and the pixel lower than minVal is all detected as non-edge.For intermediate pixel, such as
Fruit is adjacent with the pixel for being determined as edge, then is determined as edge；It otherwise is non-edge, it is accurate to can be improved using dual threshold detection
Rate.
S1224: identification picture is subjected to binary conversion treatment, obtains binaryzation picture；
Specifically, image binaryzation is exactly to set the gray value of the pixel on image to 0 or 255, that is, it will be whole
A image shows apparent black and white effect.It is judged as the pixel that all gray scales are greater than or equal to threshold value to belong to specific object
Body, gray value be 255 indicate, otherwise these pixels are excluded other than object area, gray value 0, indicate background or
The object area of person's exception.
S1225: extracting the framing mask of binaryzation picture, carries out rectangle fitting to image outline point；
S1226: the boundary point that scanning patter is selected in rectangle is cut, and obtains playing picture in frame.
It is played in frame after picture it should be noted that obtaining, screenshot picture scaling will be played using bilinear interpolation algorithm
, in practice can be in the case where guaranteeing model accuracy rate at specified pixel size (N*N), appropriate adjustment picture pixels size, with
It reduces memory consumption and shortens the training time.
Bilinear interpolation core concept is to carry out once linear interpolation respectively in both direction.If we expect position
For function f in the value of P=(x, y), formula is as follows:
Linear interpolation is carried out in the direction x first, is obtained:
Wherein, f (Q11)、f(Q21) it is function f in (x1,y1)、(x2,y1) on value,
Wherein, f (Q12)、f(Q22) it is function f in (x1,y2)、(x2,y2) on value.
Then linear interpolation is carried out in the direction y, obtained:
Wherein, y is coordinate position.
Desired result f (x, y) is thus obtained,
Further, it is also possible to using arest neighbors interpolation method, bicubic interpolation, the image interpolation algorithm based on region, due to examining
Consider processing speed, bilinear interpolation is directlyed adopt, for the situation, due to being reduced to screenshot, also it is contemplated that using base
Image interpolation algorithm in region.
It should be noted that before the step of step S11 obtains the source-information of the resources of movie & TV to be played and plays screenshot, also
Comprising steps of S10: establishing picture recognition model.
It should be noted that as shown in figure 5, the method that step S10 establishes picture recognition model, comprising:
S101: the undercarriage picture for obtaining the resources of movie & TV in each source is pre-processed；Specifically method includes:
(1) the undercarriage picture that obtains the resources of movie & TV in each source includes intercepting the resources of movie & TV in each source to broadcast
Put the processing steps such as undercarriage picture and the compressed picture in frame.
(2) for first training data, by being manually labeled to the picture of processing, normal play is labeled as 1, no
Energy normal play is labeled as 0.Subsequently through the picture constantly uploaded, gradually abundant data, is added model training；
(3) since the picture normal play quantity of extraction accounts for all picture large percentages, it is impossible to normal play according to
Certain proportion replicates multiple and is unable to normal play picture, remains consistent substantially with energy normal play picture ratio.
S102: pretreated undercarriage picture is loaded；
S103: first convolutional layer of building；
Specifically, defining convolution kernel (5*5), convolution kernel number is 32, step-length 1, and channel is 1 (only with single channel gray scale
Image), then after handling, export the picture that 32 pixel sizes are 28*28.
S104: first pond layer of building；
Specifically, by 2*2 filter, step-length 2 obtains 32 by average value pond by 32 output pictures of convolution
Open the picture of 14*14.
S105: second convolutional layer of building；
Specifically, defining convolution kernel (3*3), convolution kernel number is 64, step-length 1, and channel is 1 (only with single channel gray scale
Image), then after handling, export the picture that 64 pixel sizes are 12*12.
S106: second pond layer of building；
Specifically, by 2*2 filter, step-length 2 obtains 64 by average value pond by 64 output pictures of convolution
Open the picture of 6*6.
S107: full articulamentum is constructed；
Specifically, 16 global characteristics are exported using 16 6*6 convolution nuclear convolutions for the local feature that convolution is extracted,
It then can determine whether video display can play by these features.
S108: Dropout training deep neural network is used.
Specifically, preventing over-fitting using Dropout, Dropout can be used as one kind of trained deep neural network
Trick is selective.In each trained batch, by ignoring the property detector (the hidden node value met half way is 0) of half,
Over-fitting can significantly be reduced.This mode can reduce the interaction between property detector (hidden node), inspection
It surveys device interaction and refers to that certain detectors rely on the effect of other detector competence exertions.
Wherein, the hidden layer number of plies and convolution kernel can be adjusted according to actual effect, such as (1) Lenet model, if all using
5*5 convolution kernel, pondization is using maximum pond:
(1) size: 32*32 is inputted；
(2) convolutional layer: 3；
(3) down-sampled layer: 2；
(4) full articulamentum: 1；
(5) it exports: 2 classifications (whether can play).
It should be noted that according to default accuracy rate adjusting parameter, (input picture pixels are big after establishing picture recognition model
It is small, CNN model number of plies etc.).Finally use tensorflow-serving deployment model.
It should be noted that the specific picture recognition step of server end includes: the broadcasting for upload in step S13
Picture is scaled to specified image size (pretreatment mode and mould using bilinear interpolation algorithm after cutting by screenshot picture
Data prediction mode is consistent when type training), and structural network identical with training pattern is inputted, load has trained
Model, predicted, if being predicted as abnormal broadcasting, prompt artificial to carry out off-frame treatment or directly automatic carry out undercarriage
Processing.
As shown in fig. 6, the present invention also provides a kind of on-line monitoring devices of the resources of movie & TV undercarriage, comprising:
Screenshot obtains module 21, for obtaining the source-information of the resources of movie & TV to be played and playing screenshot；
Screenshot processing module 22, for being pre-processed to broadcasting screenshot；
Screenshot identification module 23, for broadcasting screenshot input picture recognition model to carry out the resources of movie & TV undercarriage by treated
Identification；
Off-frame treatment module 24 provides if determining the resources of movie & TV undercarriage for picture recognition model according to video display to be played
The source-information in source issues undercarriage notice or the undercarriage the resources of movie & TV.
It should be noted that screenshot acquisition module 21 includes:
Source acquiring unit obtains coming for the resources of movie & TV to be played for receiving the operational order for playing the resources of movie & TV
Source information；
Screenshot acquiring unit is used within a preset time, full frame executing if receiving the operational order of full frame the resources of movie & TV
Broadcasting screenshot is obtained after the operational order of the resources of movie & TV to execute if receiving other operational orders other than full frame the resources of movie & TV
It is obtained before other operational orders and plays screenshot, if not receiving any one operational order, obtained and play screenshot.
It should be noted that the screenshot processing module carries out pretreated method to screenshot is played, comprising:
Judge to play whether screenshot is played in full screen screenshot；
If it is not, then obtaining picture in the broadcasting frame in currently playing screenshot, will play picture compression in frame is specified pixel
Broadcasting screenshot；
If so, by the broadcasting screenshot of played in full screen screenshot boil down to specified pixel.
It should be noted that the method for playing picture in frame obtained in currently playing screenshot, including；
Currently playing screenshot is subjected to gray processing, obtains gray scale picture；
The high-frequency noise being superimposed in gray scale picture is filtered off, denoising picture is obtained；
Limb recognition is carried out to denoising picture using Canny algorithm, obtains identification picture；
Identification picture is subjected to binary conversion treatment, obtains binaryzation picture；
The framing mask for extracting binaryzation picture carries out rectangle fitting to image outline point；
The boundary point that scanning patter is selected in rectangle is cut, and obtains playing picture in frame.
It should be noted that described carry out limb recognition to denoising picture using Canny algorithm, the side of identification picture is obtained
Method, comprising:
Amplitude and the direction of denoising picture gradient are calculated with the finite difference method of single order local derviation；
Using the point of preset rules filtering non-edge, the width for denoising image edge is made to be close to 1 pixel；
According to preset two threshold values maxVal and minVal detect denoising image edge, wherein greater than maxVal all by
It is detected as edge, is all detected as non-edge lower than minVal.
As shown in fig. 7, the embodiment of the present invention also provides the on-line monitoring device of another the resources of movie & TV undercarriage, Fig. 6's
It further include model generation module 20 on the basis of on-line monitoring device.Model generation module 20 establishes the side of picture recognition model
Method, comprising:
The undercarriage picture for obtaining the resources of movie & TV in each source is pre-processed；
Load pretreated undercarriage picture；
Construct first convolutional layer；
Construct first pond layer；
Construct second convolutional layer；
Construct second pond layer；
Construct full articulamentum；
Use Dropout training deep neural network.
The embodiment of the present invention also provides a kind of on-line monitoring device of the resources of movie & TV undercarriage, comprising: memory, for storing
Computer program；Processor realizes the on-line monitoring method of above-mentioned the resources of movie & TV undercarriage for executing the computer program
The step of.
The embodiment of the present invention also provides a kind of computer readable storage medium, and computer-readable recording medium storage has calculating
Machine program realizes the step of the on-line monitoring method such as above-mentioned the resources of movie & TV undercarriage when the computer program is executed by processor
Suddenly.
The explanation of feature may refer to the related description of embodiment corresponding to Fig. 1-Fig. 5 in embodiment corresponding to Fig. 6-Fig. 7,
Here it no longer repeats one by one.
It is provided for the embodiments of the invention a kind of on-line monitoring method, apparatus of the resources of movie & TV undercarriage above and readable deposits
Storage media is described in detail.Each embodiment is described in a progressive manner in specification, and each embodiment stresses
Be the difference from other embodiments, the same or similar parts in each embodiment may refer to each other.For implementing
For device disclosed in example, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place ginseng
See method part illustration.It should be pointed out that for those skilled in the art, not departing from original of the invention
, can be with several improvements and modifications are made to the present invention under the premise of reason, these improvement and modification also fall into right of the present invention and want
In the protection scope asked.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure
And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
The step of method described in conjunction with the examples disclosed in this document or algorithm, can directly be held with hardware, processor
The combination of capable software module or the two is implemented.Software module can be placed in random access memory (RAM), memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In any other form of storage medium well known in field.
Claims (10)
1. a kind of on-line monitoring method of the resources of movie & TV undercarriage characterized by comprising
It obtains the source-information of the resources of movie & TV to be played and plays screenshot；
It is pre-processed to screenshot is played；
By treated, broadcasting screenshot input picture recognition model carries out the identification of the resources of movie & TV undercarriage；
If picture recognition model determines the resources of movie & TV undercarriage, undercarriage notice is issued according to the source-information of the resources of movie & TV to be played
Or the undercarriage the resources of movie & TV.
2. the on-line monitoring method of the resources of movie & TV undercarriage according to claim 1, which is characterized in that the acquisition is to be played
The source-information of the resources of movie & TV and the method for playing screenshot, comprising:
The operational order for playing the resources of movie & TV is received, the source-information of the resources of movie & TV to be played is obtained；
Within a preset time, if receiving the operational order of full frame the resources of movie & TV, after the operational order for executing full frame the resources of movie & TV
Broadcasting screenshot is obtained to obtain and broadcast before executing other operational orders if receiving other operational orders other than full frame the resources of movie & TV
Screenshot is put, if not receiving any one operational order, obtains and plays screenshot.
3. the on-line monitoring method of the resources of movie & TV undercarriage according to claim 1, which is characterized in that described pair of broadcasting screenshot
Carry out pretreated method, comprising:
Judge to play whether screenshot is played in full screen screenshot；
If it is not, then obtaining picture in the broadcasting frame in currently playing screenshot, will play picture compression in frame is broadcasting for specified pixel
Put screenshot；
If so, by the broadcasting screenshot of played in full screen screenshot boil down to specified pixel.
4. the on-line monitoring method of the resources of movie & TV undercarriage according to claim 3, which is characterized in that the acquisition is currently broadcast
The method for playing picture in frame in screenshot is put, including；
Currently playing screenshot is subjected to gray processing, obtains gray scale picture；
The high-frequency noise being superimposed in gray scale picture is filtered off, denoising picture is obtained；
Limb recognition is carried out to denoising picture using Canny algorithm, obtains identification picture；
Identification picture is subjected to binary conversion treatment, obtains binaryzation picture；
The framing mask for extracting binaryzation picture carries out rectangle fitting to image outline point；
The boundary point that scanning patter is selected in rectangle is cut, and obtains playing picture in frame.
5. the on-line monitoring method of the resources of movie & TV undercarriage according to claim 4, which is characterized in that described to use Canny
Algorithm carries out limb recognition to denoising picture, the method for obtaining identification picture, comprising:
Amplitude and the direction of denoising picture gradient are calculated with the finite difference method of single order local derviation；
Using the point of preset rules filtering non-edge, the width for denoising image edge is made to be close to 1 pixel；
Denoising image edge is detected according to preset two threshold values maxVal and minVal, wherein being all detected greater than maxVal
For edge, non-edge is all detected as lower than minVal.
6. the on-line monitoring method of the resources of movie & TV undercarriage according to claim 1, which is characterized in that the acquisition is to be played
Before the step of source-information and broadcasting screenshot of the resources of movie & TV, further comprises the steps of: and establish picture recognition model；It is described to establish picture
The method of identification model, comprising:
The undercarriage picture for obtaining the resources of movie & TV in each source is pre-processed；
Load pretreated undercarriage picture；
Construct first convolutional layer；
Construct first pond layer；
Construct second convolutional layer；
Construct second pond layer；
Construct full articulamentum；
Use Dropout training deep neural network.
7. a kind of on-line monitoring device of the resources of movie & TV undercarriage characterized by comprising
Screenshot obtains module, for obtaining the source-information of the resources of movie & TV to be played and playing screenshot；
Screenshot processing module, for being pre-processed to broadcasting screenshot；
Screenshot identification module, for broadcasting screenshot input picture recognition model to carry out the identification of the resources of movie & TV undercarriage by treated；
Off-frame treatment module, if determine the resources of movie & TV undercarriage for picture recognition model, according to the resources of movie & TV to be played come
Source information issues undercarriage notice or the undercarriage the resources of movie & TV.
8. the on-line monitoring device of the resources of movie & TV undercarriage according to claim 7, which is characterized in that the screenshot obtains mould
Block includes:
Source acquiring unit obtains the source letter of the resources of movie & TV to be played for receiving the operational order for playing the resources of movie & TV
Breath；
Screenshot acquiring unit, within a preset time, if receiving the operational order of full frame the resources of movie & TV, executing full frame video display
It is obtained after the operational order of resource and plays screenshot, if receiving other operational orders other than full frame the resources of movie & TV, executing other
It is obtained before operational order and plays screenshot, if not receiving any one operational order, obtained and play screenshot.
9. a kind of on-line monitoring device of the resources of movie & TV undercarriage characterized by comprising
Memory, for storing computer program；
Processor, for executing the computer program to realize the resources of movie & TV undercarriage as described in any one of claims 1 to 6
On-line monitoring method the step of.
10. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey
Sequence, when the computer program is executed by processor realize as described in any one of claims 1 to 6 the resources of movie & TV undercarriage
The step of line monitoring method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910584893.1A CN110309829B (en) | 2019-07-01 | 2019-07-01 | Online monitoring method and device for off-shelf of film and television resources and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910584893.1A CN110309829B (en) | 2019-07-01 | 2019-07-01 | Online monitoring method and device for off-shelf of film and television resources and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110309829A true CN110309829A (en) | 2019-10-08 |
| CN110309829B CN110309829B (en) | 2021-06-29 |
Family
ID=68078654
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910584893.1A Active CN110309829B (en) | 2019-07-01 | 2019-07-01 | Online monitoring method and device for off-shelf of film and television resources and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110309829B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113515670A (en) * | 2021-08-12 | 2021-10-19 | 成都极米科技股份有限公司 | Method, device and storage medium for identifying state of movie and television resource |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7672482B2 (en) * | 2006-05-16 | 2010-03-02 | Eastman Kodak Company | Shape detection using coherent appearance modeling |
| CN102024224A (en) * | 2009-09-11 | 2011-04-20 | 阿里巴巴集团控股有限公司 | E-commerce system and method for getting commodities on and/or off shelves at optimal time |
| CN106791517A (en) * | 2016-11-21 | 2017-05-31 | 广州爱九游信息技术有限公司 | Live video detection method, device and service end |
| CN106777116A (en) * | 2016-12-15 | 2017-05-31 | 腾讯科技（深圳）有限公司 | A kind of content acquisition method, subscription client, server and system |
| CN108647588A (en) * | 2018-04-24 | 2018-10-12 | 广州绿怡信息科技有限公司 | Goods categories recognition methods, device, computer equipment and storage medium |
| CN109327715A (en) * | 2018-08-01 | 2019-02-12 | 阿里巴巴集团控股有限公司 | A kind of video Risk Identification Method, device and equipment |
-
2019
- 2019-07-01 CN CN201910584893.1A patent/CN110309829B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7672482B2 (en) * | 2006-05-16 | 2010-03-02 | Eastman Kodak Company | Shape detection using coherent appearance modeling |
| CN102024224A (en) * | 2009-09-11 | 2011-04-20 | 阿里巴巴集团控股有限公司 | E-commerce system and method for getting commodities on and/or off shelves at optimal time |
| CN106791517A (en) * | 2016-11-21 | 2017-05-31 | 广州爱九游信息技术有限公司 | Live video detection method, device and service end |
| CN106777116A (en) * | 2016-12-15 | 2017-05-31 | 腾讯科技（深圳）有限公司 | A kind of content acquisition method, subscription client, server and system |
| CN108647588A (en) * | 2018-04-24 | 2018-10-12 | 广州绿怡信息科技有限公司 | Goods categories recognition methods, device, computer equipment and storage medium |
| CN109327715A (en) * | 2018-08-01 | 2019-02-12 | 阿里巴巴集团控股有限公司 | A kind of video Risk Identification Method, device and equipment |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113515670A (en) * | 2021-08-12 | 2021-10-19 | 成都极米科技股份有限公司 | Method, device and storage medium for identifying state of movie and television resource |
| CN113515670B (en) * | 2021-08-12 | 2023-09-26 | 极米科技股份有限公司 | Film and television resource state identification method, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110309829B (en) | 2021-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jiang et al. | Unsupervised decomposition and correction network for low-light image enhancement | |
| CN112330574B (en) | Portrait restoration method and device, electronic equipment and computer storage medium | |
| CN104067311B (en) | Digital makeup | |
| CN109472260B (en) | Method for removing station caption and subtitle in image based on deep neural network | |
| US11887218B2 (en) | Image optimization method, apparatus, device and storage medium | |
| CN109472193A (en) | Method for detecting human face and device | |
| Liu et al. | A motion deblur method based on multi-scale high frequency residual image learning | |
| CN110147753A (en) | Method and device for detecting small objects in image | |
| Yang et al. | Modeling the screen content image quality via multiscale edge attention similarity | |
| Chen et al. | Saliency-directed image interpolation using particle swarm optimization | |
| CN104732499A (en) | Retina image enhancement algorithm based on multiple scales and multiple directions | |
| CN109711268A (en) | A kind of facial image screening technique and equipment | |
| CN108961227A (en) | A kind of image quality evaluating method based on airspace and transform domain multiple features fusion | |
| CN109657597A (en) | Anomaly detection method towards individual live scene | |
| CN109829868A (en) | A kind of lightweight deep learning model image defogging method, electronic equipment and medium | |
| CN115100334B (en) | Image edge tracing and image animation method, device and storage medium | |
| CN110390673A (en) | Cigarette automatic testing method based on deep learning under a kind of monitoring scene | |
| CN108986145A (en) | Method of video image processing and device | |
| Yuan et al. | Single image dehazing via NIN-DehazeNet | |
| CN112036209A (en) | Portrait photo processing method and terminal | |
| CN110334590A (en) | Image Acquisition bootstrap technique and device | |
| CN110309829A (en) | The on-line monitoring method, apparatus and readable storage medium storing program for executing of the resources of movie & TV undercarriage | |
| CN108647605A (en) | A kind of combination global color and the human eye of partial structurtes feature stare point extracting method | |
| CN106651796A (en) | Image or video processing method and system | |
| CN110443780A (en) | A kind of PPT frame extracting method and relevant device based on OpenCV algorithm |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |