Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, rightThe present invention is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the present invention, notFor limiting the present invention.
It should be noted that term involved in the embodiment of the present invention " first second " be only be similar pair of differenceAs not representing the particular sorted for object, it is possible to understand that ground, " first second " can be interchanged specific in the case where permissionSequence or precedence.It should be understood that the object that " first second " is distinguished is interchangeable under appropriate circumstances, so as to retouch hereThe embodiment of the present invention stated can be performed in other sequences than those illustrated or described herein.
The processing method of video image provided by the invention, can be applied in application scenarios as shown in Figure 1, and Fig. 1 isThe application scenario diagram of the processing method of video image in one embodiment, wherein first terminal 100 can pass through network and secondTerminal 200 is communicatively coupled.Wherein, first terminal 100 and second terminal 200 can be, but not limited to be various individual calculusMachine, laptop, smart phone and tablet computer.
When sending video image, first terminal 100 can correspond to transmitting terminal, and second terminal 200 can correspond to connectVideo image can be sent to second terminal 200 by network by receiving end, first terminal 100, and second terminal 200 can receiveIt is shown after to video image.In this way, user can pass through first terminal in the application scenarios such as such as teleconferenceThe video image at 100 real-time recording meeting scenes, while the video image at meeting scene is passed through into network using first terminal 100It is transferred to second terminal 200 to be shown, the user of second terminal 200 is carried out remotely by the second terminal 200Meeting.However, certain fluctuation may occur for network state, such as by original in the process of transmission of video imagesHigh network quality bandwidth status (network flow high speed, stable or packet loss are few) become low quality network bandwidth state (flow low speed,Stability difference or packet loss are big) when, first terminal 100 would generally first reduce original image quality of video image, then by the video figureAs being transferred to second terminal 200, to adapt to current low quality network bandwidth state, and second terminal 200 is receiving reductionIt after the video image of image quality, if the video image shown, is easy to cause the image quality of display relatively low, influences long-rangeThe quality of meeting also influences the perception of user.
The processing method for the video image that various embodiments of the present invention provide, the available first terminal 100 of second terminal 200The image quality of the video image of generation, the video image is the first image quality, and then second terminal 200 can use image quality optimization modelBy the image quality of the video image from the first image quality optimization be the second image quality, wherein the image quality optimization model be to standard scene figureThe image quality corresponding relationship of picture carries out the model that machine learning obtains, and image quality corresponding relationship is the correspondence of the first image quality and the second image qualityRelationship, standard scene image are the image to match with the video scene of video image, and then second terminal 200 can be by image qualityVideo image for the second image quality is shown, so that the video image that second terminal 200 is shown is not readily susceptible to current networkThe influence of state, even if it is the first image quality that video image, which is the image quality that first terminal 100 is sent in the state of low network band width,Video image, second terminal 200 still be able to display image quality be better than first image quality video image so that such as remotely meetingIn the application scenarios such as view, the user of second terminal 200 can watch image quality preferably video image.
In one embodiment, a kind of processing method of video image is provided, is in one embodiment with reference to Fig. 2, Fig. 2The flow diagram of the processing method of video image is applied to be illustrated for the second terminal 200 in Fig. 1 in this way,The processing method of the video image may comprise steps of:
Step S101 obtains the video image that first terminal generates.
In this step, the video image of the available first terminal 100 of second terminal 200 generation.Wherein, first terminal100 video images that can be used for needing to share by user, and the transmitting terminal as video image, whole by network and secondAfter communication connection is established at end 200, which is sent to second terminal 200.Video image can be captured in real-time or recordThe video image of system, by taking the application scenarios of teleconference as an example, meeting presider can obtain meeting by first terminal 100 in real timeThe video image at scene is discussed, for first terminal 100 by the transmission of video images to second terminal 200, the quantity of second terminal 200 canBe it is multiple, then the available video image of each second terminal 200 carries out respective handling.
Step S102, using image quality optimization model by the image quality of video image from the first image quality optimization be the second image quality.
This step is mainly that second terminal 200 can use preconfigured image quality optimization model and produce first terminal 100The image quality of raw video image is the second image quality from the first image quality optimization.Wherein, image quality optimization model refers to standard scene figureThe image quality corresponding relationship of picture carries out the model that machine learning obtains, and image quality corresponding relationship refers to pair of the first image quality and the second image qualityIt should be related to, and standard scene image is the image to match with the video scene of video image.
Image quality optimization model, can be to the image quality pair of standard scene image before the image quality to video image optimizesIt should be related to carry out deep learning, i.e., image quality optimization model is to the standard under the standard scene image and the second image quality under the first image qualityThe corresponding relationship of scene image carries out deep learning, enables the image quality optimization model after training by the pattern field of the first image qualityScape image optimization is the standard scene image of the second image quality, and the video image that standard scene image and first terminal 100 are sentVideo scene match, the image quality of the video image is the first image quality, it is trained in this way after image quality optimization model also can be rightThe image quality of video image that first terminal 100 is sent optimizes, by the image quality of the video image from the first image quality optimization be theTwo image quality.
Wherein, video image has image content, and image content can reflect out the video scene of the video image, with remoteIt is illustrated for Cheng Huiyi, in the application scenarios of teleconference, the video image that first terminal 100 generates is meeting sceneVideo image, and image content is exactly the picture material shown in the video image, and such as live participant is in picturePosition, the article ornaments at meeting scene and the environment at meeting scene etc., which, which sees, can reflect out the video imageIt is to shoot or record under video scene, for teleconference, it will usually it is carried out in a relatively-stationary conference scenario,It is relatively fixed to refer to that in video image shooting or recording process too big variation, the view of teleconference occur for the conference scenarioThe video scene of frequency image is usually some meeting room, and the picture at meeting room scene is usually opposing stationary, therefore, can be with the meetingThe image of room is discussed as standard scene image, to the image quality of video image of the first terminal 100 during teleconference carries outBefore optimizing, the meeting room can be shot or recorded in the image of a variety of image quality as standard scene image, including firstThe standard scene image of image quality and the standard scene image of the second image quality, then can be excellent as image quality using neural network modelChange model to instruct come the corresponding relationship between the standard scene image of standard scene image and the second image quality to the first image qualityPractice, and video scene i.e. certain meeting room for the video image that standard scene image and first terminal 100 are sent matches, after trainingImage quality optimization model the video image that image quality is the first image quality can be optimized for the video image of the second image quality.
It is understood that if when being in an opposing stationary picture of the video image that first terminal 100 is sent,The video scene of the video image is also relatively fixed, in this way, second terminal 200 can be in the item for not increasing network bandwidth requirementIt is high-quality video figure by the low quality video image restoring that image quality optimization model sends over first terminal 100 under partPicture, even if so that first terminal 100 is influenced the video image of original high quality carrying out a degree of compression by networkThe second terminal 200 is transmitted again, which can be also reduced to the video image of high quality by second terminal 200.
Step S103, display image quality are the video image of the second image quality.
This step is mainly that the image quality of video image is by second terminal 200 from the first image quality optimization in image quality optimization modelAfter second image quality, the video image of available second image quality is simultaneously shown, so that the user of second terminal 200 can be withThe content of the video image is viewed by the second terminal 200, and the image quality of the video image passes through optimization, so that secondThe video image that terminal 200 is shown is not readily susceptible to the influence of current network state, shows that the video image of high quality is conducive toUser accurately grasps the information of video image, can also improve user to the perception of video image.
The processing method of above-mentioned video image, second terminal obtain the video image that first terminal generates, second terminal benefitWith image quality optimization model by the image quality of the video image from the first image quality optimization be the second image quality, the image quality optimization model be to markThe image quality corresponding relationship of quasi- scene image carries out the model that machine learning obtains, and image quality corresponding relationship is the first image quality and secondCorresponding relationship between image quality, standard scene image are the images to match with the video scene of the video image, and the image quality is excellentChanging model can be the second image quality from the first image quality optimization by video image, and then image quality optimization can be second by second terminalThe video image of image quality is shown, so that the video image that second terminal is shown is not readily susceptible to the influence of network state, i.e.,Make in the state of low network band width first terminal that the video image of the first image quality is sent to second terminal, second terminal is stillCan display image quality be better than the first image quality video image.
In one embodiment, video image can be first terminal for the image quality of video image from the second image quality boil down toThe video image that first image quality obtains.
In the present embodiment, for first terminal 100 before by transmission of video images to second terminal 200, first terminal 100 canFirst to compress the image quality of the video image, there are many kinds of the modes for compressing the image quality of the video image, such as reducing shouldThe resolution ratio of video image reduces frame per second etc., and the image quality of the video image is compressed to the first image quality from the second image quality, willImage quality is the transmission of video images of the first image quality to second terminal 200.The present embodiment first compresses the image quality of video imageIt transmits again, since the image quality of the video image can be the second original image quality by the first image quality optimization by second terminal 200, becauseThis this mode can ensure 200 display image quality of second terminal preferably the video image of the second image quality while, also reduceThe network bandwidth that the video image occupies in transmission process.
In one embodiment, which can be the video figure that first terminal transmits under low network band width statePicture.
In the present embodiment, first terminal 100 can be given the transmission of video images of generation in the state of low network band widthSecond terminal 200.Wherein, low network band width state refers to that current network bandwidth is in low-quality state, such as network flowThe states such as low speed, stability are poor, packet loss is big, under the low network band width state, first terminal 100 is difficult to such as fine definitionEtc. high quality transmission of video images to second terminal 200, so first terminal 100 can be by the lower video figure of image qualityAs being transferred to second terminal 200 in the state of low network band width, second terminal 200 restores the lower video image of the image qualityIt is shown for image quality preferably video image.Specifically, first terminal 100 first can will in the state of low network band widthThe image quality of video image is transferred to from preferably second the first image quality of image quality boil down to, then under current low network band width stateSecond terminal 200, so that second terminal 200 still is able to view more excellent image quality under the interactive scene of low network band widthLong-distance video image.
In one embodiment, the step of video image of acquisition first terminal generation may include:
Obtain the video image that server is sent；The server is used to receive the video image of first terminal transmission, and willThe video image is sent to corresponding receiving end.
In the present embodiment, second terminal 200 can obtain the video image that first terminal 100 generates by server.GinsengFig. 3 is examined, Fig. 3 is the signaling diagram of the processing method of video image in one embodiment, the view that first terminal 100 can be generatedFrequency image is sent to server 300, and server 300 can receive the video image of the first terminal 100 transmission, then server300 can be sent to the video image corresponding receiving end, and receiving end refers to the terminal for receiving the video image, and connectsReceive the terminal of the video image quantity may include it is multiple, which can be sent to specified by first terminal 100Video image is such as sent to second terminal 200 by receiving end, and first terminal 100 can be by the video image and second terminal200 terminal iidentification is sent to server 200, and video image can be transmitted to second according to the terminal iidentification by server 200Terminal 200, so that second terminal 200 can receive the video image and optimize to the image quality of the video image, by image qualityVideo image after optimization is shown.The scheme of the present embodiment can apply the application in multiple transmitting terminals to multiple receiving endsIn scene, server 300 can be realized the scheduling to each end stream medium data, the video image that transmitting terminal is sentIt is forwarded to corresponding receiving end by server 300, image quality optimization processing is carried out to video image on the receive side, reduces transmissionThe network bandwidth that video image occupies.
In one embodiment, after the step of obtaining the video image that first terminal generates, can also include:
Obtain the video scene of video image；The determining standard scene image to match with video scene；It is more from what is prestoredImage quality optimization model corresponding with standard scene image is extracted in a image quality optimization model.
In the present embodiment, second terminal 200 can be after the video image for receiving the generation of first terminal 100, from the viewVideo scene is extracted in the video pictures of frequency image, after determining the video scene of the video image, the available and videoThe standard scene image that scene matches.By taking the application scenarios of teleconference as an example, if the video image is in the field of meeting room AThe video image of real-time recording under scape, then after receiving the video image, the available video scene is second terminal 200Then the scene of meeting room A can obtain the standard scene image to match with the scene of meeting room A from database, shouldStandard scene image can be the image of the meeting room A shot in advance, due to video scene may include it is a variety of, such as in differenceThe video scene for the video image recorded in meeting room is different, so standard scene image also may include a variety of, various standardsScene image matches from different video scenes respectively, as standard scene image A corresponds to meeting room A, standard scene image BCorrespond to meeting room C etc. corresponding to meeting room B and standard scene image C.
Second terminal 200 can be by multiple image quality optimization models respectively to the corresponding pass of the image quality of different standard scene imagesSystem is trained, and enables multiple image quality optimization models after training that the video image progress image quality of different video scene is excellentChange.Then, video image, and the pattern field that the video scene for being determined at the video image matches are received in second terminal 200After scape image, image quality corresponding with the standard scene image can be extracted from preparatory trained multiple image quality optimization modelsOptimized model to carry out image quality optimization to the video image, thus realize to the image quality of the video image under different video scene intoRow optimization processing.
In one embodiment, can with comprising steps of
Obtain history video image；The history video image is the figure that first terminal is sent under high network bandwidth statePicture；The image quality of the history video image is not less than the second image quality；The video scene of history video image and the video of video imageScene is identical；Standard scene image is obtained according to history video image.
The present embodiment is mainly that second terminal 200 is gone through according to what first terminal 100 was sent in the state of high network bandwidthHistory video image obtains standard scene image.Wherein, first terminal 100 can be under the network state of high network bandwidth, will be highThe video image of quality is sent to second terminal 200, and second terminal 200 can such as be stored every 10 frames with certain frequency and be receivedThe video image of the high quality arrived, then second terminal 200 obtains standard scene image according to the video image of the high quality, theTwo terminals 200 can be using the video image of all high quality received as standard video image, and by these normal videosImage as image quality optimization model training data training the image quality optimization model, enable training after image quality optimization modelLow-quality video image is reduced to the video image of high quality.
Specifically, first terminal 100 can be by the transmission of video images of high quality under the network state of high network bandwidthIt is shown to second terminal 200, however when network state becomes low network band width from high network bandwidth, first terminal 100It can be that low-quality video image is transmitted with adapting to current network state by the video image compression of original high quality, thisWhen, the video image that second terminal 200 can send first terminal 100 under the network state of high network bandwidth is as historyVideo image, the image quality of the history video image are not less than the second image quality, and the video scene of the history video image and theThe video scene for the video image that one terminal 100 is sent is identical, that is to say, that the video for the video image that first terminal 100 is sentScene is identical before and after network state becomes low network band width from high network bandwidth, and second terminal 200 can be by history videoImage is as standard scene image.In this manner it is possible to first first terminal 100 is sent under the network state of high network bandwidthHigh-quality video image is stored as history video image, and using the history video image as standard scene image to image quality optimizationModel is trained, so that when network state becomes low network band width, using the image quality optimization model by the view of the first image qualityFrequency image restoring is the second image quality, finally renders by the picture of the video image of the second image quality after reduction again, to guaranteeThe image quality for the video image that user watches in second terminal 200 is not readily susceptible to the influence of current network state fluctuation.
In one embodiment, using image quality optimization model by the image quality of video image from the step of being optimized for the second image qualityMay include:
For image quality optimization model, optimization threshold value corresponding with the second image quality is set；The image quality optimization model can be using volumeProduct neural network model；Video image is inputted into image quality optimization model, triggers the image quality optimization model according to optimization threshold value for theOne image quality optimization is the second image quality.
Mainly optimization threshold value corresponding with the second image quality can be arranged for image quality optimization model in the present embodiment, so that drawingMatter Optimized model can according to the optimization threshold value by the image quality of video image from the first image quality optimization be the second image quality.Wherein, excellentChange threshold value and refer to the parameter for the target image quality optimization model to the degree of optimization of image quality, for different optimization threshold values, drawsMatter Optimized model is different to the degree of optimization of the image quality of video image, and in general, optimization threshold value is higher, to the picture of video imageThe degree of optimization of matter is higher, which can correspond to the standard scene figure of the video image after image quality optimization and high qualitySimilarity between picture, optimization threshold value it is higher, the standard scene image of video image and high quality after image quality optimization itBetween similarity it is also higher.
In the present embodiment, optimization threshold corresponding with the second image quality can be arranged for image quality optimization model in second terminal 200Value, then the video image that image quality is the first image quality is input to image quality optimization model by second terminal 200, so that the image quality optimizationModel exports the video image of the second image quality according to the optimization threshold value.It is understood that when second terminal 200 needs to export notWhen the image quality of ad eundem, different optimization threshold values can be correspondingly set, allow image quality optimization model according to the excellent of settingChange the video image of threshold value output respective level image quality.
Convolutional neural networks model can passed through to video using convolutional neural networks model as image quality optimization modelIt can be multiple picture units by video image cutting, such as by video image cutting be more when the image quality of image optimizesEach picture unit is input in convolutional neural networks model and carries out image quality optimization, so by the picture unit of a 16 × 9 pixelThe picture unit after each image quality optimization is stitched together again afterwards, forms the video image of high quality.By by video image intoRow cutting can be improved the efficiency of image quality reduction to the mode that each picture unit carries out image quality optimization.
In one embodiment, a kind of processing unit of video image is provided, is in one embodiment with reference to Fig. 4, Fig. 4The processing unit of the structural block diagram of the processing unit of video image, the video image may include:
Module 101 is obtained, for obtaining the video image of first terminal generation；
Optimization module 102, for using image quality optimization model by the image quality of video image from the first image quality optimization be secondImage quality；Wherein, image quality optimization model is to carry out the model that machine learning obtains to the image quality corresponding relationship of standard scene image；It drawsMatter corresponding relationship is the corresponding relationship of the first image quality and the second image quality；Standard scene image is the video scene phase with video imageMatched image；
Display module 103 is the video image of the second image quality for display image quality.
In one embodiment, video image can be first terminal for the image quality of video image from the second image quality boil down toThe video image that first image quality obtains.
In one embodiment, video image can be the video figure that first terminal transmits under low network band width statePicture.
In one embodiment, module 101 is obtained to be further used for:
Obtain the video image that server is sent；The server is used to receive the video image of first terminal transmission, and willThe video image is sent to corresponding receiving end.
In one embodiment, can also include:
Scene acquiring unit, for obtaining the video scene of video image；
Scene determination unit, for the determining standard scene image to match with video scene；
Model extraction unit, it is corresponding with standard scene image for being extracted from the multiple image quality optimization models prestoredImage quality optimization model.
In one embodiment, can also include:
First image acquisition unit, for obtaining history video image；The history video image is first terminal in high netThe image sent under network bandwidth status；The image quality of the history video image is not less than the second image quality；The video of history video imageScene is identical as the video scene of video image；
Second image acquisition unit, for obtaining standard scene image according to history video image.
In one embodiment, optimization module 102 is further used for:
For image quality optimization model, optimization threshold value corresponding with the second image quality is set；The image quality optimization model can be using volumeProduct neural network model；Video image is inputted into image quality optimization model, triggers the image quality optimization model according to optimization threshold value for theOne image quality optimization is the second image quality.
The processing unit of video image of the invention and the processing method of video image of the invention correspond, about viewThe specific of the processing unit of frequency image limits the restriction that may refer to the processing method above for video image, in above-mentioned viewTechnical characteristic and its advantages that the embodiment of the processing method of frequency image illustrates are suitable for the processing unit of video imageEmbodiment in, details are not described herein.Modules in the processing unit of above-mentioned video image can be fully or partially through softPart, hardware and combinations thereof are realized.Above-mentioned each module can be embedded in the form of hardware or independently of the processing in computer equipmentIt in device, can also be stored in a software form in the memory in computer equipment, in order to which processor calls execution above eachThe corresponding operation of a module.
In one embodiment, a kind of remote video system is provided, is one embodiment medium-long range with reference to Fig. 5, Fig. 5The structural block diagram of video system, the remote video system may include: transmitting terminal 400, server 300 and receiving end 500；
Transmitting terminal 400, for video image to be sent to server 300；Server 300, for forwarding video imageTo receiving end 500；Receiving end 500, for receiving the video image of the transmission of server 300, using image quality optimization model by videoThe image quality of image is the second image quality from the first image quality optimization, and the video image of the second image quality is shown；Wherein, image quality optimizationModel is to carry out the model that machine learning obtains to the image quality corresponding relationship of standard scene image；Image quality corresponding relationship is the first pictureThe corresponding relationship of matter and the second image quality；Standard scene image is the image to match with the video scene of video image.
In the present embodiment, video image can be sent to server 300, server 300 by the transmitting terminal 400 of video imageAfter receiving the video image, which is forwarded to corresponding receiving end 500, receiving end 500 can use image qualityOptimized model by the image quality of received video image from the first image quality optimization be the second image quality, and by the video figure of second image qualityAs being shown.Wherein, the quantity of transmitting terminal 400 and receiving end 500 can be multiple, and server 300 can be used to implement notReceiving end 2, receiving end are forwarded to the stream media scheduling between transmitting terminal and receiving end, such as by the video image of transmitting terminal 1M, the video image of transmitting terminal 2 is forwarded to receiving end 1 etc., the program can be applied in multiple transmitting terminals to multiple receiving endsApplication scenarios in, server 300 can be realized the scheduling to each end stream medium data so that transmitting terminal send video figureAs corresponding receiving end can be forwarded to by server 300, image quality optimization processing is carried out to video image on the receive side, is subtractedThe network bandwidth that few transmitting video image occupies.
In one embodiment, transmitting terminal is further used for drawing the image quality of video image from the second image quality boil down to firstThe video image that image quality is the first image quality is sent to receiving end by server under low network band width state by matter.
In the present embodiment, when network state is in low network band width state, transmitting terminal 400 can be by the picture of video imageThen the video image of first image quality is sent to server 300, server 300 from second the first image quality of image quality boil down to by matterThe video image of first image quality is transmitted to receiving end 500 again, so that receiving end 500 optimizes the video image of the first image qualityVideo image for the second image quality is shown, so that receiving end 500 still is able under the interactive scene of low network band widthView the long-distance video image of more excellent image quality.
In one embodiment, transmitting terminal 400 is also used to that history video image is passed through clothes under high network bandwidth stateBusiness device 300 is sent to receiving end 500；The image quality of history video image is not less than the second image quality；The video field of history video imageScape is identical as the video scene of video image；Receiving end 500 is also used to receive history video image, is obtained according to history video imageTake standard scene image.
In the present embodiment, when network state is in high network bandwidth state, transmitting terminal 400 can be by the video of high qualityImage is sent to server 300, and the video image of the high quality is transmitted to receiving end 500 by server 300, and receiving end 500 connectsIt receives the video image of the high quality and is stored as history video image, the image quality of the history video image is not less than secondImage quality, and the video scene of the history video image is identical as the video scene of video image, in this way, receiving end 500 can be withIt is trained using the history video image as the training data of image quality optimization model, when network state is from high network bandwidth stateWhen becoming low network band width state, receiving end 500 can use that trained image quality optimization model sends transmitting terminal 400The video image of one image quality is optimized to the video image of the second image quality, and the video image of second image quality is shown, withGuarantee that the image quality for the video image that user watches on receiving end 500 is not readily susceptible to the influence of current network state fluctuation.
In one embodiment, a kind of computer equipment is provided, which can be terminal, internal structureFigure can using as shown in fig. 6, Fig. 6 as the internal structure chart of computer equipment in one embodiment.The computer equipment includes passing throughProcessor, memory, network interface, display screen and the input unit of system bus connection.Wherein, the processing of the computer equipmentDevice is for providing calculating and control ability.The memory of the computer equipment includes non-volatile memory medium, built-in storage.It shouldNon-volatile memory medium is stored with operating system and computer program.The built-in storage is the behaviour in non-volatile memory mediumThe operation for making system and computer program provides environment.The network interface of the computer equipment is used to pass through net with external terminalNetwork connection communication.A kind of processing method of video image is realized when the computer program is executed by processor.The computer is setStandby display screen can be liquid crystal display or electric ink display screen, and the input unit of the computer equipment can be displayThe touch layer covered on screen is also possible to the key being arranged on computer equipment shell, trace ball or Trackpad, can also be outerKeyboard, Trackpad or mouse for connecing etc..
It will be understood by those skilled in the art that structure shown in Fig. 6, only part relevant to the present invention program is tiedThe block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to the present invention program, specific computer equipmentIt may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment, including processor and memory, the memory storage are providedThere is computer program, the processor performs the steps of when executing the computer program
Obtain the video image that first terminal generates；Using image quality optimization model by the image quality of video image from the first image qualityIt is optimized for the second image quality；Wherein, which is to carry out machine learning to the image quality corresponding relationship of standard scene imageObtained model；Image quality corresponding relationship is the corresponding relationship of the first image quality and the second image quality；Standard scene image be and video figureThe image that the video scene of picture matches；Display image quality is the video image of the second image quality.
In one embodiment, video image can be first terminal for the image quality of video image from the second image quality boil down toThe video image that first image quality obtains.
In one embodiment, video image can be the video figure that first terminal transmits under low network band width statePicture.
In one embodiment, it is also performed the steps of when processor executes computer program and obtains what server was sentVideo image；The server is used to receive the video image of first terminal transmission, and the video image is sent to and is connect accordinglyReceiving end.
In one embodiment, the view for obtaining video image is also performed the steps of when processor executes computer programFrequency scene；The determining standard scene image to match with video scene；It extracts and marks from the multiple image quality optimization models prestoredThe corresponding image quality optimization model of quasi- scene image.
In one embodiment, it is also performed the steps of when processor executes computer program and obtains history video image；The history video image is the image that first terminal is sent under high network bandwidth state；The image quality of the history video image is not lowIn the second image quality；The video scene of history video image and the video scene of video image are identical；It is obtained according to history video imageTake standard scene image.
In one embodiment, it also performs the steps of when processor executes computer program and is set for image quality optimization modelSet optimization threshold value corresponding with the second image quality；The image quality optimization model can use convolutional neural networks model；By video figureAs input image quality optimization model, triggering the image quality optimization model according to optimization threshold value is the second image quality by the first image quality optimization.
Above-mentioned computer equipment, by the computer program run on the processor, so that the view that second terminal is shownFrequency image is not readily susceptible to the influence of network state, even if first terminal is by the view of the first image quality in the state of low network band widthFrequency image is sent to second terminal, and second terminal still is able to the video image that display image quality is better than the first image quality.
Those of ordinary skill in the art will appreciate that realizing the processing side of video image described in as above any one embodimentAll or part of the process in method is relevant hardware can be instructed to complete by computer program, the computerProgram can be stored in a non-volatile computer read/write memory medium, and the computer program is when being executed, it may include as aboveState the process of the embodiment of each method.Wherein, to memory, storage, number used in each embodiment provided by the present inventionAccording to any reference of library or other media, non-volatile and/or volatile memory may each comprise.Nonvolatile memory can wrapInclude read-only memory (ROM), programming ROM (PROM), electrically programmable ROM (EPROM), electrically erasable ROM(EEPROM) or flash memory.Volatile memory may include random access memory (RAM) or external cache.MakeTo illustrate rather than limit to, RAM is available in many forms, such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram(SDRAM), double data rate sdram (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronization link (Synchlink) DRAM(SLDRAM), memory bus (Rambus) directly RAM (RDRAM), direct memory bus dynamic ram (DRDRAM) andMemory bus dynamic ram (RDRAM) etc..
Accordingly, a kind of computer readable storage medium is provided in one embodiment, is stored thereon with computer program,It is performed the steps of when computer program is executed by processor
Obtain the video image that first terminal generates；Using image quality optimization model by the image quality of video image from the first image qualityIt is optimized for the second image quality；Wherein, which is to carry out machine learning to the image quality corresponding relationship of standard scene imageObtained model；Image quality corresponding relationship is the corresponding relationship of the first image quality and the second image quality；Standard scene image be and video figureThe image that the video scene of picture matches；Display image quality is the video image of the second image quality.
In one embodiment, video image can be first terminal for the image quality of video image from the second image quality boil down toThe video image that first image quality obtains.
In one embodiment, video image can be the video figure that first terminal transmits under low network band width statePicture.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains server transmissionVideo image；The server is used to receive the video image of first terminal transmission, and the video image is sent to accordinglyReceiving end.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains video imageVideo scene；The determining standard scene image to match with video scene；From the multiple image quality optimization models prestored extract withThe corresponding image quality optimization model of standard scene image.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains history video figurePicture；The history video image is the image that first terminal is sent under high network bandwidth state；The image quality of the history video imageNot less than the second image quality；The video scene of history video image and the video scene of video image are identical；According to history video figureAs obtaining standard scene image.
In one embodiment, it is also performed the steps of when computer program is executed by processor as image quality optimization modelOptimization threshold value corresponding with the second image quality is set；The image quality optimization model can use convolutional neural networks model；By videoImage inputs image quality optimization model, and triggering the image quality optimization model according to optimization threshold value is the second image quality by the first image quality optimization.
Above-mentioned computer readable storage medium, by the computer program that it is stored, so that the video that second terminal is shownImage is not readily susceptible to the influence of network state, even if first terminal is by the video of the first image quality in the state of low network band widthImage is sent to second terminal, and second terminal still is able to the video image that display image quality is better than the first image quality.
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodimentIn each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lanceShield all should be considered as described in this specification.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneouslyIt cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the artIt says, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to protection of the inventionRange.Therefore, the scope of protection of the patent of the invention shall be subject to the appended claims.