CN116471262A - Video quality evaluation method, apparatus, device, storage medium, and program product - Google Patents

Abstract

The embodiment of the application provides a video quality assessment method, a device, equipment, a storage medium and a program product, wherein the method comprises the following steps: acquiring an image frame sequence and video parameters of a video to be evaluated; the image frame sequences are respectively input into a motion estimation module, a subjective image quality module, a content attraction module and a scene module to obtain a motion estimation score, a subjective image quality score, a content attraction score and a scene score; inputting the video parameters to an objective parameter module to obtain objective parameter scores; and calculating to obtain the video quality score of the video to be evaluated based on the motion estimation score, the subjective image quality score, the content attraction score, the scene score and the objective parameter score. The scheme fuses the motion information of the video, and combines the subjective preferences of the video scene and the audience tightly, so that the accuracy of the finally obtained video quality evaluation result is higher, and the application range is wider.

Description

Video quality evaluation method, device, equipment, storage medium and program product

technical field

The embodiments of the present application relate to the field of computer technologies, and in particular, to a video quality assessment method, device, equipment, storage medium, and program product.

Background technique

With the continuous development of real-time video streaming technology, live video streaming has become an indispensable part of people's daily life. Whether it is video conferencing, online streaming, or live TV, high-quality video transmission is required. However, there are many problems with real-time video streaming that affect user experience, such as video quality degradation, increased latency, buffering, etc.

Typically, the use of advanced encoding standards can reduce the video bit rate while maintaining objective visual quality, bringing a smoother visual experience to low-bandwidth viewers. However, existing encoders cannot fully meet the objective requirements of users. On the one hand, real-time streaming media requires the encoder to have a high encoding speed, which will affect the encoding quality. On the other hand, the adjustment of encoding parameters (such as frame rate, bit rate) does not have a linear relationship with the viewer's subjective experience. For example, when the video already has a high frame rate and bit rate, further increasing the encoding parameters does not significantly improve the viewer's subjective experience. On the contrary, when the video is at a low frame rate and bit rate, appropriately increasing the frame rate will bring a better viewing experience than increasing the bit rate. Therefore, an accurate and efficient video quality evaluation method is needed to accurately evaluate the video quality for subsequent parameter decision-making to better meet user needs.

Contents of the invention

Embodiments of the present application provide a video quality assessment method, device, device, storage medium, and program product, which integrate video motion information and closely combine video scenes and subjective preferences of viewers, so that the final video quality assessment results are more accurate and have a wider range of applications.

In the first aspect, the embodiment of the present application provides a video quality assessment method, the method comprising:

Obtain the image frame sequence and video parameters of the video to be evaluated;

Inputting the image frame sequence into the motion estimation module, the subjective image quality module, the content attractiveness module and the scene module respectively, and obtaining the motion estimation score, the subjective image quality score, the content attractiveness score and the scene score, and the content attractiveness module is obtained based on sample training after data layering processing;

Inputting the video parameters into the objective parameter module to obtain the objective parameter score;

A video quality score of the video to be evaluated is calculated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and the objective parameter score.

In the second aspect, the embodiment of the present application also provides a video quality evaluation device, including:

A data acquisition unit configured to acquire an image frame sequence and video parameters of a video to be evaluated;

The data fusion processing unit is configured to input the sequence of image frames into the motion estimation module, the subjective image quality module, the content attractiveness module and the scene module respectively, to obtain the motion estimation score, the subjective image quality score, the content attractiveness score and the scene score, and the content attractiveness module is obtained based on the sample training after data layering processing; and, the video parameters are input into the objective parameter module to obtain the objective parameter score;

A score calculation unit configured to calculate a video quality score of the video to be evaluated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score, and the objective parameter score.

In the third aspect, the embodiment of the present application also provides a video quality evaluation device, the device includes:

one or more processors;

storage means for storing one or more programs,

When the one or more programs are executed by the one or more processors, the one or more processors are made to implement the video quality assessment method described in the embodiment of the present application.

In a fourth aspect, the embodiment of the present application further provides a non-volatile storage medium storing computer-executable instructions, and the computer-executable instructions are used to execute the video quality assessment method described in the embodiment of the present application when executed by a computer processor.

In the fifth aspect, the embodiment of the present application further provides a computer program product, the computer program product includes a computer program, the computer program is stored in a computer-readable storage medium, at least one processor of the device reads and executes the computer program from the computer-readable storage medium, so that the device executes the video quality assessment method described in the embodiment of the present application.

In the embodiment of the present application, by obtaining the image frame sequence and video parameters of the video to be evaluated, the image frame sequence is respectively input into the motion estimation module, the subjective image quality module, the content attractiveness module and the scene module to obtain the motion estimation score, the subjective image quality score, the content attractiveness score and the scene score. The video quality score of the video to be evaluated. This video quality evaluation method integrates motion estimation, subjective image quality, content attractiveness, video scene and objective video parameters, and closely combines the video scene and the subjective preferences of the audience. The resulting video quality evaluation results are more accurate and can be applied in a wider range. Among them, the training data of the content attractiveness module uses samples after data layering processing, and different content attractiveness scores under the same subjective image quality can be obtained to significantly improve the accuracy of subsequent video quality evaluation.

Description of drawings

Fig. 1 is the flowchart of a kind of video quality evaluation method that the embodiment of the present application provides;

FIG. 2 is a flow chart of a method for obtaining a motion estimation score through a motion estimation module provided in an embodiment of the present application;

FIG. 3 is a flow chart of a method for training a content attractiveness module provided by an embodiment of the present application;

FIG. 4 is a flow chart of a method for video frame rate decision-making provided in an embodiment of the present application;

FIG. 5 is a structural block diagram of a video quality assessment device provided in an embodiment of the present application;

FIG. 6 is a schematic structural diagram of a video quality assessment device provided by an embodiment of the present application.

Detailed ways

The embodiments of the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It can be understood that the specific embodiments described here are only used to explain the embodiments of the present application, but not to limit the embodiments of the present application. In addition, it should be noted that, for the convenience of description, only a part but not all structures related to the embodiment of the present application are shown in the drawings.

The terms "first", "second" and the like in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific sequence or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second" and so on are generally of one type, and the number of objects is not limited. For example, there can be one or more first objects. In addition, "and/or" in the specification and claims means at least one of the connected objects, and the character "/" generally means that the related objects are an "or" relationship.

The video quality evaluation method provided in the embodiment of the present application can be applied to a live broadcast scene, such as evaluating the quality of a live video to obtain a video quality score accordingly. Or, any other scenario where video quality needs to be evaluated.

Fig. 1 is the flow chart of a kind of video quality evaluation method that the embodiment of the present application provides, as shown in Fig. 1, specifically comprises the following steps:

Step S101. Obtain an image frame sequence and video parameters of a video to be evaluated.

Wherein, the video to be evaluated may be an acquired live streaming video, for example, a video acquired when it is necessary to evaluate the quality of the live image content during the live broadcast. The subject of execution may be a server, that is, the server evaluates the video quality to obtain an evaluation result, so as to provide auxiliary decision-making for adjusting video parameters.

Wherein, the image frame sequence is a sequence composed of multiple image frames. Optionally, a sequence may be composed of consecutive multiple frames of images obtained by sampling the video to be evaluated. The video parameter represents an objective video quality parameter of the video to be evaluated, for example, it may be a frame rate, a bit rate, a quantization parameter value, a resolution, and the like.

Step S102, input the sequence of image frames into motion estimation module, subjective image quality module, content attractiveness module and scene module respectively, and obtain motion estimation score, subjective image quality score, content attractiveness score and scene score.

Among them, the motion estimation module, subjective image quality module, content attractiveness module and scene module are pre-trained neural network modules, which can output corresponding category scores for the input image frame sequence. For example, the image frame sequence is input to the motion estimation module to output the motion estimation score; the image frame sequence is input to the subjective image quality module to output the subjective image quality score; the image frame sequence is input to the content attractiveness module to output the content attractiveness score; the image frame sequence is input to the scene module to output the scene score. Among them, during the training process of the subjective image quality module and the content attractiveness module, a large number of labeled data from different regions and countries are used to reflect the preferences of people in different regions.

Among them, the motion estimation module is based on the optical flow model, which will have a negative impact on user experience when motion is perceived as discontinuous or undulating. Optical flow models accurately capture motion by mapping the displacement of each pixel from one image to another.

In one embodiment, as shown in FIG. 2, FIG. 2 is a flowchart of a method for obtaining a motion estimation score through a motion estimation module provided in an embodiment of the present application. As shown in the figure, it includes:

Step S1021, inputting two consecutive frames of images in the image frame sequence as a group to the motion estimation module.

In one embodiment, for each frame image in the image frame sequence, two consecutive frames of images are input as a group to the motion estimation module, so as to output the relative displacement of the pixels corresponding to each group of images.

Step S1022, calculate the relative displacement of the pixels in each group of frame images through the motion estimation module, and calculate the motion estimation score based on the relative displacement of the pixels.

Wherein, the relative displacement of pixels includes displacement in the horizontal direction and displacement in the vertical direction, and for an image frame sequence including N frames of images, it outputs a motion evaluation result of the relative displacement of N/2 pixels. Exemplarily, the recorded content of each motion evaluation result is the relative displacement of each pixel of one frame of images in two consecutive frames of images relative to the other frame of images. After the relative displacement of each pixel is obtained, the average value of the relative displacements of all pixels in the image can be calculated, that is, the average amplitude to characterize the motion situation to obtain the corresponding motion estimation score. Exemplarily, the higher the average amplitude, the more obvious the motion situation, and the higher the corresponding motion estimation score, and vice versa.

In another embodiment, when the motion estimation score is calculated by the motion estimation module, the obtained relative displacement of the pixel is further superimposed on the three-channel color value of the pixel, and then the obtained features are input into the motion estimation module to obtain the motion estimation score. Exemplarily, the relative displacement of the pixel point is represented by a two-dimensional vector, the three-channel color value of the pixel point is represented by a three-dimensional vector, and the result after the two are superimposed is represented by a five-dimensional vector. Optionally, each dimension represents the horizontal displacement, the vertical displacement, the R component color value, the G component color value, and the B component color value of the pixel point in turn, and the superimposed result is input to the motion estimation module as an image feature to obtain a motion estimation score of the entire image.

Wherein, the subjective image quality module is used to capture the image quality of each frame in the image frame sequence. Optionally, it uses an image quality assessment model with a feature extraction head for regression output. Use supervised learning techniques to train on large datasets including images of variations in brightness, contrast, saturation, noise, distortion, scene, and object type. A quality score for person labels is associated with each image in the dataset to reflect the subjective quality of the image. Among them, the model used by the subjective image quality module is different from the objective quality model. It mainly uses the neural network to fit people's subjective perception of image blur, noise, frame loss, etc., not the objective image quality.

Wherein, the content attractiveness module is obtained based on sample training after data layering processing. Optionally, FIG. 3 is a flow chart of a content attractiveness module training method provided in the embodiment of the present application, as shown in FIG. 3 , including:

Step S1023 , based on the scoring values of the sample data by the motion estimation module, the subjective image quality module, the scene module and the objective parameter module, perform data layering processing on the sample data to obtain layered samples.

In one embodiment, when performing sample layering processing, the trained motion estimation module, subjective image quality module, scene module, and objective parameter module are used to score the sample data to obtain corresponding scoring values, and then perform data layering processing on the sample data according to the scoring points to obtain layered samples. It should be noted that in data layering processing, any two or three of the above-mentioned motion estimation module, subjective image quality module, scene module, and objective parameter module can also be used to perform corresponding scoring, and layered processing of sample data is performed based on the scoring value to obtain layered samples.

Optionally, after the used module scores the sample data to obtain a scoring value, the sample data with the same scoring value or a score difference smaller than a preset value can be formed into a stratified sample. For example, the scoring result for each sample image in the sample data can be exemplarily recorded as [a, b, c, d], where a is the scoring value of the motion estimation module, b is the scoring value of the subjective image quality module, c is the scoring value of the scene module, and d is the scoring value of the objective parameter module. Exemplarily, the value range of each scoring value is [0, 1] after being normalized by the corresponding module. In the layering process, images with the same scoring value in the result can be used as a layer of samples, or sample data whose score difference is smaller than a preset value can be used to form a layer of samples, and the preset value can be 0.1 for example.

Step S1024, train the content attractiveness module based on hierarchical samples.

In one embodiment, the data is layered through other four modules (motion estimation module, subjective image quality module, scene module, and objective parameter module). The data in each layer has similar motion estimation, subjective image quality, scene, and objective parameter scores, but they represent different content. For the sample data in each layer, the labeling teams of different countries are scored as regional attractiveness labels, and then the neural network method is used for regression training.

In one embodiment, the output result of the scene module represents the category scene of the image, such as a game live broadcast scene, a personal show live broadcast scene, a PK live broadcast scene, and the like. Exemplarily, 8 different live broadcast scenes can be set in advance, and the output result is an 8-dimensional vector, each dimension corresponds to a category, the category represented by the dimension whose vector value is 1 is the scene category, and the category represented by the dimension whose vector value is 0 is the non-recognized category.

Step S103, input the video parameters into the objective parameter module to obtain the objective parameter score.

Wherein, the video parameter may be frame rate, bit rate, quantization parameter value, resolution, etc., and the objective quality parameter of the video is input to the objective parameter module to obtain the objective parameter score. Exemplarily, it is assumed that the input video parameters include three types of frame rate, bit rate, and quantization parameter value, and the corresponding output result is a score corresponding to each parameter, so as to jointly form an objective parameter score. Among them, the specific score falls within the range of [0, 1] after normalization.

Step S104: Calculate and obtain the video quality score of the video to be evaluated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and the objective parameter score.

In one embodiment, after the motion estimation score, subjective image quality score, content attractiveness score, scene score and objective parameter score are obtained, fusion calculation is performed to obtain the video quality score of the video to be evaluated. Optionally, the motion estimation score, subjective image quality score, content attractiveness score, scene score and objective parameter score may be multiplied by their corresponding weights, and the product results may be added to obtain the video quality score of the video to be evaluated. Wherein, the size of the weight value is obtained through neural network training and optimization. In another embodiment, other calculation formulas can also be set, and the obtained motion estimation score, subjective image quality score, content attractiveness score, scene score and objective parameter score are brought into the formula to calculate the video quality score of the video to be evaluated.

It can be seen from the above that by obtaining the image frame sequence and video parameters of the video to be evaluated, the image frame sequence is respectively input into the motion estimation module, subjective image quality module, content attractiveness module and scene module to obtain the motion estimation score, subjective image quality score, content attractiveness score and scene score. The video quality score of the video to be evaluated. This video quality evaluation method integrates motion estimation, subjective image quality, content attractiveness, video scene and objective video parameters, and closely combines the video scene and the subjective preferences of the audience. The resulting video quality evaluation results are more accurate and can be applied in a wider range. Among them, the training data of the content attractiveness module uses samples after data layering processing, and different content attractiveness scores under the same subjective image quality can be obtained to significantly improve the accuracy of subsequent video quality evaluation.

Fig. 4 is a flow chart of a method for video frame rate decision-making provided in the embodiment of the present application, as shown in Fig. 4 , including:

Step S201 , adjust the frame rate of the video to be evaluated to obtain videos to be evaluated with different frame rates, and input the videos to be evaluated with different frame rates to the objective parameter module to obtain multiple objective parameter scores.

In one embodiment, the video parameters of the same video stream/video content can be adjusted. Taking the frame rate as an example, videos to be evaluated with different frame rates can be adjusted and input to the objective parameter module to obtain multiple objective parameter scores.

Step S202 , based on the motion estimation score, subjective image quality score, content attractiveness score, scene score and each objective parameter score, the video quality scores of the videos to be evaluated at different frame rates are calculated.

Among them, since the same video frame sequence is aimed at, its motion estimation score, subjective image quality score, content attractiveness score, and scene score remain unchanged, and the value is obtained following the scheme described in the previous example, and only different objective parameter scores are used as variables for fusion calculation to obtain the video quality score of videos to be evaluated at different frame rates.

Step S203, based on multiple video quality scores corresponding to different frame rates, make a video frame rate decision.

In one embodiment, multiple video quality scores corresponding to different frame rates are obtained, and when making a video frame rate decision, the video frame rate decision may be made according to changes in video quality scores at different frame rates. Optionally, when the video frame rate is increased and the video quality score is significantly increased, the video frame rate is increased; when the video frame rate is decreased and the video quality score is not significantly decreased, the video frame rate is decreased. Exemplarily, taking the original video frame rate to be evaluated at 15fps as an example, by inputting 9, 12, 15, 18, 21, and 24 to the objective parameter module to observe the relative benefits when the frame rate changes, if the frame rate is 15-18fps, the final video quality score is greatly improved, and the visual experience is optimized by interpolating frames. If at 12-15fps, the resulting final video quality score is less degraded, then frame extraction is used to reduce the bit rate and bandwidth cost.

It should be noted that the frame rate is used as the video quality parameter above to explain its decision-making, and the same decision-making on the bit rate and quantization parameters, etc. can also be made based on the same idea and design logic.

It can be known from the above that by adjusting the frame rate of the video to be evaluated, the video to be evaluated at different frame rates is obtained, and the video to be evaluated at different frame rates is input to the objective parameter module to obtain a plurality of objective parameter scores, and the video quality scores of the video to be evaluated at different frame rates are calculated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and each objective parameter score, and the video frame rate decision is made based on the corresponding video quality scores at multiple different frame rates, which combines the evaluation of subjective video quality to adjust video parameters Decision-making is not simply a mechanical adjustment based on objective factors and video parameters. It can significantly improve the user's viewing experience while balancing bandwidth and device computing power.

FIG. 5 is a structural block diagram of a video quality assessment device provided by an embodiment of the present application. As shown in FIG. 5 , the device is used to implement the video quality assessment method provided by the above embodiment, and has corresponding functional modules and beneficial effects for executing the method. As shown in Figure 5, the device specifically includes: a data acquisition unit 101, a data fusion processing unit 102, and a score calculation unit 103, wherein the data fusion processing unit 102 includes a motion estimation module 1021, a subjective image quality module 1022, a content attractiveness module 1023, a scene module 1024 and an objective parameter module 1025,

The data acquisition unit 101 is configured to acquire an image frame sequence and video parameters of the video to be evaluated;

The data fusion processing unit 102 is configured to input the sequence of image frames into the motion estimation module 1021, the subjective image quality module 1022, the content attractiveness module 1023, and the scene module 1024, respectively, to obtain motion estimation scores, subjective image quality scores, content attractiveness scores, and scene scores, and the content attractiveness module is obtained based on sample training after data layering processing; and, input the video parameters into the objective parameter module 1025 to obtain objective parameter scores;

The score calculation unit 103 is configured to calculate the video quality score of the video to be evaluated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and the objective parameter score.

It can be seen from the above that by obtaining the image frame sequence and video parameters of the video to be evaluated, the image frame sequence is respectively input into the motion estimation module, subjective image quality module, content attractiveness module and scene module to obtain the motion estimation score, subjective image quality score, content attractiveness score and scene score. The video quality score of the video to be evaluated. This video quality evaluation method integrates motion estimation, subjective image quality, content attractiveness, video scene and objective video parameters, and closely combines the video scene and the subjective preferences of the audience. The resulting video quality evaluation results are more accurate and can be applied in a wider range. Among them, the training data of the content attractiveness module uses samples after data layering processing, and different content attractiveness scores under the same subjective image quality can be obtained to significantly improve the accuracy of subsequent video quality evaluation.

In a possible embodiment, the data fusion processing unit 102 is specifically configured as:

Two consecutive frames of images in the image frame sequence are input as a group to the motion estimation module;

The relative displacement of pixels in each group of frame images is calculated by the motion estimation module, and a motion estimation score is calculated based on the relative displacement of the pixels.

In a possible embodiment, the data fusion processing unit 102 is further configured to:

After the relative displacement of the pixels in each group of frame images is calculated by the motion estimation module, the relative displacement of the pixels in each group of frame images is superimposed with the three-channel color value of the pixels, and input to the motion estimation module to obtain a motion estimation score.

In a possible embodiment, the subjective image quality module is generated based on an image quality assessment model, and includes a feature extraction head with a regression output.

In a possible embodiment, the data fusion processing unit 102 is further configured to:

Before the multi-frame images in the video to be evaluated are acquired and the image frame sequence is generated, based on the scoring points of the sample data by the motion estimation module, the subjective image quality module, the scene module and the objective parameter module, data layering processing is performed on the sample data to obtain layered samples;

A content attractiveness module is trained based on the stratified samples.

In a possible embodiment, the data fusion processing unit 102 is specifically configured as:

The sample data with the same scoring value or the difference between the scoring values is smaller than the preset value to form a stratified sample.

In a possible embodiment, the score calculation unit 103 is configured to:

multiplying the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score, and the objective parameter score by their corresponding weights;

Adding the product results to obtain the video quality score of the video to be evaluated.

In a possible embodiment, the video parameters include a video frame rate, and the data fusion processing unit 102 is further configured to:

Adjusting the frame rate of the video to be evaluated to obtain video to be evaluated with different frame rates, and inputting the video to be evaluated with different frame rates to the objective parameter module to obtain a plurality of objective parameter scores;

Based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and each objective parameter score, the video quality scores of the video to be evaluated at different frame rates are calculated;

The device further includes a decision unit configured to: make a video frame rate decision based on multiple video quality scores corresponding to different frame rates.

Figure 6 is a schematic structural diagram of a video quality assessment device provided in the embodiment of the present application. As shown in Figure 6, the device includes a processor 201, a memory 202, an input device 203, and an output device 204; the number of processors 201 in the device can be one or more, and one processor 201 is used as an example in Figure 6; the processor 201, memory 202, input device 203, and output device 204 in the device can be connected through a bus or in other ways, and in Figure 6, the connection through a bus is used as an example. The memory 202, as a computer-readable storage medium, can be used to store software programs, computer-executable programs and modules, such as program instructions/modules corresponding to the video quality assessment method in the embodiment of the present application. The processor 201 executes various functional applications and data processing of the device by running the software programs, instructions and modules stored in the memory 202, that is, realizes the above-mentioned video quality assessment method. The input device 703 can be used to receive input numbers or character information, and generate key signal input related to user settings and function control of the device. The output device 204 may include a display device such as a display screen.

An embodiment of the present application also provides a non-volatile storage medium containing computer-executable instructions, the computer-executable instructions are used to execute a video quality assessment method described in the above-mentioned embodiments when executed by a computer processor, including:

Obtain the image frame sequence and video parameters of the video to be evaluated;

Inputting the image frame sequence into the motion estimation module, the subjective image quality module, the content attractiveness module and the scene module respectively, and obtaining the motion estimation score, the subjective image quality score, the content attractiveness score and the scene score, and the content attractiveness module is obtained based on sample training after data layering processing;

Inputting the video parameters into the objective parameter module to obtain the objective parameter score;

A video quality score of the video to be evaluated is calculated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and the objective parameter score.

It should be noted that, in the above embodiment of the video quality assessment device, the various units and modules included are only divided according to the functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction, and are not used to limit the scope of protection of the embodiments of the present application.

In some possible implementations, various aspects of the method provided in this application can also be implemented in the form of a program product, which includes program code, and when the program product is run on a computer device, the program code is used to make the computer device execute the steps in the method according to various exemplary embodiments of the application described above in this specification. For example, the computer device can execute the video quality evaluation method described in the embodiment of the application. The program product can be implemented using any combination of one or more readable media.

Claims (12)

1. The video quality assessment method is characterized in that, comprising: Obtain the image frame sequence and video parameters of the video to be evaluated; Inputting the image frame sequence into the motion estimation module, the subjective image quality module, the content attractiveness module and the scene module respectively, and obtaining the motion estimation score, the subjective image quality score, the content attractiveness score and the scene score, and the content attractiveness module is obtained based on sample training after data layering processing; Inputting the video parameters into the objective parameter module to obtain the objective parameter score; A video quality score of the video to be evaluated is calculated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and the objective parameter score. 2. The video quality assessment method according to claim 1, wherein the image frame sequence is input to a motion estimation module to obtain a motion estimation score, comprising: Two consecutive frames of images in the image frame sequence are input as a group to the motion estimation module; The relative displacement of pixels in each group of frame images is calculated by the motion estimation module, and a motion estimation score is calculated based on the relative displacement of the pixels. 3. The video quality assessment method according to claim 2, wherein, after the relative displacement of the pixels in each group of frame images is obtained through the calculation of the motion estimation module, further comprising: The relative displacement of the pixel in each group of frame images is superimposed with the three-channel color value of the pixel, and input to the motion estimation module to obtain a motion estimation score. 4. The video quality assessment method according to claim 1, wherein the subjective image quality module is generated based on an image quality assessment model, comprising a feature extraction head with regression output. 5. The video quality assessment method according to claim 1, wherein, before said obtaining multi-frame images in the video to be evaluated, and generating an image frame sequence, it also includes: Based on the scoring values of the sample data by the motion estimation module, the subjective image quality module, the scene module and the objective parameter module, performing data layering processing on the sample data to obtain layered samples; A content attractiveness module is trained based on the stratified samples. 6. The video quality assessment method according to claim 5, wherein said performing data layering processing on said sample data to obtain layered samples comprises: The sample data with the same scoring value or the difference between the scoring values is smaller than the preset value to form a stratified sample. 7. The video quality assessment method according to any one of claims 1-6, wherein the calculation of the video quality score of the video to be evaluated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and the objective parameter score includes: multiplying the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score, and the objective parameter score by their corresponding weights; Adding the product results to obtain the video quality score of the video to be evaluated. 8. The video quality assessment method according to any one of claims 1-6, wherein the video parameters include video frame rate, and the video quality assessment method further comprises: Adjusting the frame rate of the video to be evaluated to obtain video to be evaluated with different frame rates, and inputting the video to be evaluated with different frame rates to the objective parameter module to obtain a plurality of objective parameter scores; Based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score and each objective parameter score, the video quality scores of the video to be evaluated at different frame rates are calculated; Based on the video quality scores corresponding to multiple different frame rates, the video frame rate decision is made. 9. A video quality assessment device, characterized in that it comprises: A data acquisition unit configured to acquire an image frame sequence and video parameters of a video to be evaluated; The data fusion processing unit is configured to input the image frame sequence to the motion estimation module, the subjective image quality module, the content attractiveness module and the scene module respectively, to obtain the motion estimation score, the subjective image quality score, the content attractiveness score and the scene score, and the content attractiveness module is obtained based on the sample training after data layering processing; and, the video parameters are input to the objective parameter module to obtain the objective parameter score; A score calculation unit configured to calculate a video quality score of the video to be evaluated based on the motion estimation score, the subjective image quality score, the content attractiveness score, the scene score, and the objective parameter score. 10. A video quality evaluation device, the device comprising: one or more processors; storage means for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the video quality evaluation method according to any one of claims 1-8. 11. A non-volatile storage medium storing computer-executable instructions, the computer-executable instructions being used to execute the video quality assessment method according to any one of claims 1-8 when executed by a computer processor. 12. A computer program product, comprising a computer program, characterized in that, when the computer program is executed by a processor, the video quality assessment method according to any one of claims 1-8 is implemented.