Symbol Parameter Calculation Device, Symbol Parameter Calculation Method, and Program

Abstract

1. An encoding parameter calculation device comprising: a processor; and a memory, the memory including instructions for causing the processor to execute the steps of: dividing an original video into a plurality of scenes and selecting one of the plurality of scenes; applying a plurality of combinations of bit rate and resolution to the selected one of the scenes to generate a plurality of encoded scenes; evaluating quality of each of the encoded scenes; estimating a relationship between a viewing abandonment rate and a bit rate for the one of the scenes; applying the relationship to throughput history data for the one of the scenes to obtain a plurality of values ​​of the viewing abandonment rate; applying a cluster algorithm to the plurality of values ​​of the viewing abandonment rate to calculate N discrete quantization values ​​of the viewing abandonment rate; and identifying a plurality of combinations of bit rate, resolution, and quality corresponding to the N discrete quantization values.

Description

Technical Field

[0001] The present invention relates to an encoding parameter calculation device, an encoding parameter calculation method, and a program.

Background Art

[0002] The increasing use of video streaming services such as YouTube (registered trademark), Netflix, and Twitch has presented new challenges. With the increasing amount of data to be transmitted, it is necessary to optimize video encoding in order to secure storage capacity and reduce network usage. In addition, since the network state varies depending on the load on the network and the user's location, adaptive bitrate (ABR) video streaming should be used to ensure the optimal viewing quality for viewers. To use ABR video streaming, it is necessary to define a plurality of parameters such as combinations of resolution, frame rate, and bitrate. For each set of parameters, different throughput requirements are determined, and streaming in various network situations becomes possible with these sets of the plurality of parameters. Therefore, the video is encoded at a plurality of different bitrate values. Encoding the video at a plurality of different bitrate values is called a bitrate ladder.

[0003] There are multiple different ways to define a bitrate ladder. First, there are pre-defined coding recepts. These are tables published by video streaming platforms that provide recommendations regarding encoding settings such as bitrate values corresponding to various resolutions. A clear limitation of this method is that the bitrate requirements for achieving high quality are not the same for all content. Therefore, using the same bitrate value for different content will result in an excessive allocation of bits for videos that are easy to encode, wasting space on the server and unnecessarily using the network. On the other hand, for videos that are not easy to encode, encoding artifacts may appear and the quality may become low.

[0004] To address this problem, it is necessary to define a bitrate ladder based on content encoding complexity. Defining a bitrate ladder based on content encoding complexity is called "per-title video encoding". A common method to achieve this is to encode videos with multiple resolutions and bitrate values and evaluate the videos using a quality metric such as Video Multi-method Assessment Fusion (VMAF). VMAF is a computer program that uses machine learning to predict the quality of encoded and / or resampled videos. In this algorithm, the encoded video and the source content before encoding are input, and the change in pixel values before and after encoding is measured to predict the quality of the encoded video. Using such an algorithm, measure the rate-distortion curve for each resolution while considering the encoding complexity of the video content and specifying the resolution to be used at a given bitrate. Regarding the bitrate value, it is common to set the value so that the quality of the video is separated by one Just Noticeable Difference (JND). A scene-by-scene analysis that extends this method has also been proposed.

[0005] However, several challenges still remain. First, this method requires a huge amount of computational effort and relies on multiple encoding trials. Second, while this method reduces the dimensionality of the search space of encoding conditions to one dimension (bit rate), it does not determine the optimal value to be used. To address the computational complexity, a support vector regression method that uses both spatial and temporal texture features has been proposed. This method predicts the range of bit rates for which a given resolution should be used while significantly reducing the number of encodings. Additionally, a unique solution based on machine learning that directly predicts the encoding complexity or bit rate ladder has also been proposed. However, the details of the algorithms have not been published, making it difficult to compare or evaluate these methods. To address the problem of bit rate ladders optimized for a network, researchers have used actual throughput history data to define a bit rate ladder based on a probabilistic model with the aim of ensuring a minimum quality while minimizing network usage. Analytical evaluation methods for bit rate ladders based on non-linear constrained optimization are known. A player simulation model is defined, and a bit rate ladder is defined to maximize session quality while considering throughput probabilities based on history data. This method has been extended to support multiple video codecs with the same ladder. The problem of identifying a bit rate ladder has been solved by using integer linear programming while maximizing the perceived quality of experience (QoE) measured using the National Telecommunications and Information Administration Video Quality Metric (NTIA VQM).

[0006] However, these studies have certain common specific constraints. That is, the above methods address the problem of optimizing the ladder for the network, but these studies are based on a model of a video player with simple operation (selecting the highest available quality based on known future throughput values). A player that always selects the highest available quality makes undesirable frequent quality changes, so when doing as described above, there are more restrictions. Therefore, a player-independent method for selecting an optimized bitrate ladder for content and network, corresponding to a more advanced video player model that utilizes a more advanced algorithm, has been proposed.

Prior Art Documents

Non-Patent Documents

[0007]

Non-Patent Document 1

Non-Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0008] There are many challenges in streaming video to a large number of viewers. Each user accesses the Internet with different performance in terms of available throughput, and the available throughput can vary as a function of time. Therefore, in order to stream video to all users, it is necessary to use different representations (videos encoded with various resolutions, bitrates, and frame rate values) to enable the adaptation of the video bitrate requirements to match the user's available throughput. Such a method of adapting quality to the available throughput is called adaptive bitrate video streaming, and the bitrate values corresponding to each possible representation are called the bitrate ladder.

[0009] When defining the bitrate ladder, it is necessary to set sufficient granularity with respect to the bitrate values so that each user can enjoy the highest quality allowed by the Internet access. However, from the operator's perspective, it is also necessary to avoid unnecessary use of the network and unnecessary use of storage space on the server. Therefore, in order to optimize the user experience while reducing operating costs, it is necessary to optimally encode the video.

[0010] Finding the optimal bitrate ladder has already been addressed. However, some challenges still remain. It is necessary to define a bitrate ladder that improves the user experience and reduces the likelihood of the user aborting the service.

Means for Solving the Problems

[0011] According to one aspect of the present invention, there is provided an encoding parameter calculation device including a processor and a memory. The memory causes the processor to divide the original video into a plurality of scenes, select one of the plurality of scenes DoSteps; applying a plurality of combinations of bitrate and resolution to the selected one scene to generate a plurality of encoded scenes; evaluating the quality of each of the encoded scenes; estimating the relationship between the video pause rate and the bitrate for the one scene; applying the relationship to the throughput history data for the one scene to obtain a plurality of values of the video pause rate; applying a clustering algorithm to the plurality of values of the video pause rate to calculate N discrete quantization values of the video pause rate; and specifying a plurality of combinations of bitrate, resolution, and quality corresponding to the N discrete quantization values; An encoding parameter calculation device including instructions for executing the steps is provided.

Advantages of the Invention

[0012] According to the embodiment, it is possible to define a bitrate ladder so as to improve the user experience and reduce the possibility that the user stops the service.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Modes for Carrying Out the Invention

[0014] As described in the background art section, finding the optimal bitrate ladder has already been addressed. However, several issues still remain. In fact, according to previous research, it is possible to optimize the bitrate ladder based on quality, but in previous research, when defining the bitrate ladder, the operation of pausing viewing has not been addressed. Since the relationship between the quality of the service controlled by the bitrate value in the ladder and whether the user requests the continuation of the service usage is not clear, it is difficult to address the operation of pausing viewing. Since the service provider is interested in whether the user wants to use the service rather than the quality itself, such a type of ladder determination mechanism is important. Therefore, it is necessary to develop an optimization strategy for the bitrate ladder considering the influence of quality on whether the user continues or discontinues the service usage. Since the relationship between quality and acceptability is time-dependent, considering the duration of the stimulus when defining the ladder is also a new issue. This point has not been addressed in previous research and will be considered in the following embodiments.

[0015] Considering that the definition of the encoding conditions is not clear, in the following embodiments, a solution for optimizing the bitrate ladder with respect to the characteristics of the content and network statistics will be described.

[0016] Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing a hardware configuration example of an encoding parameter calculation device 10 according to an embodiment of the present invention. The encoding parameter calculation device 10 in FIG. 1 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a central processing unit (CPU) 104, and an interface device 105, etc., which are mutually connected by a bus B.

[0017] The program that realizes the processing in the parameter calculation device 10 for symbolization is provided by a recording medium 101 such as a CD-ROM. When the recording medium 101 storing the program is set in the drive device 100, the program is installed from the recording medium 101 via the drive device 100 into the auxiliary storage device 102. However, the installation of the program does not necessarily have to be performed from the recording medium 101, and it may be downloaded from another computer via a network. The auxiliary storage device 102 stores the installed program and also stores necessary files, data, etc.

[0018] When there is an instruction to start the program, the memory device 103 reads out and stores the program from the auxiliary storage device 102. The CPU 104 executes the functions related to the parameter calculation device 10 for symbolization according to the program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network.

[0019] FIG. 2 is a diagram showing a functional configuration example of the parameter calculation device 10 for symbolization in an embodiment of the present invention. In FIG. 2, the parameter calculation device 10 for symbolization includes a division unit 11, an encoding unit 12, a viewing interruption rate calculation unit 13, a throughput input unit 14, a conversion unit 15, and an analysis unit 16. Each of these units is realized by the processing executed by the CPU 104 by one or more programs installed in the parameter calculation device 10 for symbolization. That is, each of these units is realized by the cooperation of the hardware resources of the parameter calculation device 10 for symbolization and the program (software) installed in the parameter calculation device 10 for symbolization.

[0020] Hereinafter, an example of the processing procedure executed by the parameter calculation device 10 for symbolization will be described. FIG. 3 is a flowchart for explaining an example of the processing procedure executed by the parameter calculation device 10 for symbolization.

[0021] In step S101, the splitting unit 11 analyzes the input original video and splits the original video into a plurality of scenes. Generally, a scene refers to a segment of an action at a certain location. A scene in a video is a collection of a plurality of consecutive shots at a certain location.

[0022] FIG. 4 is a diagram showing an example of splitting an original video into a plurality of scenes. As the purpose of splitting the original video into a plurality of scenes, mainly the following two purposes can be considered. Since the encoding complexity in the original video varies for each scene, it is conceivable to define an optimal bitrate ladder for each scene. Also, by splitting the original video into a plurality of scenes, the duration of each scene can be obtained. Since there is a temporal aspect to the acceptability of a given quality model, it is important to specify the duration of the scenes in the video.

[0023] As a method for splitting the original video into a plurality of scenes, a known method can be used. For example, the method described in Non-Patent Document 2 can be used. The method of Non-Patent Document 2 is one example, and the method for splitting the original video into a plurality of scenes is not limited to the method of Non-Patent Document 2.

[0024] In the following processing, as described above, after splitting the original video into a plurality of scenes, a bitrate ladder is defined for each scene. Hereinafter, the processing for each scene obtained by splitting the original video into a plurality of scenes is shown.

[0025] In step S102, the encoding unit 12 encodes one scene to be evaluated with a plurality of combinations of a plurality of resolutions and a plurality of bitrates.

[0026] Since the complexity of symbolization and the bitrate requirements vary for each video (scene), the encoding unit 12 encodes the video (scene) using a constant rate factor (CRF) that is separated by 4 within the range of 12 to 48 for AVC and HEVC, and within the range of 12 to 60 for VP8, VP9, and AV1. As a result, for each resolution, 10 to 14 different encoded videos can be obtained. By encoding using a constant rate factor, it becomes possible to encode the video with a constant quantization that depends on the motion within the frame. For frames containing large motion, higher values of the quantization parameter (QP) are used, resulting in the effect of masking the encoding compression distortion due to motion and making it possible to achieve higher encoding efficiency. Furthermore, by using encoded videos with a constant rate factor over a wide range, almost uniform sampling can be obtained regarding the quality of possible encoding conditions.

[0027] In step S103, the encoding unit 12 evaluates the quality of a plurality of encoded scenes obtained using different quantization parameters by using a content-dependent quality evaluation method such as Video Multimethod Assessment Fusion (VMAF). For example, the encoding unit 12 calculates, for each of the plurality of encoded scenes, the value of an index indicating an estimated value of the viewer's perceived quality (QoE) using VMAF.

[0028] VMAF is a full-reference method that calculates the difference in quality by comparing the original video signal and the received video signal. That is, in VMAF, the encoded video is compared with the original video, and the quality of the video is evaluated in the range of 0 to 100. In this case, 0 represents the lowest quality and 100 represents the highest quality. When using a quality evaluation method other than VMAF, the evaluation value of the quality may be different from the range of 0 to 100. For example, the range of the output quality value may be in the range of 1 to 5.

[0029] By executing the above processing, the characteristics shown in FIG. 5 (the relationship between the values of the encoding parameters and QoE) can be obtained. Further, by drawing a curve so as to obtain the maximum QoE with respect to the bit rate, the dashed curve in FIG. 5 is obtained. This dashed line is generally called a Covex hull. The encoding unit 12 derives the Covex hull indicated by the dashed curve in FIG. 5.

[0030] The Convex hull is represented by the curve shown by the dashed line in FIG. 5. The Covex hull indicates the highest quality obtained when changing the resolution among all resolutions (all finite resolutions) for one bit rate.

[0031] FIG. 6 is a diagram showing an example of the process of deriving the Convex hull. In the example of FIG. 6, the encoding unit 12 specifies the highest quality value obtained when changing the resolution among all resolutions (all finite resolutions) for each bit rate value.

[0032] In FIG. 6, each encoded scene is indicated by a star mark. To derive the Convex hull, the encoding unit 12 compares the quality evaluation results of scenes with the same bit rate as the encoded scene but different resolutions for each encoded scene.

[0033] FIG. 6 shows an example of this procedure. The scene encoded at 720p and bit rate value b shows the quality value Q 720 at a resolution of 720p and shows the quality value Q 1080 at a resolution of 1080p. Q 1080 >Q 720 and since Q 1080 is the highest quality achievable at bit rate b, the Convex hull is calculated based on the data point [b, Q 1080 .

[0034] Here, since there is no encoded video at a resolution of 1080p and a bit rate of b, Q 1080It is necessary to estimate. As shown in FIG. 6, the encoding unit 12 uses linear regression for two quality values corresponding to bitrate values before and after the bitrate value b among the 1080p encoded videos, thereby obtaining Q 1080 estimation.

[0035] By applying this procedure, the encoding unit 12 obtains the highest quality value that can be obtained when changing the resolution among all resolutions (all finite resolutions) for each bitrate.

[0036] A model (for example, Convex hull) that associates each bitrate with the highest quality value that can be obtained when changing the resolution among all resolutions (all finite resolutions) can be formulated by the sigmoid function shown in Equation 1 below.

[0037]

Equation

[0038] When VMAF is used, the quality value is defined in the range of [0, 100]. However, in the user viewing interruption prediction model that predicts the possibility of stopping viewing, the video quality is evaluated in the range of [1, 5]. Therefore, the encoding unit 12 maps the quality value of VMAF evaluated using Convex hull to the range [1, 5] using the linear function defined in Equation 2.

[0039]

Equation

[0040] When a quality evaluation method other than VMAF is used, in order to make the evaluation method give a unified quality value, the quality prediction may be adjusted to match the reference data by maintaining this linear function.

[0041] After identifying the relationship between the bitrate and the content encoding quality, the viewing interruption rate calculation unit 13 models the probability that the user stops viewing at the end of a scene due to low quality using Equation 3.

[0042]

Equation

[0043] The viewing interruption rate calculation unit 13 obtains MOSv, which is the quality of the video, using Equation 2. The viewing interruption rate calculation unit 13 obtains MOSa, which is the quality of the audio, using Equation 4.

[0044]

Equation

[0045] In Equation 4, B1, B2, and B3 are model parameters obtained using regression analysis in the database for learning the evaluation of audio quality. The model parameters B1, B2, and B3 do not depend on the content and are constant in all scenes. Since the encoding efficiency varies for each audio codec, the model parameters B1, B2, and B3 depend only on the audio codec used for audio encoding.

[0046] Using regression analysis in a plurality of different databases, a table of model parameters B1, B2, and B3 for the audio codec is obtained. The viewing interruption rate calculation unit 13 selects appropriate model parameters B1, B2, and B3 based on the codec used for audio encoding.

[0047] In Equation 4, b a is the bit rate of the audio. For the bit rate of the audio, different options are conceivable.

[0048] In the first option, the bit rate of the audio is set to a high quality level, for example, 384 kbps. In this case, the interruption of viewing will depend only on the quality of the video.

[0049] However, in the user viewing interruption prediction model, the quality of the audio is also considered. Therefore, it is also conceivable to define a relationship between the video bit rate and the audio bit rate. For example, it is also conceivable to define a linear relationship such that the audio bit rate is limited to the range from 32 kbps to 384 kbps and then set to 1 / 42 of the video bit rate. Alternatively, it is also conceivable to define a more general logistic function between the video bit rate and the audio bit rate.

[0050] In step S104, the viewing interruption rate calculation unit 13 evaluates the viewing interruption rate after the viewing session of period d using the evaluation of the video quality and the evaluation of the audio quality.

[0051] The period of the scene is obtained using the scene segmentation analysis shown in FIG. 4. Based on this model, in step S105, the viewing interruption rate calculation unit 13 obtains the relationship between the viewing interruption rate and the bit rate for each scene. FIG. 7 is a diagram showing an example of the relationship between the viewing stop rate and the bit rate for the scene to be evaluated.

[0052] In step S106, the throughput input unit 14 inputs network throughput history data for the scene to be evaluated. Note that when multiple viewings (deliveries) are performed over a certain period, a plurality of time series data may be input. The time axis of each time series data corresponds to the viewing time of the video. The throughput input unit 14 can obtain characteristics (relationship between viewing time and throughput) as shown in FIG. 8 for each past viewing (delivery) from such time series data.

[0053] FIG. 8 is a diagram showing an example of a time series of throughput values for the scene to be evaluated. The throughput values shown in FIG. 8 indicate the data transfer rate (bandwidth) available to the client.

[0054] In step S107, the conversion unit 15 converts the time series of throughput values into achievable quality values. For example, in formulas (1)-(4), the conversion unit 15 applies the throughput value as the bit rate to convert the time series of throughput into achievable quality values. FIG. 9 is a diagram showing an example of converting the time series of throughput values into achievable quality.

[0055] In step S108, the conversion unit 15 converts the time series of throughput values into a viewing interruption rate. For example, the conversion unit 15 converts the time series of throughput values into a time series of viewing interruption rates based on the relationship between the viewing stop rate obtained in step S105 and the bit rate. FIG. 10 is a diagram showing an example of converting the time series of throughput values into a viewing interruption rate.

[0056] In step S109, the analysis unit 16 analyzes the viewing interruption rate for the scene to be evaluated based on the network throughput history data for the scene to be evaluated, thereby deriving the distribution function of the viewing interruption rate. For example, the analysis unit 16 converts the time series of throughput values for the scene to be evaluated obtained for each past viewing (distribution) into the time series of the viewing interruption rate. From the plurality of time series of the viewing interruption rate thus obtained, a plurality of viewing interruption rates may be collected, and the distribution function of the viewing interruption rate may be derived by analyzing the frequency of the values of each viewing interruption rate.

[0057] FIG. 11 is a diagram showing an example of the distribution function of the viewing interruption rate for the scene to be evaluated.

[0058] When considering all the achievable quality values and all the values of the viewing interruption rate, specifying the bitrate ladder for the scene to be evaluated can be expressed as an optimal quantization problem.

[0059] Finding the optimal bitrate ladder can be considered to be identifying N discrete values that better approximate the continuous values taken by the viewing interruption rate. Let L in Equation 5 shown below be defined as the set of N discrete values of the bitrate ladder.

[0060]

Equation

[0061] The problem of specifying the optimal bitrate ladder can be formulated by the following Equation 6.

[0062]

Equation

Equation

Equation

Number

[0063] In Equation 6

Number

Number

[0064] As described in Equation 6, the quantization error is calculated for all values of the viewing interruption rate

Number

[0065] In step S110, the ladder calculation unit 17 sets the number of data as the value of the viewing interruption rate

Number

Number

[0066] Note that the ladder calculation unit 17 should consider the viewing interruption rate

Number

Number

[0067] After the cluster assignment by the k - means method converges, among the N values

Number

[0068] FIG. 12 is a diagram showing an example of the result of classification by the k - means method.

[0069] The value of the viewing interruption rate

Number

Number

Number

[0070] In step S110, the ladder calculation unit 17

Number

Number

[0071] In step S110, the ladder calculation unit obtains the resolution value used together with these bit rate values from the encoding complexity analysis performed when specifying the Covex hull. Further, the ladder calculation unit 17 specifies the QoE corresponding to the resolution used using the Covex hull. FIG. 14 is a diagram showing an example of specifying the resolution used corresponding to the viewing interruption rate and the corresponding QoE using the encoding complexity analysis.

[0072] Therefore, the encoding parameter calculation device 10 can obtain a set of a bit rate value, a resolution, and a QoE for each quantization representative value of the viewing interruption rate value. As a result, a bit rate ladder in which the resolution, bit rate (VBR), and QoE are associated is obtained as shown in FIG. 15.

[0073] It has been shown that it is possible to perform the encoding complexity analysis of video from the viewpoint of the user's viewing interruption rate. Compared with the conventional analysis method using the quality-based evaluation method and the quality-based rate-distortion curve, by performing the analysis from the viewpoint of the user's viewing interruption rate, it becomes possible to consider the duration of the stimulus and the temporal influence on the viewing time of low-quality users.

[0074] Quality is an important indicator, but knowing the influence of quality on viewing time and viewing behavior is an important point for network operators. Therefore, performing the analysis based on the user's viewing interruption rate gives a high interpretability regarding the influence on the service of the selection of the bit rate ladder.

[0075] According to this embodiment, it is possible to define a bitrate ladder so as to minimize the user's video pause behavior. The optimization of the bitrate ladder is based on historical data of content encoding complexity and network throughput. It has been shown that it is possible to estimate the user's video pause rate by combining the historical data of content encoding complexity and network throughput. Based on the combination of the historical data of content encoding complexity and network throughput, it is possible to select the bitrate ladder used when encoding video.

[0076] Also, according to this embodiment, it is possible to define a bitrate ladder that optimizes the video encoding complexity measured by the user's video pause rate without prior knowledge of the video player and the adaptive bitrate control method used by the video player.

[0077] Also, according to this embodiment, by using a model of the video pause rate, it is possible to define the bitrate ladder used for video encoding in consideration of audio quality as well as video quality.

[0078] As described above, the embodiments of the present invention have been described in detail, but the present invention is not limited to such specific embodiments, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims.

Claims

1. An apparatus for calculating encoding parameters, comprising: a processor; and a memory, wherein the memory causes the processor to:[[]] divide the original video into a plurality of scenes and select one scene from the plurality of scenes; apply a plurality of combinations of bitrates and resolutions to the selected one scene to generate a plurality of encoded scenes; evaluate the quality of each of the encoded scenes; estimate the relationship between the viewing interruption rate and the bitrate for the one scene; apply the relationship to the throughput history data for the one scene to obtain a plurality of values of the viewing interruption rate; apply a clustering algorithm to the plurality of values of the viewing interruption rate to calculate N discrete quantization values of the viewing interruption rate; and identify a plurality of combinations of bitrates, resolutions, and qualities corresponding to the N discrete quantization values; and includes instructions for causing the above to be executed, an apparatus for calculating encoding parameters.

2. In the step of calculating the N discrete quantization values of the viewing interruption rate, the processor applies a clustering classification algorithm to values of the plurality of values of the viewing interruption rate that are below a predetermined threshold, The encoding parameter calculation apparatus according to claim 1.

3. In the step of estimating the relationship, the processor includes the audio bitrate and the video bitrate as the bitrate, The encoding parameter calculation apparatus according to claim 1.

4. The audio bitrate is a predetermined bitrate in the range of 32 kbps to 384 kbps, The encoding parameter calculation apparatus according to claim 3.

5. The audio bitrate is a bitrate obtained by multiplying the video bitrate by a specific coefficient, The encoding parameter calculation apparatus according to claim 3.

6. An encoding parameter calculation method executed by a computer, comprising: divide the original video into a plurality of scenes and select one scene from the plurality of scenes; apply a plurality of combinations of bitrates and resolutions to the selected one scene to generate a plurality of encoded scenes; evaluate the quality of each of the encoded scenes; estimate the relationship between the viewing interruption rate and the bitrate for the one scene; ​ applying the relationship to the throughput history data for the one scene to obtain a plurality of values of the viewing interruption rate; applying a clustering algorithm to the plurality of values of the viewing interruption rate to calculate N discrete quantization values of the viewing interruption rate; and identifying a plurality of combinations of bitrate, resolution, and quality corresponding to the N discrete quantization values; A coding parameter calculation method comprising the above steps.

7. A non-transitory computer-readable storage medium storing a coding parameter calculation program, wherein when the coding parameter calculation program is executed by a computer, the computer dividing the original video into a plurality of scenes and selecting one scene from the plurality of scenes; applying a plurality of combinations of bitrate and resolution to the selected one scene to generate a plurality of encoded scenes; evaluating the quality of each of the encoded scenes; estimating a relationship between the viewing interruption rate and the bitrate for the one scene; applying the relationship to the throughput history data for the one scene to obtain a plurality of values of the viewing interruption rate; applying a clustering algorithm to the plurality of values of the viewing interruption rate to calculate N discrete quantization values of the viewing interruption rate; and identifying a plurality of combinations of bitrate, resolution, and quality corresponding to the N discrete quantization values; A non-transitory computer-readable storage medium that executes the above steps.

Images