Streaming media platform optimization method and system based on video image processing

Abstract

The invention discloses a streaming media platform optimization method and system based on video image processing, and relates to the technical field of streaming media, and the method comprises the steps: collecting an image frame sequence, extracting a brightness histogram feature value and color saturation data, calculating the quality score of the image frame sequence, and building a dynamic bit rate distribution model; performing coding compression processing on the image frame sequence through a dynamic bit rate distribution model to generate a video data stream with a self-adaptive code rate, and transmitting the video data stream to a streaming media server; and collecting playing quality feedback data of the streaming media server, and optimizing and adjusting the dynamic bit rate distribution model based on the playing quality feedback data. According to the method, the brightness histogram and the color saturation data of the video frame are extracted, the quality scoring model based on the double-branch neural network is established, and adaptive video coding compression is realized in combination with dynamic bit rate allocation, so that the bandwidth occupation is effectively reduced while the video quality is ensured, and the user experience and the system performance of streaming media service are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of streaming media, and in particular to a streaming media platform optimization method and system based on video image processing. Background Art

[0002] Traditional streaming media transmission technology mainly adopts adaptive bitrate adjustment schemes with fixed bitrate or preset thresholds, and dynamically adjusts the video bitrate by detecting the network bandwidth status. However, such schemes often ignore the characteristic differences of the video content itself, and fail to fully consider the impact of visual quality factors such as picture complexity, brightness distribution, and color saturation on the transmission effect. At the same time, the existing technology lacks a real-time feedback mechanism for the user experience of the playback end during the bitrate adaptive adjustment process, making it difficult to accurately evaluate and optimize the actual playback effect, which can easily lead to problems such as unstable picture quality and playback freezes.

[0003] The TCP-based adaptive bitrate streaming (ABR) technology currently widely used in the industry can achieve dynamic bitrate adjustment to a certain extent, but its decision-making mechanism is too dependent on network status parameters and fails to establish an effective mapping relationship between video content features and transmission strategies. Especially in complex and changeable network environments, due to the lack of accurate evaluation and intelligent prediction capabilities of video picture quality, existing technologies are difficult to achieve optimal bitrate allocation under bandwidth constraints. In addition, traditional solutions generally use fixed compression parameters and encoding modes in the video encoding process, and fail to flexibly adjust the compression strategy according to the image characteristics of different scenes, which not only affects the utilization efficiency of bandwidth resources, but also makes it difficult to ensure the continuity of video playback and the fidelity of picture details. In practical applications, these technical limitations often lead to significant fluctuations in user experience quality, restricting the further improvement of the service performance and user satisfaction of streaming media platforms. Summary of the invention

[0004] In view of the fact that the existing streaming media transmission technology has many shortcomings in video quality assessment, bit rate adaptive adjustment and user experience optimization, especially the lack of in-depth analysis of video content characteristics and intelligent transmission control mechanism, which makes it impossible to achieve the optimal transmission effect in a complex network environment, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to establish an adaptive optimization mechanism based on video content features and playback quality feedback through video image processing and machine learning technology to improve the transmission efficiency and user experience quality of the streaming media platform.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a streaming media platform optimization method based on video image processing, which includes collecting an image frame sequence in a target video stream, and extracting brightness histogram eigenvalues ​​and color saturation data from the image frame sequence; calculating a quality score of the image frame sequence based on the brightness histogram eigenvalues ​​and the color saturation data, and establishing a dynamic bitrate allocation model; encoding and compressing the image frame sequence through the dynamic bitrate allocation model to generate a video data stream with an adaptive bitrate, and transmitting it to a streaming media server; collecting playback quality feedback data of the streaming media server, and optimizing and adjusting the dynamic bitrate allocation model based on the playback quality feedback data.

[0008] As a preferred solution of the streaming media platform optimization method based on video image processing described in the present invention, wherein: the playback quality feedback data of the streaming media server is collected, and the dynamic bit rate allocation model is optimized and adjusted based on the playback quality feedback data, including: obtaining the playback quality feedback data of the streaming media server, wherein the playback quality feedback data includes playback fluency, buffer status and delay parameters; constructing a playback quality evaluation system, performing weighted calculation on the playback quality feedback data, generating a standardized playback quality score, and comprehensively evaluating the playback effect; establishing a feedback optimization mechanism based on the playback quality score, and optimizing and adjusting the parameters of the dynamic bit rate allocation model; and feeding back the optimized parameters to the dynamic bit rate allocation model in real time to optimize the transmission quality of the video data stream.

[0009] As a preferred solution of the streaming media platform optimization method based on video image processing described in the present invention, the method in which: the dynamic bit rate allocation model encodes and compresses the image frame sequence is to divide the image frame sequence into image groups based on the target bit rate output by the dynamic bit rate allocation model; according to the frames of the image group, a differential compression method is used for compression, and a rate-distortion optimization strategy is used to divide each frame of the image into macroblock units; for the encoding mode selection of the macroblock unit, an adaptive cost function is introduced to generate a video data stream with an adaptive bit rate; the video data stream is encapsulated, and the encapsulated video data stream is transmitted to the streaming media server through the RTMP protocol.

[0010] As a preferred solution of the streaming media platform optimization method based on video image processing described in the present invention, the method for constructing the dynamic bit rate allocation model is as follows: based on the brightness histogram eigenvalues ​​and color saturation data, an image quality assessment model is established using a dual-branch neural network structure; the features extracted by the dual-branch neural network structure are weightedly fused to obtain a quality score for the image frame sequence; based on the temporal variation law of the quality score, a sliding time window method is used to establish a dynamic bit rate allocation model.

[0011] As a preferred solution of the streaming media platform optimization method based on video image processing described in the present invention, wherein: the dual-branch neural network structure includes a first branch neural network structure and a second branch neural network structure; the first branch neural network structure receives the brightness histogram feature value, and extracts the brightness distribution feature through a three-layer convolution layer; the second branch neural network structure receives the color saturation data, and extracts the color feature through a fully connected layer; the dynamic bitrate allocation model includes a quality threshold judgment module and a bitrate adjustment module; the quality threshold judgment module sets an adaptive adjustment threshold according to the fluctuation amplitude of the quality score; the bitrate adjustment module dynamically calculates the target bitrate based on the adaptive adjustment threshold.

[0012] As a preferred solution of the streaming media platform optimization method based on video image processing of the present invention, the specific formula of the dynamic bit rate allocation model is as follows:

[0013]

[0014] Q(t)=1+β·tanh(Δq / σ)

[0015]

[0016] Among them, R(t) is the target bit rate value at time t, R b is the base bit rate, T is the time window length, α is the time decay coefficient, Q(t) is the quality score function, ΔH i is the image entropy change value of the i-th time window, n is the number of sampling windows, is the visual feature response function, β is the quality adjustment coefficient, Δq is the quality score deviation, σ is the quality score standard deviation, v is the normalized image complexity feature value, λ is the response function gain coefficient, and μ is the response function threshold.

[0017] As a preferred solution of the streaming media platform optimization method based on video image processing described in the present invention, wherein: an image frame sequence in a target video stream is collected, and brightness histogram feature values ​​and color saturation data are extracted from the image frame sequence, including: collecting an image frame sequence in a target video stream, sampling the image frame sequence frame by frame according to a preset sampling interval, and obtaining a key image frame; converting the key image frame from an RGB color space to an HSV color space, and extracting a brightness component in the HSV color space; counting the pixel value distribution of the brightness component, establishing a brightness histogram, and calculating a brightness histogram feature value; based on the saturation component in the HSV space, analyzing the color space distribution characteristics, and obtaining regional saturation characteristics and color saturation data.

[0018] In a second aspect, an embodiment of the present invention provides a streaming media platform optimization system based on video image processing, comprising: an extraction module, used to collect an image frame sequence in a target video stream, and extract brightness histogram eigenvalues ​​and color saturation data from the image frame sequence; an establishment module, used to calculate the quality score of the image frame sequence based on the brightness histogram eigenvalues ​​and the color saturation data, and establish a dynamic bit rate allocation model; a generation module, used to encode and compress the image frame sequence through the dynamic bit rate allocation model, generate a video data stream with an adaptive bit rate, and transmit it to a streaming media server; an optimization module, used to collect playback quality feedback data of the streaming media server, and optimize and adjust the dynamic bit rate allocation model based on the playback quality feedback data.

[0019] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the streaming media platform optimization method based on video image processing as described in the first aspect of the present invention are implemented.

[0020] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the streaming media platform optimization method based on video image processing as described in the first aspect of the present invention are implemented.

[0021] The beneficial effects of the present invention are as follows: the present invention extracts the brightness histogram and color saturation data of video frames, establishes a quality scoring model based on a dual-branch neural network, and realizes adaptive video coding compression in combination with dynamic bit rate allocation; the system introduces a closed-loop optimization mechanism based on playback quality feedback, continuously adjusts the bit rate allocation strategy, effectively reduces bandwidth occupancy while ensuring video quality, and improves the user experience and system performance of streaming media services; especially in network fluctuations and complex video content scenarios, it shows good adaptability and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0023] Figure 1 This is a flow chart of a streaming media platform optimization method based on video image processing according to Example 1.

[0024] Figure 2This is a brightness histogram of the streaming media platform optimization method based on video image processing in Example 1.

[0025] Figure 3 This is a comparison chart of video quality scores of the streaming media platform optimization method based on video image processing in Example 2.

[0026] Figure 4 This is a system delay comparison chart of the streaming media platform optimization method based on video image processing in Example 2. DETAILED DESCRIPTION

[0027] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0028] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0029] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0030] Example 1

[0031] Reference Figure 1-2 , which is the first embodiment of the present invention, and provides a streaming media platform optimization method based on video image processing, comprising:

[0032] S1: Acquire an image frame sequence in a target video stream, and extract brightness histogram feature values ​​and color saturation data from the image frame sequence.

[0033] Specifically, an image frame sequence in a target video stream is collected, and the image frame sequence is sampled frame by frame according to a preset sampling interval to obtain a key image frame; the key image frame is converted from an RGB color space to an HSV color space, and a brightness component in the HSV color space is extracted.

[0034] It should be noted that the principle of converting the key image frame from RGB color space to HSV color space is: RGB color space is an additive color model based on the three components of red (R), green (G) and blue (B); HSV color space is based on human perception and consists of hue (Hue), saturation (Saturation) and brightness value (Value). The conversion process requires nonlinear mapping.

[0035] Furthermore, the conversion processing steps are as follows: normalize and preprocess the RGB image data of the input key image frame, and uniformly map the value ranges of the three RGB components to the interval [0,1]; compare the sizes of the three components R, G and B, and take the maximum value as the brightness component V in the HSV space; calculate the saturation S: find the maximum and minimum values ​​of R, G and B; when the maximum value of R, G and B is not 0, the saturation S is equal to the difference between the maximum and minimum values ​​divided by the maximum value; when the maximum value of R, G and B is 0, the saturation S is directly set to 0.

[0036] Furthermore, the hue H is calculated: when the maximum value among R, G and B is equal to the minimum value, the hue H is set to 0; if R is the maximum value, the hue H is equal to 60 degrees multiplied by G minus B divided by the difference between the maximum and minimum values ​​among R, G and B; if G is the maximum value, the hue H is equal to 60 degrees multiplied by (2 plus B minus R divided by the difference between the maximum and minimum values ​​among R, G and B); if B is the maximum value, the hue H is equal to 60 degrees multiplied by (4 plus R minus G divided by the difference between the maximum and minimum values ​​among R, G and B); if the calculated hue H is a negative value, add 360 to normalize it to the range of [0,360].

[0037] Specifically, extract the brightness component: extract the calculated brightness value V separately, for an 8-bit image, remap the brightness value range to the [0,255] interval, and save the extracted brightness component data for subsequent histogram feature analysis.

[0038] Furthermore, the pixel value distribution of the brightness component is statistically analyzed, a brightness histogram is established, and the characteristic value of the brightness histogram is calculated. The specific formula is as follows:

[0039]

[0040] Where P(k) is the probability of occurrence of a pixel with brightness value k, H(k) is the number of pixels with brightness value k, M is the number of pixels of image height, N is the number of pixels of image width, a is the mean brightness, reflecting the overall brightness level of the image, b is the standard deviation, indicating the degree of discreteness of brightness distribution, and k is the brightness level value, ranging from [0,255].

[0041] Furthermore, based on the saturation component in the HSV space, the color space distribution characteristics are analyzed to obtain regional saturation characteristics and color saturation data.

[0042] It should be noted that the mean and standard deviation of the brightness histogram are used as the brightness histogram feature values; the mean and variance indicators of the color saturation of the entire image are used as color saturation data.

[0043] S2: Calculate the quality score of the image frame sequence according to the brightness histogram feature value and the color saturation data, and establish a dynamic bit rate allocation model.

[0044] Specifically, the method for constructing the dynamic bit rate allocation model is to establish an image quality assessment model using a dual-branch neural network structure based on brightness histogram eigenvalues ​​and color saturation data.

[0045] It should be noted that the dual-branch neural network structure includes a first branch neural network structure and a second branch neural network structure; the first branch neural network structure receives the brightness histogram feature value and extracts the brightness distribution characteristics through three convolutional layers; the second branch neural network structure receives the color saturation data and extracts the color characteristics through a fully connected layer.

[0046] Furthermore, the features extracted by the dual-branch neural network structure are weighted and fused to obtain the quality score of the image frame sequence.

[0047] It should be noted that the basic parameters of the dynamic bitrate allocation model initialized based on the quality score include: the base bitrate range is [4Mbps, 12Mbps], the sliding time window length is set to 30 frames; the quality score value range is [0, 100], and the bitrate adjustment step is set to 10% of the base bitrate.

[0048] Furthermore, a layered bit rate allocation strategy is established. When the quality score is greater than 80, the target bit rate is set to 1.2 times the baseline bit rate; when the quality score is between 50 and 80, the baseline bit rate is kept unchanged; when the quality score is less than 50, the target bit rate is reduced to 0.8 times the baseline bit rate; a bit rate smooth transition mechanism is designed. When the bit rate change amplitude of adjacent time windows exceeds 20% of the current bit rate, a linear interpolation algorithm is started for smoothing, and the bit rate adjustment value is stored in a buffer with a capacity of 5 frames; based on the bit rate adjustment value in the buffer, it is gradually released at a fixed time interval of 100ms to achieve a smooth transition of the bit rate and avoid drastic fluctuations in picture quality.

[0049] Specifically, based on the temporal variation law of the quality score, a dynamic bit rate allocation model is established by using a sliding time window method. The specific formula is as follows:

[0050]

[0051] Q(t)=1+β·tanh(Δq / σ)

[0052]

[0053] Among them, R(t) is the target bit rate value at time t, R b is the base bit rate, T is the time window length, α is the time decay coefficient, Q(t) is the quality score function, ΔH i is the image entropy change value of the i-th time window, n is the number of sampling windows, is the visual feature response function, β is the quality adjustment coefficient, Δq is the quality score deviation, σ is the quality score standard deviation, v is the normalized image complexity feature value, λ is the response function gain coefficient, and μ is the response function threshold.

[0054] It should be noted that the dynamic bit rate allocation model includes a quality threshold judgment module and a bit rate adjustment module; the quality threshold judgment module sets an adaptive adjustment threshold according to the fluctuation range of the quality score; the bit rate adjustment module dynamically calculates the target bit rate based on the adaptive adjustment threshold.

[0055] Furthermore, the quality threshold judgment module: when the fluctuation range of the quality score continues to exceed the preset threshold, the quality threshold judgment module will automatically increase the adaptive adjustment threshold; when the quality scores of multiple consecutive time windows show a downward trend, the system will lower the adaptive adjustment threshold; when the image entropy change value ΔH i When a sudden change occurs, the system will temporarily adjust the adaptive adjustment threshold; when the quality score standard deviation σ is large, it indicates that the picture quality is unstable, and the system will expand the adjustment range of the adaptive adjustment threshold.

[0056] Furthermore, the bitrate adjustment module: when the quality score deviation Δq exceeds the adaptive adjustment threshold, the bitrate adjustment module triggers the bitrate adjustment and calculates the new target bitrate value R(t); when the image complexity feature value v grows rapidly, the system increases the base bitrate R b weight; when the output value of the visual feature response function μ is close to saturation, the system will limit the growth of the target bit rate; when the quality score within the time window is stable, the system will reduce the influence of the time decay coefficient λ and maintain the current bit rate allocation strategy.

[0057] S3: Encode and compress the image frame sequence through the dynamic bit rate allocation model to generate a video data stream with an adaptive bit rate, and transmit it to a streaming media server.

[0058] Specifically, the method for encoding and compressing the image frame sequence using the dynamic bit rate allocation model is to divide the image frame sequence into image groups based on the target bit rate output by the dynamic bit rate allocation model.

[0059] It should be noted that the picture group sets the picture group length to 60 frames, wherein each picture group includes 1 I frame, 14 P frames and 45 B frames, and sets the maximum number of reference frames to 4.

[0060] Furthermore, according to the frames of the image group, a differential compression method is used for compression, and a rate-distortion optimization strategy is used to divide each frame of the image into macroblock units; an adaptive cost function is introduced for the encoding mode selection of the macroblock unit to generate a video data stream with an adaptive bit rate; the video data stream is encapsulated and transmitted to the streaming media server through the RTMP protocol.

[0061] It should be noted that the macroblock coding strategy divides each frame of the image into macroblock units of 16×16 pixels; the formula of the adaptive cost function is: J=D+cB, where D is the distortion, B is the number of bits, and c is the Lagrange multiplier.

[0062] S4: Collecting playback quality feedback data of the streaming media server, and optimizing and adjusting the dynamic bit rate allocation model based on the playback quality feedback data.

[0063] Specifically, playback quality feedback data of the streaming media server is obtained, wherein the playback quality feedback data includes playback fluency, buffer status and delay parameters.

[0064] Furthermore, a playback quality evaluation system is constructed to perform weighted calculation on playback quality feedback data, generate standardized playback quality scores, and comprehensively evaluate playback effects;

[0065] It should be noted that the weighted calculation is: setting the frame rate weight to 0.3, the buffer occupancy weight to 0.2, the playback delay weight to 0.2, the audio and video synchronization weight to 0.15, and the freeze number weight to 0.15 to generate a standardized playback quality score.

[0066] Furthermore, a feedback optimization mechanism is established based on the playback quality score to optimize and adjust the parameters of the dynamic bitrate allocation model.

[0067] It should be noted that the feedback optimization mechanism is: when the playback quality score is lower than 75 points, the benchmark bit rate in the dynamic bit rate allocation model will be reduced by 15%; when the playback quality score is between 75-85 points, the current parameters will be kept unchanged; when the playback quality score is higher than 85 points and the network bandwidth is sufficient, the benchmark bit rate will be increased, and the increase will not exceed 10% of the current bit rate.

[0068] Specifically, the optimized parameters are fed back to the dynamic bit rate allocation model in real time to optimize the transmission quality of the video data stream.

[0069] Furthermore, the present embodiment also provides a streaming media platform optimization system based on video image processing, including: an extraction module, used to collect image frame sequences in a target video stream, and extract brightness histogram eigenvalues ​​and color saturation data from the image frame sequence; an establishment module, used to calculate the quality score of the image frame sequence based on the brightness histogram eigenvalues ​​and the color saturation data, and establish a dynamic bit rate allocation model; a generation module, used to encode and compress the image frame sequence through the dynamic bit rate allocation model, generate a video data stream with an adaptive bit rate, and transmit it to a streaming media server; an optimization module, used to collect playback quality feedback data of the streaming media server, and optimize and adjust the dynamic bit rate allocation model based on the playback quality feedback data.

[0070] This embodiment also provides a computer device, which is suitable for the case of a streaming media platform optimization method based on video image processing, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the streaming media platform optimization method based on video image processing proposed in the above embodiment.

[0071] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0072] The present embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps: acquiring an image frame sequence in a target video stream, and extracting brightness histogram feature values ​​and color saturation data from the image frame sequence; calculating a quality score of the image frame sequence based on the brightness histogram feature values ​​and the color saturation data, and establishing a dynamic bit rate allocation model; encoding and compressing the image frame sequence through the dynamic bit rate allocation model to generate a video data stream with an adaptive bit rate, and transmitting the stream to a streaming media server; acquiring playback quality feedback data of the streaming media server, and optimizing and adjusting the dynamic bit rate allocation model based on the playback quality feedback data.

[0073] In summary, the present invention extracts the brightness histogram and color saturation data of video frames, establishes a quality scoring model based on a dual-branch neural network, and realizes adaptive video coding compression in combination with dynamic bitrate allocation; the system introduces a closed-loop optimization mechanism based on playback quality feedback, continuously adjusts the bitrate allocation strategy, effectively reduces bandwidth occupancy while ensuring video quality, and improves the user experience and system performance of streaming media services; especially in network fluctuations and complex video content scenarios, it shows good adaptability and stability.

[0074] Example 2

[0075] Reference Figure 3-4 , which is the second embodiment of the present invention, provides a streaming media platform optimization method based on video image processing. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0076] Specifically, we selected a 4K resolution (3840×2160) test video sequence, including four typical scenes: sports events, news broadcasts, online education, and game broadcasts. The video duration of each scene is 30 minutes. The test environment uses an Intel i9-12900K processor, a 64GB RAM server configuration, and a Gigabit Ethernet network environment.

[0077] Furthermore, the test video sequence is frame sampled, and the sampling interval is set to 0.5 seconds to ensure that the key moments of scene changes are captured. The OpenCV library is used to convert the RGB image to the HSV color space, extract the V channel data to construct a 256-level brightness histogram, and calculate the mean, variance, skewness and kurtosis of the histogram as brightness feature values; at the same time, the S channel data is extracted to calculate the regional saturation mean and standard deviation as color saturation features.

[0078] Furthermore, in the dual-branch neural network structure, the first branch uses a three-layer convolution structure (the convolution kernel size is 7×7, 5×5, and 3×3 respectively) to process the brightness feature, and the second branch uses a three-layer fully connected layer (the number of nodes is 1024, 512, and 256 respectively) to process the color feature. The features of the two branches are combined through a weighted fusion layer, and the weight coefficient is optimized by the back propagation algorithm.

[0079] Specifically, the dynamic bitrate allocation model uses a 60-second sliding time window with a window sliding step of 1 second. The initial value of the time attenuation coefficient λ is set to 0.95, the quality adjustment coefficient β is set to 0.8, the response function gain coefficient α is 1.2, and the response function threshold θ is 0.85. The benchmark bit rate is set in the range of 6-15Mbps according to different scene types. In order to verify the system performance, the traditional method (fixed bit rate CBR) and the present invention are used for comparative tests. The system records the performance indicators of the video quality score (VQS) and the end-to-end delay of each solution.

[0080] Further, such as Figure 3 As shown in the video quality score comparison chart, the score of the method of the present invention is always maintained between 90-98 points, the curve trend is stable, and the fluctuation range is small (about ±3 points), indicating that the method can continuously provide high-quality video output. In contrast, the scores of traditional methods generally fluctuate between 80-90 points, and the fluctuation range is large (about ±5 points), indicating that the traditional method has obvious deficiencies in maintaining the stability of video quality.

[0081] Furthermore, Figure 4 As shown in the system delay comparison diagram, the end-to-end delay of the method of the present invention is stably maintained at about 400 milliseconds, with a fluctuation range of about ±50 milliseconds, reflecting excellent system response performance. The system delay of the traditional method is obviously higher, with an average delay of 800 milliseconds and a large fluctuation (about ±100 milliseconds). This high delay will lead to a significant decline in user experience, especially in real-time interactive scenarios.

[0082] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A streaming media platform optimization method based on video image processing, characterized in that: include, Acquire an image frame sequence in a target video stream, and extract brightness histogram feature values ​​and color saturation data from the image frame sequence; Calculating the quality score of the image frame sequence according to the brightness histogram feature value and the color saturation data, and establishing a dynamic bit rate allocation model; The image frame sequence is encoded and compressed by the dynamic bit rate allocation model to generate a video data stream with an adaptive bit rate, and transmitted to a streaming media server; The playback quality feedback data of the streaming media server is collected, and the dynamic bit rate allocation model is optimized and adjusted based on the playback quality feedback data.

2. The method for optimizing a streaming media platform based on video image processing according to claim 1, characterized in that: Collecting playback quality feedback data of the streaming media server, and optimizing and adjusting the dynamic bit rate allocation model based on the playback quality feedback data, including: Acquire playback quality feedback data from the streaming media server, wherein the playback quality feedback data includes playback smoothness, buffering status, and delay parameters; Constructing a playback quality evaluation system, performing weighted calculation on the playback quality feedback data, generating a standardized playback quality score, and comprehensively evaluating the playback effect; Establishing a feedback optimization mechanism based on the playback quality score to optimize and adjust the parameters of the dynamic bit rate allocation model; The optimized parameters are fed back to the dynamic bit rate allocation model in real time to optimize the transmission quality of the video data stream.

3. The method for optimizing a streaming media platform based on video image processing according to claim 2, characterized in that: The method for encoding and compressing the image frame sequence by the dynamic bit rate allocation model is: Dividing the image frame sequence into image groups based on the target bit rate output by the dynamic bit rate allocation model; According to the frames of the image group, a differential compression method is used for compression, and each frame of the image is divided into macroblock units by using a rate-distortion optimization strategy; Selecting the encoding mode of the macroblock unit, introducing an adaptive cost function, and generating a video data stream with an adaptive bit rate; The video data stream is encapsulated and transmitted to a streaming media server via the RTMP protocol.

4. The method for optimizing a streaming media platform based on video image processing according to claim 3, characterized in that: The method for constructing the dynamic bit rate allocation model is as follows: According to the brightness histogram eigenvalues ​​and color saturation data, a double-branch neural network structure is used to establish an image quality assessment model. The features extracted by the dual-branch neural network structure are weighted and fused to obtain the quality score of the image frame sequence; Based on the temporal variation law of the quality score, a dynamic bit rate allocation model is established by adopting a sliding time window method.

5. The method for optimizing a streaming media platform based on video image processing according to claim 4, characterized in that: The dual-branch neural network structure includes a first branch neural network structure and a second branch neural network structure; the first branch neural network structure receives the brightness histogram feature value and extracts brightness distribution features through a three-layer convolution layer; the second branch neural network structure receives the color saturation data and extracts color features through a fully connected layer; The dynamic bit rate allocation model includes a quality threshold judgment module and a bit rate adjustment module; the quality threshold judgment module sets an adaptive adjustment threshold according to the fluctuation range of the quality score; and the bit rate adjustment module dynamically calculates the target bit rate based on the adaptive adjustment threshold.

6. The method for optimizing a streaming media platform based on video image processing according to claim 5, characterized in that: The specific formula of the dynamic bit rate allocation model is as follows: Q(t)=1+β·tanh(Δq / σ) Among them, R(t) is the target bit rate value at time t, R b is the base bit rate, T is the time window length, α is the time decay coefficient, Q(t) is the quality score function, ΔH i is the image entropy change value of the i-th time window, n is the number of sampling windows, is the visual feature response function, β is the quality adjustment coefficient, Δq is the quality score deviation, σ is the quality score standard deviation, v is the normalized image complexity feature value, λ is the response function gain coefficient, and μ is the response function threshold.

7. The method for optimizing a streaming media platform based on video image processing according to claim 6, characterized in that: Acquiring an image frame sequence in a target video stream, and extracting brightness histogram feature values ​​and color saturation data from the image frame sequence, including: Collecting an image frame sequence in a target video stream, sampling the image frame sequence frame by frame according to a preset sampling interval to obtain a key image frame; Convert the key image frame from the RGB color space to the HSV color space, and extract the brightness component in the HSV color space; Counting the pixel value distribution of the brightness component, establishing a brightness histogram, and calculating the brightness histogram feature value; Based on the saturation component in the HSV space, the color space distribution characteristics are analyzed to obtain regional saturation characteristics and color saturation data.

8. A streaming media platform optimization system based on video image processing, based on the streaming media platform optimization method based on video image processing according to any one of claims 1 to 7, characterized in that: Also includes, An extraction module, used for collecting an image frame sequence in a target video stream, and extracting a brightness histogram feature value and color saturation data from the image frame sequence; Establishing a module, used for calculating the quality score of the image frame sequence according to the brightness histogram characteristic value and the color saturation data, and establishing a dynamic bit rate allocation model; A generation module, used to encode and compress the image frame sequence through the dynamic bit rate allocation model, generate a video data stream with an adaptive bit rate, and transmit it to a streaming media server; The optimization module is used to collect the playback quality feedback data of the streaming media server and optimize and adjust the dynamic bit rate allocation model based on the playback quality feedback data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the streaming media platform optimization method based on video image processing described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the streaming media platform optimization method based on video image processing described in any one of claims 1 to 7 are implemented.

Images