Unmanned aerial vehicle visual information frame residual distribution prediction model construction method and prediction method

Abstract

The application provides a method for constructing and predicting a residual distribution prediction model of a UAV visual information frame, which comprises the following steps: obtaining frame residual distribution standard deviation actual values corresponding to a plurality of encoding parameter groups after the UAV visual information frame is encoded by the encoding parameter groups; performing first-order linear regression on the quantization step and the frame residual distribution standard deviation actual values in each encoding parameter group; determining first-order estimation error actual values between first-order frame residual distribution standard deviation prediction values and actual values; performing second-order linear regression on the search range and the first-order estimation error actual values in each encoding parameter group; determining second-order estimation error actual values between the first-order estimation error prediction values and the actual values; performing third-order linear regression on the related frame number and the second-order estimation error actual values in each encoding parameter group; and obtaining the frame residual distribution prediction model according to the three linear regression equations, which reduces the calculation complexity, improves the model construction speed and the distribution prediction precision.

Description

Technical Field

[0001] This invention belongs to the field of electronic information, and specifically relates to a method for constructing a residual distribution prediction model for UAV visual information frames, a method for predicting the residual distribution of UAV visual information frames, a device for constructing a residual distribution prediction model for UAV visual information frames, a device for predicting the residual distribution of UAV visual information frames, an electronic device, and a storage medium. Background Technology

[0002] Visual information from drones is typically acquired by onboard visual acquisition equipment and encoded into video for transmission and display. Raw visual information occupies a significant amount of storage space, which is detrimental to transmission; therefore, encoding of the visual information frames is necessary to reduce space consumption. The commonly used encoding standard is H.264, and the coding rate is controlled by configuring different encoding parameters. During the encoding process, the frame residual distribution directly affects the output bit rate and visual information distortion. Accurately estimating and predicting the frame residual distribution is crucial for optimizing the encoding control of visual information frames.

[0003] In related technologies, exponential models are typically used to estimate and predict the residual distribution of visual information frames. However, due to the limited computing power of the processors on UAVs and the high latency requirements for real-time visual information transmission, this method is limited in its application on UAV flight platforms, and suffers from technical problems such as high computational complexity, slow model building speed, and insufficient distribution prediction accuracy. Summary of the Invention

[0004] In a first aspect, embodiments of the present invention provide a method for constructing a prediction model for the residual distribution of UAV visual information frames, comprising: acquiring multiple encoding parameter groups and determining the actual value of the standard deviation of the frame residual distribution corresponding to each encoding parameter group after encoding the UAV visual information frame, wherein the encoding parameter group includes a quantization step size, a search range, and a number of relevant frames; performing linear regression on the quantization step size and the actual value of the standard deviation of the frame residual distribution in each encoding parameter group to obtain a first-order linear regression equation; calculating the predicted value of the first-order frame residual distribution standard deviation corresponding to each quantization step size based on the first-order linear regression equation, and determining the predicted value of the first-order frame residual distribution standard deviation and the frame residual distribution standard deviation. The actual value of the first-level estimation error between the actual value of the standard deviation of the distribution; linear regression is performed on the search range and the actual value of the first-level estimation error in each coding parameter group to obtain the second-level linear regression equation; the predicted value of the first-level estimation error corresponding to each search range is calculated based on the second-level linear regression equation, and the actual value of the second-level estimation error between the predicted value of the first-level estimation error and the actual value of the first-level estimation error is determined; linear regression is performed on the number of relevant frames and the actual value of the second-level estimation error in each coding parameter group to obtain the third-level linear regression equation; the frame residual distribution prediction model is determined based on the first-level linear regression equation, the second-level linear regression equation, and the third-level linear regression equation.

[0005] In some embodiments, after determining the frame residual distribution prediction model, the method further includes: calculating the predicted standard deviation of the frame residual distribution for each coding parameter group based on the frame residual distribution prediction model; and constructing an error distribution estimation model corresponding to the frame residual distribution prediction model based on the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution.

[0006] In some embodiments, constructing an error distribution estimation model corresponding to the frame residual distribution prediction model based on the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution includes: determining the final error value between the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution; determining an initial Gaussian mixture model; and using the expectation-maximization algorithm to iteratively optimize the initial Gaussian mixture model based on the final error value to obtain a final Gaussian mixture model, wherein the final Gaussian mixture model serves as the error distribution estimation model.

[0007] In some embodiments, obtaining multiple encoding parameter sets includes: obtaining an encoding parameter set, the encoding parameter set including a quantization step size parameter set, a search range parameter set, and a related frame number parameter set; randomly selecting a quantization step size, a search range, and a related frame number from the encoding parameter set to form an encoding parameter set; and repeatedly performing the step of randomly selecting a quantization step size, a search range, and a related frame number from the encoding parameter set to obtain multiple encoding parameter sets.

[0008] Secondly, embodiments of the present invention provide a method for predicting the residual distribution of UAV visual information frames, applied to a frame residual distribution prediction model constructed by the UAV visual information frame residual distribution prediction model construction method described in any one of the first aspects. The method includes: obtaining a set of encoding parameters to be predicted for UAV visual information, the encoding parameter set including quantization step size, search range and number of related frames; inputting the set of encoding parameters to be predicted into the frame residual distribution prediction model to obtain the standard deviation of the frame residual distribution of UAV visual information.

[0009] In some embodiments, after obtaining the standard deviation of the frame residual distribution of the UAV visual information, the method further includes: in the case of obtaining the error distribution estimation model corresponding to the frame residual distribution prediction model, randomly determining the error value of the standard deviation of the frame residual distribution according to the error distribution estimation model; and determining the final value of the standard deviation of the frame residual distribution according to the standard deviation of the frame residual distribution and the error value.

[0010] Thirdly, embodiments of the present invention provide a device for constructing a prediction model for the residual distribution of UAV visual information frames, comprising: a first acquisition module, configured to acquire multiple sets of encoding parameters and determine the actual value of the standard deviation of the frame residual distribution corresponding to each encoding parameter set after encoding the UAV visual information frame, wherein the encoding parameter set includes a quantization step size, a search range, and a number of relevant frames; a first-level regression module, configured to perform linear regression on the quantization step size and the actual value of the standard deviation of the frame residual distribution in each encoding parameter set to obtain a first-level linear regression equation; calculate the predicted value of the standard deviation of the first-level frame residual distribution corresponding to each quantization step size according to the first-level linear regression equation, and determine the predicted value of the standard deviation of the first-level frame residual distribution and the standard deviation of the frame residual distribution. The system comprises three modules: a first-level estimation error actual value; a second-level regression module, used to perform linear regression on the search range and the first-level estimation error actual value in each coding parameter group to obtain a second-level linear regression equation; a second-level linear regression module, used to calculate the first-level estimation error prediction value corresponding to each search range based on the second-level linear regression equation, and to determine the second-level estimation error actual value between the first-level estimation error prediction value and the first-level estimation error actual value; a third-level regression module, used to perform linear regression on the relevant frame number and the second-level estimation error actual value in each coding parameter group to obtain a third-level linear regression equation; and a prediction model construction module, used to determine the frame residual distribution prediction model based on the first-level linear regression equation, the second-level linear regression equation, and the third-level linear regression equation.

[0011] Fourthly, embodiments of the present invention provide a device for predicting the frame residual distribution of UAV visual information, characterized in that it is applied to a frame residual distribution prediction model constructed by the UAV visual information frame residual distribution prediction model construction method described in any one of the first aspects, the device comprising: a second acquisition module, configured to acquire a set of encoding parameters to be predicted for UAV visual information, the encoding parameter set including a quantization step size, a search range, and a number of related frames; and a prediction module, configured to input the set of encoding parameters to be predicted into the frame residual distribution prediction model to obtain the standard deviation of the frame residual distribution of UAV visual information.

[0012] Fifthly, embodiments of the present invention provide an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to execute the program stored in the memory to implement the steps of the UAV visual information frame residual distribution prediction model construction method according to any one of the first aspects or the UAV visual information frame residual distribution prediction method according to the second aspect when executing the program stored in the memory.

[0013] In a sixth aspect, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method for constructing a residual distribution prediction model of UAV visual information frames as described in any of the first aspects or the method for predicting residual distribution of UAV visual information frames as described in the second aspect.

[0014] The beneficial effects of this invention are as follows:

[0015] As can be seen from the above scheme, the UAV visual information frame residual distribution prediction model construction method and prediction method provided in this embodiment of the invention firstly perform a first-level linear regression between the quantization step size and the standard deviation of the frame transform residual distribution. Based on the difference between the predicted value and the actual value of this first-level linear regression, a first-level estimation error is obtained. Then, a second-level linear regression is performed between the first-level estimation error and the search range to obtain a second-level estimation error. Next, a third-level linear regression is performed between the number of relevant frames and the second-level estimation error. Finally, the results of these three linear regressions are linearly combined to complete the construction of the visual information frame transform residual distribution prediction model. This method reduces computational complexity by performing three-level linear regression modeling on encoding parameters such as quantization step size, search range, and number of relevant frames, and then linearly combining the results. This effectively improves the model construction speed and distribution prediction accuracy, providing a solution for rate control and quality evaluation of UAV visual information encoding. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a method for constructing a residual distribution prediction model for UAV visual information frames, provided in an embodiment of the present invention;

[0017] Figure 2 A flowchart illustrating another method for constructing a residual distribution prediction model for UAV visual information frames provided in an embodiment of the present invention;

[0018] Figure 3 A flowchart illustrating a method for predicting the residual distribution of visual information frames from unmanned aerial vehicles (UAVs) according to an embodiment of the present invention;

[0019] Figure 4 This is a flowchart illustrating a method for constructing and predicting the residual distribution of UAV visual information frames, as provided in an embodiment of the present invention.

[0020] Figure 5 This is a schematic diagram of a device for constructing a prediction model for the residual distribution of UAV visual information frames, provided in an embodiment of the present invention.

[0021] Figure 6 This is a schematic diagram of the structure of a UAV visual information frame residual distribution prediction device provided in an embodiment of the present invention;

[0022] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0024] In related technologies, exponential models are typically used to model the standard deviation of the visual information frame residual distribution. However, these models are time-consuming to build and computationally complex, making them poorly applicable on UAV flight platforms with limited computing resources. Furthermore, UAVs are affected by airflow and the environment during flight, causing the collected visual information to frequently experience jitter and severe shaking, which significantly impacts the prediction of the frame residual distribution. Frame residual standard deviation models based on exponential models are prone to overfitting, which in turn reduces the model's prediction accuracy.

[0025] To address the aforementioned technical problems, the technical concept of this invention is as follows: a three-level linear regression method is used to model encoding parameters such as quantization step size, search range, and number of related frames, and then the results of these three linear regressions are linearly added together to complete the construction of a visual information frame residual distribution prediction model.

[0026] Figure 1 This is a flowchart illustrating a method for constructing a residual distribution prediction model for UAV visual information frames, provided in an embodiment of the present invention. Figure 1 As shown, the construction method includes:

[0027] Step S101: Obtain multiple encoding parameter groups and determine the actual value of the standard deviation of the frame residual distribution corresponding to each encoding parameter group after encoding the UAV visual information frame. The encoding parameter group includes quantization step size, search range and number of related frames.

[0028] Specifically, assume that M (M>2) encoding parameter sets are obtained, each encoding parameter set including quantization step size Q, search range λ, and number of relevant frames θ, which can be denoted as {Q,λ,θ}. i Let i = 1, 2, ..., M; iterate through all the encoding parameter groups and encode them, i.e., use {Q, λ, θ}. i The visual information frames of the UAV are encoded, and the actual value of the standard deviation σ of the frame transform residual distribution of the corresponding encoding parameter group is obtained and stored. i .

[0029] In some embodiments, obtaining multiple encoding parameter groups in step S101 includes: obtaining an encoding parameter set, which includes a quantization step size parameter set, a search range parameter set, and a related frame number parameter set; randomly selecting a quantization step size, a search range, and a related frame number from the encoding parameter set to form an encoding parameter group; and repeatedly performing the step of randomly selecting a quantization step size, a search range, and a related frame number from the encoding parameter set to obtain multiple encoding parameter groups.

[0030] Specifically, a set of encoding parameters is set, including a quantization step size parameter set, a search range parameter set, and a related frame number parameter set; a quantization step size Q is randomly selected from the quantization step size parameter set, a search range λ is randomly selected from the search range parameter set, and a related frame number θ is randomly selected from the related frame number parameter set to form an encoding parameter group; the above steps are repeated to construct multiple encoding parameter groups.

[0031] Step S102: Perform linear regression on the actual values ​​of the quantization step size and frame residual distribution standard deviation in each coding parameter group to obtain the first-order linear regression equation.

[0032] Specifically, for all groups of coding parameters, the quantization step size Q and its corresponding actual value σ of the frame transform residual standard deviation are... i Performing linear regression yields the first-order linear regression equation, as shown below:

[0033]

[0034] in, denoted by , where represents the predicted standard deviation of the first-order frame transform residual for the corresponding quantization step size, Q represents the quantization step size, and a and b represent the parameters of the first-order linear regression equation.

[0035] It should be noted that experiments have shown that the quantization step size has the greatest impact on the residual distribution of visual information frame transformation among the coding parameters. Therefore, in this embodiment, a first-order linear regression equation for the quantization step size is first constructed.

[0036] Step S103: Calculate the predicted standard deviation of the first-level frame residual distribution corresponding to each quantization step size according to the first-level linear regression equation, and determine the actual value of the first-level estimation error between the predicted standard deviation of the first-level frame residual distribution and the actual value of the standard deviation of the frame residual distribution.

[0037] Specifically, for each group of encoded parameters, the quantization step size Q i Substituting into formula (1), we obtain the predicted standard deviation of the corresponding first-order frame residual distribution. and the actual value of the standard deviation σ of the frame transform residual distribution. i By comparison, the actual value of the first-order estimation error, e, is obtained. Q,iAfter traversing the quantization step size in all encoding parameter sets, the set of actual values ​​of the first-level estimation error {e} is obtained. Q}

[0038] Step S104: Perform linear regression on the search range and the actual value of the first-level estimation error in each group of coding parameters to obtain the second-level linear regression equation.

[0039] Specifically, a linear regression is performed on the search range and its corresponding first-level estimation error for all coding parameter groups to construct a second-level linear regression equation, as shown below:

[0040]

[0041] in, λ represents the predicted value of the first-level estimation error, λ represents the search range, and c and d represent the parameters of the second-level linear regression equation.

[0042] Step S105: Calculate the predicted value of the first-level estimation error corresponding to each search range based on the second-level linear regression equation, and determine the actual value of the second-level estimation error between the predicted value of the first-level estimation error and the actual value of the first-level estimation error.

[0043] Specifically, for the search range λ in each group of encoded parameters i Substituting into formula (2), we obtain the corresponding first-order estimation error prediction value. and the actual value of the first-order estimation error e Q,i By comparison, the actual value of the second-order estimation error e is obtained. Q,λ,i After traversing the search range in all encoding parameter sets, the set of actual values ​​of the second-order estimation error {e} is obtained. Q,λ,i}

[0044] Step S106: Perform linear regression on the relevant frame number and the actual value of the second-level estimation error in each coding parameter group to obtain the third-level linear regression equation.

[0045] Specifically, a linear regression is performed on the relevant frame number in all coding parameter groups and their corresponding actual second-level estimation errors to construct a third-level linear regression equation, as shown below:

[0046]

[0047] in, θ represents the predicted value of the second-order estimation error, g and h represent the parameters of the third-order linear regression equation.

[0048] Step S107: Determine the frame residual distribution prediction model based on the first-level linear regression equation, the second-level linear regression equation, and the third-level linear regression equation.

[0049] Specifically, by linearly superimposing formulas (1), (2), and (3), the frame residual distribution prediction model (i.e., the frame residual distribution standard deviation model) is constructed, as shown below:

[0050]

[0051] in, This represents the predicted standard deviation of the frame transform residual distribution.

[0052] The UAV visual information frame residual distribution prediction model construction method provided in this embodiment of the invention first performs a first-level linear regression between the quantization step size and the standard deviation of the frame transform residual distribution. Based on the difference between the predicted value and the actual value of this first-level linear regression, a first-level estimation error is obtained. Then, a second-level linear regression between the search range and the first-level estimation error is performed. Based on the comparison between the predicted value of this second-level linear regression and the first-level estimation error, a second-level estimation error is obtained. Then, a third-level linear regression between the number of relevant frames and the second-level estimation error is performed. Finally, the three linear regressions are linearly added to complete the construction of the visual information frame transform residual distribution prediction model. That is, this embodiment solves the problems of high computational complexity and long computation time of traditional visual information frame transform residual distribution prediction models, as well as the problems of overfitting and insufficient accuracy when applied to UAV flight platforms, by adopting a three-level linear regression method.

[0053] Based on the aforementioned embodiments, Figure 2 This is a flowchart illustrating another method for constructing a residual distribution prediction model for UAV visual information frames, provided in an embodiment of the present invention. Figure 2 As shown, the construction method includes:

[0054] Step S201: Obtain multiple encoding parameter groups and determine the actual value of the standard deviation of the frame residual distribution corresponding to each encoding parameter group after encoding the UAV visual information frame. The encoding parameter group includes quantization step size, search range and number of related frames.

[0055] Step S202: Perform linear regression on the actual values ​​of the quantization step size and frame residual distribution standard deviation in each coding parameter group to obtain the first-order linear regression equation.

[0056] Step S203: Calculate the predicted standard deviation of the first-level frame residual distribution corresponding to each quantization step size according to the first-level linear regression equation, and determine the actual value of the first-level estimation error between the predicted standard deviation of the first-level frame residual distribution and the actual value of the standard deviation of the frame residual distribution.

[0057] Step S204: Perform linear regression on the search range and the actual value of the first-level estimation error in each group of coding parameters to obtain the second-level linear regression equation.

[0058] Step S205: Calculate the predicted value of the first-level estimation error corresponding to each search range based on the second-level linear regression equation, and determine the actual value of the second-level estimation error between the predicted value of the first-level estimation error and the actual value of the first-level estimation error.

[0059] Step S206: Perform linear regression on the relevant frame number and actual value of the second-level estimation error in each coding parameter group to obtain the third-level linear regression equation.

[0060] Step S207: Determine the frame residual distribution prediction model based on the first-level linear regression equation, the second-level linear regression equation, and the third-level linear regression equation.

[0061] Step S208: Calculate the predicted standard deviation of the frame residual distribution for each coding parameter group based on the frame residual distribution prediction model.

[0062] Step S209: Construct an error distribution estimation model corresponding to the frame residual distribution prediction model based on the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution.

[0063] It should be noted that steps S201-S207 in this embodiment are similar to steps S101-S107 in the previous embodiment, and will not be described again here.

[0064] The difference from the previous embodiments is that, in order to further improve the prediction accuracy of frame residual distribution of unmanned visual information, in this embodiment, the predicted value of the standard deviation of frame residual distribution corresponding to each coding parameter group is calculated according to the frame residual distribution prediction model; and the error distribution estimation model corresponding to the frame residual distribution prediction model is constructed according to the predicted value of the standard deviation of frame residual distribution and the actual value of the standard deviation of frame residual distribution.

[0065] Specifically, for each encoding parameter group {Q,λ,θ} i Substituting these values ​​into formula (4), we obtain the predicted standard deviation of the corresponding frame transform residual distribution. according to And the actual value of the standard deviation of the frame transform residual distribution σ i Construct an error distribution estimation model corresponding to the standard deviation model of the residual distribution of this frame.

[0066] In some embodiments, step S209 includes: determining the final error value between the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution; determining an initial Gaussian mixture model; using the expectation-maximization algorithm to iteratively optimize the initial Gaussian mixture model based on the final error value to obtain a final Gaussian mixture model, wherein the final Gaussian mixture model is the error distribution estimation model.

[0067] Specifically, the third-level linear regression causes the error of the frame residual distribution prediction model to exhibit a mixture Gaussian distribution. In this embodiment, firstly... With σ i In comparison, the final error value e is calculated. σ Then, determine the initial Gaussian mixture model, that is, set the initial values ​​μ of the Gaussian mixture model parameters. k and σ k μ k σ represents the initial value of the mean of the Gaussian mixture model. k Let represent the initial value of the variance of the Gaussian mixture model, and set the maximum number of iterations N. Then, the Expectation-Maximization algorithm (EM algorithm) is used to analyze the error of the frame residual distribution standard deviation model. By iteratively solving E and M, the convergent final Gaussian mixture model (μ) is obtained. c , σ c ).

[0068] Error estimation is performed based on the error distribution estimation model, and the error estimation results are combined with the three-level linear regression model to achieve the estimation of the visual information frame transformation residual distribution.

[0069] Based on the aforementioned embodiments, the predicted standard deviation of the frame residual distribution corresponding to each coding parameter group is calculated according to the frame residual distribution prediction model; an error distribution estimation model corresponding to the frame residual distribution prediction model is constructed based on the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution; that is, for the estimation error of the linear regression model, machine learning methods are used to further improve its accuracy.

[0070] Figure 3 This is a flowchart illustrating a method for predicting the residual distribution of UAV visual information frames according to an embodiment of the present invention. It is applied to the frame residual distribution prediction model constructed by the method for constructing the UAV visual information frame residual distribution prediction model in the aforementioned embodiments. Figure 3 As shown, the prediction method includes:

[0071] Step S301: Obtain the set of encoding parameters to be predicted for the UAV visual information. The set of encoding parameters includes the quantization step size, search range, and number of related frames.

[0072] Step S302: Input the group of encoding parameters to be predicted into the frame residual distribution prediction model to obtain the standard deviation of the frame residual distribution of UAV visual information.

[0073] Specifically, for the encoding parameter set used to predict the standard deviation of frame residual distribution of UAV visual information, it is substituted into the frame residual distribution standard deviation model corresponding to formula (4) to obtain the predicted value of frame residual distribution standard deviation. The frame residual distribution frame residual distribution can be analyzed and predicted based on the predicted value of frame residual distribution standard deviation.

[0074] Continue to refer to Figure 3 As shown, in some embodiments, after step S302, the following steps are further included:

[0075] Step S303: After obtaining the error distribution estimation model corresponding to the frame residual distribution prediction model, randomly determine the error value of the frame residual distribution standard deviation according to the error distribution estimation model.

[0076] Step S304: Determine the final value of the standard deviation of the frame residual distribution based on the standard deviation value and the error value of the frame residual distribution.

[0077] Specifically, the error estimation distribution model is a convergent final Gaussian mixture model (μ c , σ c ), from which error e is randomly selected c , will e c Substituting into formula (5), calculate the final value of the standard deviation of the visual information frame transformation residual.

[0078]

[0079] Based on the aforementioned embodiments, a set of encoding parameters to be predicted for UAV visual information is obtained, including quantization step size, search range, and number of relevant frames. This set of parameters is then input into the frame residual distribution prediction model to obtain the standard deviation of the frame residual distribution of the UAV visual information. The three-level linear regression model employed improves computational efficiency and reduces computational complexity. Furthermore, given the error distribution estimation model corresponding to the frame residual distribution prediction model, the error value of the frame residual distribution standard deviation is randomly determined based on the error distribution estimation model. The final value of the frame residual distribution standard deviation is determined based on the standard deviation and the error value. By combining the error estimation result with the three-level linear regression model, the accuracy of the visual information frame transform residual distribution prediction is further improved.

[0080] Figure 4 This is a flowchart illustrating a method for constructing and predicting the residual distribution of UAV visual information frames, as provided in an embodiment of the present invention. Now, in conjunction with... Figure 4 The embodiments of the present invention will be described in detail below:

[0081] (1) Set the encoding parameter set: including the quantization step size parameter set, the search range parameter set, and the related frame number parameter set.

[0082] (2) Randomly select encoding parameters and construct multiple encoding parameter groups: randomly select a quantization step size from the quantization step size parameter set, randomly select a search range from the search range parameter set, and randomly select a related frame number from the related frame number parameter set to form an encoding parameter group. Repeat the above process to construct multiple encoding parameter groups.

[0083] (3) Encode all encoding parameter groups to obtain the actual value of the standard deviation of the frame transform residual distribution: Encode the UAV information frame using each encoding parameter group, determine the difference between the UAV information frame before and after encoding, and further obtain the actual value of the standard deviation of the frame transform residual distribution.

[0084] (4) First-order linear regression: Perform linear regression on the actual values ​​of the quantization step size and the standard deviation of the frame residual distribution in all coding parameter groups to obtain the first-order linear regression equation; substitute the quantization step size in each coding parameter group into the first-order linear regression equation to obtain the corresponding first-order frame residual distribution standard deviation prediction value, and determine the actual value of the first-order estimation error between the first-order frame residual distribution standard deviation prediction value and the actual value of the frame residual distribution standard deviation.

[0085] (5) Secondary linear regression: Perform linear regression on the search range and the actual value of the first-level estimation error in all coding parameter groups to obtain the secondary linear regression equation; substitute the search range in each coding parameter group into the secondary linear regression equation to obtain the corresponding predicted value of the first-level estimation error, and determine the actual value of the secondary estimation error between the predicted value of the first-level estimation error and the actual value of the first-level estimation error.

[0086] (6) Third-level linear regression: Perform linear regression on the relevant frame number and the actual value of the second-level estimation error in all coding parameter groups to obtain the third-level linear regression equation.

[0087] (7) Model construction is completed by superposition: The first-level linear regression equation, the second-level linear regression equation and the third-level linear regression equation are linearly added together to obtain the frame residual distribution standard deviation model.

[0088] (8) Initialize the error Gaussian distribution and EM iterative algorithm parameters: Set the initial values ​​of the mean and variance of the Gaussian mixture model, as well as the maximum number of iterations of the EM algorithm.

[0089] (9) Iterative solution to generate error Gaussian distribution: Substitute each group of coding parameters into the frame residual distribution standard deviation model to obtain the predicted value of the frame residual distribution standard deviation, and compare it with the actual value of the frame residual distribution standard deviation to determine the error value; use the EM algorithm to iteratively solve based on the error value to obtain a converged error Gaussian distribution.

[0090] (10) Randomly generate error distribution based on error Gaussian distribution: Randomly select error from converged error Gaussian distribution.

[0091] (11) Combine model building and error estimation to predict the standard deviation of visual information frame transformation residuals: Input the coding parameters to be predicted into the constructed frame residual distribution prediction model to obtain the standard deviation of frame residual distribution. Combine the standard deviation of frame residual distribution with the error in (10) to obtain the final value of the standard deviation of frame transformation residuals.

[0092] In summary, this embodiment employs a three-level linear regression method to model encoding parameters such as quantization step size, search range, and number of relevant frames. First, a first-level linear regression is performed between the quantization step size and the standard deviation of the frame transform residual distribution. The difference between the predicted and actual values ​​from this first-level linear regression yields the first-level estimation error. Then, a second-level linear regression is performed between the first-level estimation error and the search range to obtain the second-level estimation error. Next, a third-level linear regression is performed between the number of relevant frames and the second-level estimation error. Finally, the results of these three linear regressions are linearly combined to construct the visual information frame transform residual distribution prediction model. Furthermore, considering the Gaussian mixture distribution of the model error caused by the three-level linear regression, an expectation-maximization algorithm is used to iteratively solve for the model error, ultimately obtaining the error distribution estimation model. Error estimation is performed based on the error distribution estimation model, and the error estimation results are combined with the three-level linear regression model to achieve visual information frame transform residual distribution estimation. This reduces computational complexity, improves modeling speed and distribution prediction accuracy, and is suitable for UAV flight platforms.

[0093] Figure 5 This is a schematic diagram of a device for constructing a prediction model for the residual distribution of UAV visual information frames, provided in an embodiment of the present invention. Figure 5 As shown, the construction apparatus includes:

[0094] The first acquisition module 501 is used to acquire multiple encoding parameter groups and determine the actual value of the standard deviation of the frame residual distribution corresponding to each encoding parameter group after encoding the UAV visual information frame. The encoding parameter group includes quantization step size, search range and number of related frames.

[0095] The first-level regression module 502 is used to perform linear regression on the actual values ​​of the quantization step size and the standard deviation of the frame residual distribution in each coding parameter group to obtain the first-level linear regression equation.

[0096] The predicted standard deviation of the first-level frame residual distribution corresponding to each quantization step size is calculated based on the first-level linear regression equation, and the actual value of the first-level estimation error between the predicted standard deviation of the first-level frame residual distribution and the actual value of the standard deviation of the frame residual distribution is determined.

[0097] The secondary regression module 503 is used to perform linear regression on the search range and the actual value of the first-level estimation error in each group of encoded parameters to obtain the secondary linear regression equation.

[0098] The predicted value of the first-level estimation error corresponding to each search range is calculated based on the second-level linear regression equation, and the actual value of the second-level estimation error between the predicted value of the first-level estimation error and the actual value of the first-level estimation error is determined.

[0099] The third-level regression module 504 is used to perform linear regression on the relevant frame number and the actual value of the second-level estimation error in each coding parameter group to obtain the third-level linear regression equation.

[0100] The prediction model building module 505 is used to determine the frame residual distribution prediction model based on the first-level linear regression equation, the second-level linear regression equation, and the third-level linear regression equation.

[0101] In some embodiments, the apparatus further includes an error model construction module 506, which is used to calculate the predicted standard deviation of the frame residual distribution for each coding parameter group based on the frame residual distribution prediction model.

[0102] An error distribution estimation model corresponding to the frame residual distribution prediction model is constructed based on the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution.

[0103] In some embodiments, the error model construction module 506 is specifically used for:

[0104] Determine the final error value between the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution;

[0105] Determine the initial Gaussian mixture model;

[0106] The initial Gaussian mixture model is iteratively optimized using the expectation-maximization algorithm based on the final error value to obtain the final Gaussian mixture model, which serves as the error distribution estimation model.

[0107] In some embodiments, the first acquisition module 501 is specifically used for:

[0108] Obtain the encoding parameter set, which includes a quantization step size parameter set, a search range parameter set, and a related frame number parameter set;

[0109] Randomly select the quantization step size, search range, and number of relevant frames from the set of encoding parameters to form an encoding parameter group;

[0110] Repeat the steps of randomly selecting the quantization step size, search range, and relevant frame number from the encoding parameter set to obtain multiple encoding parameter groups.

[0111] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process and corresponding beneficial effects of the above-described UAV visual information frame residual distribution prediction model construction device can be referred to the corresponding process in the aforementioned UAV visual information frame residual distribution prediction model construction method example, and will not be repeated here.

[0112] Figure 6 This is a schematic diagram of a UAV visual information frame residual distribution prediction device provided in an embodiment of the present invention. It is applied to the frame residual distribution prediction model constructed by the UAV visual information frame residual distribution prediction model construction method in the aforementioned embodiments, such as... Figure 6 As shown, the construction apparatus includes:

[0113] The second acquisition module 601 is used to acquire a set of encoding parameters to be predicted for UAV visual information, the encoding parameter set including quantization step size, search range and number of related frames;

[0114] The prediction module 602 is used to input the set of encoding parameters to be predicted into the frame residual distribution prediction model to obtain the standard deviation of the frame residual distribution of UAV visual information.

[0115] In some embodiments, the prediction module 602 is further configured to:

[0116] Given the error distribution estimation model corresponding to the frame residual distribution prediction model, the error value of the frame residual distribution standard deviation is randomly determined according to the error distribution estimation model.

[0117] The final value of the standard deviation of the frame residual distribution is determined based on the standard deviation and error value of the frame residual distribution.

[0118] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process and corresponding beneficial effects of the UAV visual information frame residual distribution prediction device described above can be referred to the corresponding process in the aforementioned UAV visual information frame residual distribution prediction method example, and will not be repeated here.

[0119] like Figure 7 As shown, this embodiment of the invention provides an electronic device, including a processor 701, a communication interface 702, a memory 703, and a communication bus 704, wherein the processor 701, the communication interface 702, and the memory 703 communicate with each other via the communication bus 704.

[0120] Memory 703 is used to store computer programs;

[0121] In one embodiment of the present invention, when the processor 701 executes the program stored in the memory 703, it implements the steps of the UAV visual information frame residual distribution prediction model construction method or the UAV visual information frame residual distribution prediction method provided in any of the foregoing method embodiments.

[0122] The electronic device provided in this embodiment of the invention has a similar implementation principle and technical effect to the above embodiments, and will not be described again here.

[0123] The aforementioned memory 703 can be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 703 has storage space for program code used to perform any of the method steps described above. For example, the storage space for program code may include individual program codes for implementing the various steps in the methods described above. This program code can be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, optical discs (CDs), memory cards, or floppy disks. Such computer program products are typically portable or fixed storage units. The storage unit may have storage segments or storage spaces arranged similarly to memory 703 in the aforementioned electronic device. The program code may be compressed, for example, in a suitable form. Typically, the storage unit includes programs for performing the method steps according to embodiments of the invention, i.e., code that can be read by a processor such as 701, which, when run by the electronic device, causes the electronic device to perform the various steps in the methods described above.

[0124] Embodiments of the present invention also provide a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for constructing a residual distribution prediction model for UAV visual information frames or the method for predicting the residual distribution of UAV visual information frames as described above.

[0125] The computer-readable storage medium may be included in the device / apparatus described in the above embodiments; or it may exist independently and not assembled into the device / apparatus. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of the present invention.

[0126] According to embodiments of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0127] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0128] The above are preferred embodiments of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for constructing a prediction model for the residual distribution of visual information frames from unmanned aerial vehicles (UAVs), characterized in that, include: Multiple encoding parameter groups are obtained, and the actual value of the standard deviation of the frame residual distribution corresponding to each encoding parameter group after encoding the UAV visual information frame is determined. The encoding parameter group includes quantization step size, search range and number of related frames. Linear regression was performed on the actual values ​​of the quantization step size and the standard deviation of the frame residual distribution in each coding parameter group to obtain the first-order linear regression equation; The predicted standard deviation of the first-level frame residual distribution corresponding to each quantization step size is calculated based on the first-level linear regression equation, and the actual value of the first-level estimation error between the predicted standard deviation of the first-level frame residual distribution and the actual value of the standard deviation of the frame residual distribution is determined. Linear regression was performed on the search range and the actual value of the first-level estimation error in each group of coding parameters to obtain the second-level linear regression equation; The predicted value of the first-level estimation error corresponding to each search range is calculated based on the second-level linear regression equation, and the actual value of the second-level estimation error between the predicted value of the first-level estimation error and the actual value of the first-level estimation error is determined. Linear regression was performed on the relevant frame number and the actual value of the second-level estimation error in each coding parameter group to obtain the third-level linear regression equation; The frame residual distribution prediction model is determined based on the first-level linear regression equation, the second-level linear regression equation, and the third-level linear regression equation.

2. The method according to claim 1, characterized in that, After determining the frame residual distribution prediction model, the method further includes: The predicted standard deviation of the frame residual distribution for each coding parameter group is calculated based on the frame residual distribution prediction model. An error distribution estimation model corresponding to the frame residual distribution prediction model is constructed based on the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution.

3. The method according to claim 2, characterized in that, The step of constructing the error distribution estimation model corresponding to the frame residual distribution prediction model based on the predicted value of the frame residual distribution standard deviation and the actual value of the frame residual distribution standard deviation includes: Determine the final error value between the predicted standard deviation of the frame residual distribution and the actual standard deviation of the frame residual distribution; Determine the initial Gaussian mixture model; The initial Gaussian mixture model is iteratively optimized using the expectation-maximization algorithm based on the final error value to obtain the final Gaussian mixture model, which serves as the error distribution estimation model.

4. The method according to any one of claims 1-3, characterized in that, The acquisition of multiple encoding parameter groups includes: Obtain the encoding parameter set, which includes a quantization step size parameter set, a search range parameter set, and a related frame number parameter set; Randomly select the quantization step size, search range, and number of relevant frames from the set of encoding parameters to form an encoding parameter group; Repeat the steps of randomly selecting the quantization step size, search range, and relevant frame number from the encoding parameter set to obtain multiple encoding parameter groups.

5. A method for predicting the residual distribution of visual information frames from unmanned aerial vehicles (UAVs), characterized in that, The method is applied to a frame residual distribution prediction model constructed by the UAV visual information frame residual distribution prediction model construction method according to any one of claims 1-4, the method comprising: A set of encoding parameters to be predicted for obtaining the visual information of the UAV, the set of encoding parameters including quantization step size, search range and number of relevant frames; The set of encoding parameters to be predicted is input into the frame residual distribution prediction model to obtain the standard deviation of the frame residual distribution of UAV visual information.

6. The method according to claim 5, characterized in that, After obtaining the standard deviation of the frame residual distribution of the UAV visual information, the method further includes: Given the error distribution estimation model corresponding to the frame residual distribution prediction model, the error value of the frame residual distribution standard deviation is randomly determined according to the error distribution estimation model. The final value of the standard deviation of the frame residual distribution is determined based on the standard deviation and error value of the frame residual distribution.

7. A device for constructing a prediction model for the residual distribution of visual information frames from unmanned aerial vehicles (UAVs), characterized in that, include: The first acquisition module is used to acquire multiple encoding parameter groups and determine the actual value of the standard deviation of the frame residual distribution corresponding to each encoding parameter group after encoding the UAV visual information frame. The encoding parameter group includes quantization step size, search range and number of related frames. The first-level regression module is used to perform linear regression on the actual values ​​of the quantization step size and the standard deviation of the frame residual distribution in each coding parameter group to obtain the first-level linear regression equation. The predicted standard deviation of the first-level frame residual distribution corresponding to each quantization step size is calculated based on the first-level linear regression equation, and the actual value of the first-level estimation error between the predicted standard deviation of the first-level frame residual distribution and the actual value of the standard deviation of the frame residual distribution is determined. The secondary regression module is used to perform linear regression on the search range and the actual value of the primary estimation error in each group of encoded parameters to obtain the secondary linear regression equation. The predicted value of the first-level estimation error corresponding to each search range is calculated based on the second-level linear regression equation, and the actual value of the second-level estimation error between the predicted value of the first-level estimation error and the actual value of the first-level estimation error is determined. The three-level regression module is used to perform linear regression on the relevant frame number and the actual value of the second-level estimation error in each coding parameter group to obtain the three-level linear regression equation. The prediction model building module is used to determine the frame residual distribution prediction model based on the first-level linear regression equation, the second-level linear regression equation, and the third-level linear regression equation.

8. A device for predicting the residual distribution of visual information frames from unmanned aerial vehicles (UAVs), characterized in that, The apparatus is applied to a frame residual distribution prediction model constructed by the UAV visual information frame residual distribution prediction model construction method according to any one of claims 1-4, the apparatus comprising: The second acquisition module is used to acquire a set of encoding parameters to be predicted from the UAV's visual information. The encoding parameter set includes quantization step size, search range and number of relevant frames. The prediction module is used to input the set of encoding parameters to be predicted into the frame residual distribution prediction model to obtain the standard deviation of the frame residual distribution of UAV visual information.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When executing a program stored in memory, the processor implements the steps of the method for constructing a residual distribution prediction model for UAV visual information frames as described in any one of claims 1-4, or the method for predicting residual distribution of UAV visual information frames as described in claim 5 or 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the UAV visual information frame residual distribution prediction model construction method as described in any one of claims 1-4 or the UAV visual information frame residual distribution prediction method as described in claim 5 or 6.

Images