Screen content video code rate control initial quantization parameter selection method

Abstract

The invention discloses a method for selecting an initial quantization parameter for screen content video code rate control. The method comprises the following steps: determining a video category, determining a video motion speed degree, determining video texture complexity, determining initial frame texture complexity, determining each pixel of a video bit, determining an average bit of each frame of the video, normalizing data, determining a minimum Euclidean distance, determining a minimum absolute value difference, determining a weight factor and a parameter, and determining a normalized bit of an initial frame; the method comprises the following steps: determining a video category, determining a weight factor and a parameter, determining an initial frame bit, determining a Lagrange factor, limiting the Lagrange factor, determining an initial frame quantization parameter, limiting the initial frame quantization parameter, and carrying out coding, and comprises the steps of determining the video category, determining the weight factor and the parameter, and determining the initial frame bit. And the code rate control performance is relatively low. The method has the advantages of high bit control precision, good rate distortion performance and the like, and can be applied to the technical field of screen content video code rate control.

Description

Technical Field

[0001] The invention belongs to the technical field of video encoding, and in particular relates to the selection of initial quantization parameters for screen content video bit rate control. Background Art

[0002] As video develops towards high resolution, high frame rate, and multiple color components, the contradiction between the rapidly growing amount of video data and channel bandwidth is becoming increasingly acute. In order to solve this problem, research on effective video coding and rate control technology has been widely concerned. The core goal of video coding is to achieve higher reconstructed video quality at a lower coding bit rate. The size of the coding bit rate and the channel bandwidth are both time-varying. Rate control is dedicated to studying how to effectively transmit the time-varying coding bit rate on the time-varying channel. Therefore, rate control is the key to determining the practicality of video codecs. In rate control, since a large number of frames directly or indirectly refer to the first frame during the video encoding process, the method of selecting the quantization parameter for the first frame of the video, that is, the initial quantization parameter selection method, will significantly affect the performance of rate control. With the rapid development of multimedia technology, screen content videos generated or rendered by computers are widely used, such as computer desktop sharing, distance education, document presentation, etc. Correspondingly, the initial quantization parameter selection method for screen content videos came into being.

[0003] The currently proposed method for selecting the initial quantization parameters of screen content video has the following problems:

[0004] Only the spatial content characteristics of the screen content video are considered, ignoring the significant impact of the temporal content characteristics on the selection of initial quantization parameters. The spatial content characteristics of the video mainly reflect the complexity of the video texture, and the temporal content characteristics of the video mainly reflect the speed of the video motion. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a method for selecting initial quantization parameters for rate control of screen content video with good rate control performance, high bit control accuracy and good rate distortion performance.

[0006] The technical solution adopted to solve the above technical problems consists of the following steps:

[0007] (1) Determine the video category

[0008] 1) Determine block difference and frame difference

[0009] Divide each frame in the video into pieces of size Non-overlapping pixel blocks, τ∈{1,2,…,h}, h represents the height of the video, w represents the width of the video, and the values ​​of h and w are finite positive integers.

[0010] According to formula (1), the average value of the sum of pixel differences of the δth block in the xth frame is determined, that is, the block difference σ of the δth block in the xth frame x,δ :

[0011]

[0012] Where x∈{1,2,…,N fra},N fra Indicates the total number of video frames, N fra The value is a finite positive integer, δ∈{1,2,…,N blo},N blo Indicates the total number of pixel blocks per frame of the video, N blo The value is a finite positive integer, m and n represent the position coordinates of the pixel in the pixel block, m∈{1,2,…,τ}, u x,δ,m,n represents the pixel value of the mth row and nth column in the δth block of the xth frame of the video, u x,δ,m,n ∈{0,1,…,Γ},u x-1,δ,m,n represents the pixel value of the mth row and nth column in the δth block of the x-1th frame of the video, u x-1,δ,m,n ∈{0,1,...,Γ}, Γ is a finite positive integer.

[0013] According to formula (2), the average value of all block differences in the xth frame is determined, that is, the frame difference σ x :

[0014]

[0015] 2) Determine the video scene switching frame

[0016] ① Determine the difference between adjacent frames

[0017] Determine the frame difference σ of the x-1th frame according to formula (3): γ :

[0018]

[0019] Where, γ∈{1,2,…,N fra -1},σ γ,δ Represents the block difference of the δth block in the γth frame.

[0020] Determine the frame difference σ of the x+1th frame according to formula (4): Ψ :

[0021]

[0022] Where Ψ∈{2,…,N fra},σ Ψ,δ It represents the block difference of the δth block in the Ψth frame.

[0023] ②Determine the scene switching frame

[0024] Arrange all frame differences in ascending order to obtain the set H, and determine the average value σ of the first φ elements of the set H according to formula (5): ele :

[0025]

[0026] Where φ∈{1,2,…,N fra -1},σ ρ represents the frame difference corresponding to the ρth element of the set H, σ x -σ γ >σ ele And σ x -σ Ψ >σ ele , determine the xth frame as the scene switching frame.

[0027] 3) Determine the category to which the video belongs

[0028] If the video contains scene switching frames, execute steps ①-⑥ to determine the category to which the video belongs. If the video does not contain scene switching frames, directly execute steps ⑤-⑥ to determine the category to which the video belongs.

[0029] ①Determine the average value of the remaining frame difference

[0030] Delete all scene switching frames and determine the average value σ of the remaining frame differences according to formula (6): rem :

[0031]

[0032]

[0033] Where N fra,rem Indicates the number of frames remaining after deleting all scene switching frames from the video, N fra,rem ∈{1,2,…,N fra -1},r∈{1,2,...,N fra,rem},σ r represents the frame difference of the rth frame, σ r,δ represents the block difference of the δth block in the rth frame, u r,δ,m,n represents the pixel value of the mth row and nth column of the δth block of the rth frame, u r,δ,m,n ∈{0,1,...,Γ},u r-1,δ,m,n represents the pixel value of the mth row and nth column of the δth block in the r-1th frame, u r-1,δ,m,n ∈{0,1,...,Γ}.

[0034] ②Determine the average value of all frame differences

[0035] Determine the average value σ of all frame differences according to formula (7): vid :

[0036]

[0037] ③Determine the ratio of frame difference to block difference

[0038] According to formula (8), the frame difference of all scene switching frames in the video is determined as σ vid The ratio θ:

[0039] θ=|σ vid -σ rem | / σ vid ×100% (8)

[0040] According to formula (9), the ratio ω of the number of pixel blocks with a block difference of 0 in all scene switching frames of the video to the total number of pixel blocks is determined as follows:

[0041] ω=N SC,blo,0 / N SC,blo ×100% (9)Where N SC,blo,0 Indicates the number of pixel blocks with a block difference of 0 in all scene switching frames, N SC,blo,0 The value is a finite non-negative integer, N SC,blo Represents the total number of pixel blocks in all scene switching frames, N SC,blo The value is a finite positive integer.

[0042] ④ Initially determine the category of the video

[0043] If θ>Θ and ω<Θ, the video is determined to be a scene switching video, and steps ⑤ and ⑥ are no longer executed; if θ>Θ and ω<Θ are not satisfied, the video is determined to be a large-area motion video or a small-area motion video, Θ is the video category classification threshold, Θ∈(0,1).

[0044] ⑤ Determine the category of the video again

[0045] The video does not contain frames with a frame difference greater than 0, the video is in a static state, which is an extreme case of small-area motion. The video is determined to be a small-area motion video, and step ⑥ is no longer executed; the video contains frames with a frame difference greater than 0, and the video is determined to be a large-area motion video or a small-area motion video.

[0046] ⑥Finally determine the category to which the video belongs

[0047] According to formula (10), the ratio of the number of pixel blocks with a block difference of 0 in all frames with a frame difference greater than 0 to the total number of pixel blocks is determined.

[0048]

[0049] Where N γ,blo,0 Indicates the number of pixel blocks with a block difference of 0 in all frames with a frame difference greater than 0, N γ,blo,0The value is a finite non-negative integer, N γ,blo Indicates the total number of pixel blocks of all frames whose frame difference is greater than 0, N γ,blo The value is a finite positive integer. Determine that the video is a large-area motion video. Determine that the video is a small-area motion video.

[0050] (2) Determine the speed of video motion

[0051] The existing method of taking the maximum value of the standard deviation of the pixel value differences between adjacent frames is used to determine the speed T of the video motion.

[0052] (3) Determine the video texture complexity

[0053] The existing minimum transform domain absolute error and averaging method are used to determine the texture complexity S of the video.

[0054] (4) Determine the initial frame texture complexity

[0055] The texture complexity S of the initial frame of the video is determined by using the existing minimum transform domain absolute error and method. ini .

[0056] (5) Determine video bits per pixel

[0057] Determine the video bit per pixel (bpp) according to formula (11):

[0058] bpp=R tar / (ζ×h×w) (11)

[0059] In the formula, R tar represents the target bit rate of the video, ζ represents the frame rate of the video, and ζ is a finite positive integer.

[0060] (6) Determine the average bit per frame of the video

[0061] According to formula (12), the average number of bits R allocated to each frame of the video is determined fra :

[0062] R fra =R tar / N fra (12)

[0063] (7) Normalized data

[0064] The existing minimum-maximum normalization method is used to normalize the video motion speed T, the video texture complexity S, and the texture complexity S of the initial frame of the video. ini , the bits per pixel (bpp) of the video, the average number of bits allocated to each frame of the video R fraPerform normalization to obtain the normalized video motion speed Texture complexity of the video Texture complexity of the initial frame of the video Video bits per pixel The average number of bits allocated to each frame of the video

[0065] (8) Determine the minimum Euclidean distance

[0066] Determine according to formula (13) and The Euclidean distance L between K,P :

[0067]

[0068] In the formula, represents the motion speed obtained by normalizing the motion speed of the Pth video in the Kth category of videos using the existing minimum-maximum normalization method, The texture complexity of the Pth video in the Kth category is normalized. When K is 0, it means that the video belongs to the large area motion video category. When K is 1, it means that the video belongs to the small area motion video category. When K is 2, it means that the video belongs to the scene switching video category. P∈{1,…,Z}, Z represents the number of videos in the category to which the video belongs. Z is a finite positive integer. The minimum Euclidean distance is L opt .

[0069] (9) Determine the minimum absolute value difference

[0070] Determine according to formula (14) and The absolute value difference A y :

[0071]

[0072] In the formula, Indicates that L opt The corresponding video obtains the normalized bits per pixel at the yth target bit rate, y∈{0,1,…,U}, where U represents the number of quantization parameter values ​​allowed by the actual video coding standard, and U is a finite positive integer. The minimum absolute value difference is A opt .

[0073] (10) Determine weight factors and parameters

[0074] According to formula (15), determine the difference A with the minimum absolute value opt The corresponding intra-reference dependency weight factor a opt :

[0075]

[0076] In the formula, l 0 , o 0 、p 0 represents the parameters corresponding to the large area motion video category, l 0 The value is a finite positive real number, o 0 The value is a finite positive real number, p 0 The value is a finite non-negative real number, l 1 , o 1 、p 1 represents the parameters corresponding to the small area motion video category, l 1 The value is a finite positive real number, o 1 The value is a finite positive real number, p 1 The value is a finite non-negative real number, f 2 , l 2 , o 2 、p 2 Indicates the parameters corresponding to the scene switching video category, f 2 The value is a finite positive real number, l 2 The value is a finite negative real number, o 2 The value is a finite positive real number, p 2 The value is a finite real number.

[0077] The existing minimum-maximum normalization method is used to normalize a opt Perform normalization to obtain the normalized difference A from the minimum absolute value opt The corresponding intra-reference dependency weight factor

[0078] Determine the difference A with the minimum absolute value according to formula (16) opt The corresponding intra-frame reference dependency weight factor b opt :

[0079]

[0080] Determine the difference A with the minimum absolute value according to formula (17) opt The corresponding parameter c opt :

[0081]

[0082] In the formula, e 0 、s 0 ,t 0 represents the parameters corresponding to the large area motion video category, e 0 The value is a finite non-negative real number, s 0 The value is a finite positive real number, t0 The value is a finite positive real number, ε 1 、e 1 、s 1 ,t 1 represents the parameter corresponding to the small area motion video category, ε 1 The value is a finite positive real number, e 1 The value is a finite negative real number, s 1 The value is a finite positive real number, t 1 The value is a finite negative real number, e 2 、s 2 ,t 2 Indicates the parameters corresponding to the scene switching video category, e 2 The value is a finite positive real number, s 2 The value is a finite negative real number, t 2 The value is a finite positive real number.

[0083] According to formula (18), determine the difference A with the minimum absolute value opt The corresponding parameter k opt :

[0084]

[0085] In the formula, ι 0 , η 0 ,Ω 0 Indicates the parameters corresponding to the large area motion video category, The value is a finite positive real number, ι 0 The value is a finite negative real number, η 0 The value is a finite positive real number, Ω 0 The value is a finite negative real number, ι 1 , η 1 ,Ω 1 represents the parameters corresponding to the small area motion video category, ι 1 The value is a finite real number, η 1 The value is a finite real number, Ω 1 The value is a finite real number, ι 2 , η 2 ,Ω 2 Indicates the parameters corresponding to the scene switching video category. The value is a finite negative real number, ι 2 The value is a finite positive real number, η 2 The value is a finite negative real number, Ω 2 The value is a finite positive real number.

[0086] According to formula (19), determine the difference A with the minimum absolute value opt The corresponding parameter aopt :

[0087]

[0088] In the formula, C 0 、E 0 、F 0 , G 0 represents the parameters corresponding to the large area motion video category, C 0 The value is a finite negative real number, E 0 The value is a finite positive real number, F 0 The value is a finite negative real number, G 0 The value is a finite positive real number, C 1 、E 1 、F 1 , G 1 represents the parameters corresponding to the small area motion video category, C 1 The value is a finite positive real number, E 1 The value is a finite negative real number, F 1 The value is a finite positive real number, G 1 The value is a finite negative real number, C 2 、E 2 、F 2 , G 2 Indicates the parameters corresponding to the scene switching video category, C 2 The value is a finite real number, E 2 The value is a finite positive real number, F 2 The value is a finite negative real number, G 2 The value is a finite positive real number.

[0089] (11) Determine the initial frame normalized bit

[0090] According to formula (20), the normalized bits allocated to the initial frame are determined

[0091]

[0092] Where g, I, V, χ, i, j represent parameters, and their values ​​are finite positive real numbers.

[0093] (12) Determine the initial frame bit

[0094] According to the existing minimum-maximum normalization method, Perform denormalization to obtain the initial frame bit R ini .

[0095] (13) Determine the Lagrangian factor

[0096] Determine the Lagrangian factor λ of the initial frame according to formula (21):ini :

[0097] λ ini =(α / κ)×(spp ini ξ / bpp ini ) β (twenty one)

[0098] spp ini =S ini / (hw)

[0099] bpp ini =R ini / (hw)

[0100] Where α, κ, ξ, and β are model parameters, which are finite positive real numbers. ini Indicates the initial frame texture complexity per pixel, bpp ini Indicates the original frame bits per pixel.

[0101] (14) Limiting the Lagrangian factor

[0102] According to formula (22), the Lagrangian factor λ ini Make a restriction and obtain the Lagrangian factor after restriction

[0103] In the formula, λ max ∈[10,μ], μ is an intermediate variable, λ min ∈[0.01,0.2].

[0104] (15) Determine the initial frame quantization parameter

[0105] Determine the quantization parameter Q of the initial frame according to formula (23): ini :

[0106]

[0107] In the formula, and ψ as parameters, which are finite positive real numbers, d lum Indicates the bit depth scaling factor of the luma component, d lum is an intermediate variable.

[0108] (16) Limiting the initial frame quantization parameters

[0109] According to formula (24), Q ini Limit to obtain the quantitative parameters after limitation

[0110]

[0111] (17) Encoding

[0112] Use the quantization parameters after limiting The initial video frame is encoded according to the encoding software recommended by the video encoding standard.

[0113] In the formula (1) for determining the video category in step (1) of the present invention, m and n represent the position coordinates of the pixels in the pixel block, m∈{1,2,…,τ}, τ∈{1,2,…,h}, and h is 768-2160. The value of w is 1024 to 3840.

[0114] In the formula (14) for determining the minimum absolute value difference in step (9) of the present invention, the y∈{0,1,…,U}, U represents the number of quantization parameter values ​​allowed by the video coding standard actually used, and U is 51-63.

[0115] In the formula (15) for determining the weight factor and parameter in step (10) of the present invention, the l 0 , o 0 、p 0 represents the parameters corresponding to the large area motion video category, l 0 ∈[1,126.58],o 0 ∈[1,18.86], p 0 ∈[0,0.05292],l 1 , o 1 、p 1 represents the parameters corresponding to the small area motion video category, l 1 ∈[0.0048,0.0378],o 1 ∈[1,702.2], p 1 ∈[0,3.179], f 2 , l 2 , o 2 、p 2 Indicates the parameters corresponding to the scene switching video category, f 2 ∈[3.875e+07,4.112e+07],l 2 ∈[-6.589e+05,-6.315e+05],o 2 ∈[3000,3832], p 2 ∈[-100,100].

[0116] In the formula (17) for determining the weight factor and parameter in step (10), the e 0 、s 0 ,t 0 represents the parameters corresponding to the large area motion video category, e 0 ∈[0,0.001034], s0 ∈[1,11.08], t 0 ∈[1,18.82],ε 1 、e 1 、s 1 ,t 1 represents the parameter corresponding to the small area motion video category, ε 1 ∈[1.1456e+10,1.1684e+10],e 1 ∈[-1.633e+08,-1.433e+08], s 1 ∈[5.645e+05,5.696e+05], t 1 ∈[-592,-500], e 2 、s 2 ,t 2 Indicates the parameters corresponding to the scene switching video category, e 2 ∈[2.699e+05,2.839e+05],s 2 ∈[-4598,-2188], t 2 ∈[6.797,13.3].

[0117] In the formula (18) for determining the weight factor and parameter in step (10) of the present invention, the ι 0 , η 0 ,Ω 0 Indicates the parameters corresponding to the large area motion video category, ι 0 ∈[-1.323e+05,-0.723e+05], η 0 ∈[2320,3000],Ω 0 ∈[-20,-17.78],ι 1 , η 1 ,Ω 1 represents the parameters corresponding to the small area motion video category, l 1 ∈[-5.568e+06,2.863e+06], η 1 ∈[-3.03e+04,5.081e+04],Ω 1 ∈[-80.77,82.62], ι 2 , η 2 ,Ω 2 Indicates the parameters corresponding to the scene switching video category. ι 2 ∈[4.915e+07,4.937e+07],η 2∈[-2.618e+05,-2.600e+05],Ω 2 ∈[440.2,450.2].

[0118] In the formula (19) for determining the weight factor and parameter in step (10), the C 0 、E 0 、F 0 , G 0 represents the parameters corresponding to the large area motion video category, C 0 ∈[-1.569e+06,-1.419e+06], E 0 ∈[1.117e+05,1.359e+05],F 0 ∈[-3059,-2997],G 0 ∈[24,28.6], C 1 、E 1 、F 1 , G 1 represents the parameters corresponding to the small area motion video category, C 1 ∈[4.975e+08,4.981e+08],E 1 ∈[-6.213e+06,-6.173e+06], F 1 ∈[2.121e+04,2.151e+04],G 1 ∈[-24,-20.68],C 2 、E 2 、F 2 , G 2 Indicates the parameters corresponding to the scene switching video category, C 2 ∈[-2.325e+09,2.005e+09],E 2 ∈[3.265e+07,3.843e+07],F 2 ∈[-2.112e+05,-1.598e+05],G 2 ∈[301,317.2].

[0119] In the formula (20) for determining the normalized bits of the initial frame in step (11) of the present invention, the g, V, χ, i, and j represent parameters, g∈[1,10], V∈[0.1,1], χ∈[1,100], i∈[0.1,1], and j∈[0.1,1].

[0120] In the formula (21) for determining the Lagrangian factor in step (13) of the present invention, the α, κ, ξ, and β are parameters, α∈[1,13.5084], κ∈[1,512], ξ∈[1,5], and β∈[1,3.5720].

[0121] In the formula (23) for determining the initial frame quantization parameter in step (15) of the present invention, the following is said: With ψ as parameter, ψ∈[1,20].

[0122] Since the present invention adopts the steps of determining the video category, determining the weight factor and parameters, and determining the initial frame bit, the technical problem that the existing method only considers the spatial content characteristics of the screen content video and ignores the influence of the temporal content characteristics on the selection of the initial quantization parameters, resulting in low bit rate control performance, is solved. The present invention has the advantages of good bit rate control performance, high bit control accuracy, good rate distortion performance, etc., and can be used in the field of screen content video bit rate control technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0123] Figure 1 This is a flow chart of Example 1 of the present invention.

[0124] Figure 2 1 is a rate-distortion curve of Example 1 of the present invention and a comparative experimental method. DETAILED DESCRIPTION

[0125] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments, but the present invention is not limited to the following embodiments.

[0126] Example 1

[0127] exist Figure 1 In the embodiment, the method for selecting the initial quantization parameter for the screen content video bit rate control comprises the following steps:

[0128] (1) Determine the video category

[0129] 1) Determine block difference and frame difference

[0130] Divide each frame in the video into pieces of size Non-overlapping pixel blocks, τ∈{1,2,…,h}, h represents the height of the video, w represents the width of the video, and the value of h is 768-2160. In this embodiment, the value of h is 1080, and the value of w is 1024-3840. In this embodiment, the value of w is 1920.

[0131] According to formula (1), the average value of the sum of pixel differences of the δth block in the xth frame is determined, that is, the block difference σ of the δth block in the xth frame x,δ :

[0132]

[0133] Where x∈{1,2,…,N fra},N fra Indicates the total number of video frames, Nfra The value is a finite positive integer, δ∈{1,2,…,N blo},N blo Indicates the total number of pixel blocks per frame of the video, N blo The value is a finite positive integer, m and n represent the position coordinates of the pixel in the pixel block, u x,δ,m,n represents the pixel value of the mth row and nth column in the δth block of the xth frame of the video, m∈{1,2,…,τ}, τ∈{1,2,…,h}, u x,δ,m,n ∈{0,1,…,Γ},u x-1,δ,m,n represents the pixel value of the mth row and nth column in the δth block of the x-1th frame of the video, u x-1,δ,m,n ∈{0,1,…,Γ}, Γ is a finite positive integer.

[0134] According to formula (2), the average value of all block differences in the xth frame is determined, that is, the frame difference σ x :

[0135]

[0136] 2) Determine the video scene switching frame

[0137] ① Determine the difference between adjacent frames

[0138] Determine the frame difference σ of the x-1th frame according to formula (3): γ :

[0139]

[0140] Where, γ∈{1,2,…,N fra -1},σ γ,δ Represents the block difference of the δth block in the γth frame.

[0141] Determine the frame difference σ of the x+1th frame according to formula (4): Ψ :

[0142]

[0143] Where Ψ∈{2,…,N fra},σ Ψ,δ It represents the block difference of the δth block in the Ψth frame.

[0144] ②Determine the scene switching frame

[0145] Arrange all frame differences in ascending order to obtain the set H, and determine the average value σ of the first φ elements of the set H according to formula (5): ele :

[0146]

[0147] Where φ∈{1,2,…,Nfra -1},σ ρ represents the frame difference corresponding to the ρth element of the set H, σ x -σ γ >σ ele And σ x -σ Ψ >σ ele , determine the xth frame as the scene switching frame.

[0148] 3) Determine the category to which the video belongs

[0149] If the video contains scene switching frames, execute steps ①-⑥ to determine the category to which the video belongs. If the video does not contain scene switching frames, directly execute steps ⑤-⑥ to determine the category to which the video belongs.

[0150] ①Determine the average value of the remaining frame difference

[0151] Delete all scene switching frames and determine the average value σ of the remaining frame differences according to formula (6): rem :

[0152]

[0153] Where N fra,rem Indicates the number of frames remaining after deleting all scene switching frames from the video, N fra,rem ∈{1,2,…,N fra -1},r∈{1,2,…,N fra,rem},σ r represents the frame difference of the rth frame, σ r,δ represents the block difference of the δth block in the rth frame, u r,δ,m,n represents the pixel value of the mth row and nth column of the δth block of the rth frame, u r,δ,m,n ∈{0,1,…,Γ},u r-1,δ,m,n represents the pixel value of the mth row and nth column of the δth block in the r-1th frame, u r-1,δ,m,n ∈{0,1,…,Γ}.

[0154] ②Determine the average value of all frame differences

[0155] Determine the average value σ of all frame differences according to formula (7): vid :

[0156]

[0157] ③Determine the ratio of frame difference to block difference

[0158] According to formula (8), the frame difference of all scene switching frames in the video is determined as σ vid The ratio θ:

[0159] θ=|σ vid -σ rem | / σvid ×100% (8)

[0160] According to formula (9), the ratio ω of the number of pixel blocks with a block difference of 0 in all scene switching frames of the video to the total number of pixel blocks is determined.

[0161] ω=N SC,blo,0 / N SC,blo ×100% (9)Where N SC,blo,0 Indicates the number of pixel blocks with a block difference of 0 in all scene switching frames, N SC,blo,0 The value is a finite non-negative integer, N SC,blo Represents the total number of pixel blocks in all scene switching frames, N SC,blo The value is a finite positive integer.

[0162] ④ Initially determine the category of the video

[0163] If θ>Θ and ω<Θ, the video is determined to be a scene switching video, and steps ⑤ and ⑥ are no longer executed; if θ>Θ and ω<Θ are not satisfied, the video is determined to be a large-area motion video or a small-area motion video, Θ is the video category classification threshold, Θ∈(0,1).

[0164] ⑤ Determine the category of the video again

[0165] The video does not contain frames with a frame difference greater than 0, the video is in a static state, which is an extreme case of small-area motion. The video is determined to be a small-area motion video, and step ⑥ is no longer executed; the video contains frames with a frame difference greater than 0, and the video is determined to be a large-area motion video or a small-area motion video.

[0166] ⑥Finally determine the category to which the video belongs

[0167] According to formula (10), the ratio of the number of pixel blocks with a block difference of 0 in all frames with a frame difference greater than 0 to the total number of pixel blocks is determined.

[0168]

[0169] Where N γ,blo,0 Indicates the number of pixel blocks with a block difference of 0 in all frames with a frame difference greater than 0, N γ,blo,0 The value is a finite non-negative integer, N γ,blo Indicates the total number of pixel blocks of all frames whose frame difference is greater than 0, N γ,blo The value is a finite positive integer. Determine that the video is a large-area motion video. Determine that the video is a small-area motion video.

[0170] (2) Determine the speed of video motion

[0171] The existing method of taking the maximum value of the standard deviation of the pixel value differences between adjacent frames is used to determine the speed T of the video motion.

[0172] (3) Determine the video texture complexity

[0173] The existing minimum transform domain absolute error and averaging method are used to determine the texture complexity S of the video.

[0174] (4) Determine the initial frame texture complexity

[0175] The texture complexity S of the initial frame of the video is determined by using the existing minimum transform domain absolute error and method. ini .

[0176] (5) Determine video bits per pixel

[0177] Determine the video bit per pixel (bpp) according to formula (11):

[0178] bpp=R tar / (ζ×h×w) (11) In the formula, R tar represents the target bit rate of the video, ζ represents the frame rate of the video, and ζ is a finite positive integer.

[0179] (6) Determine the average bit per frame of the video

[0180] According to formula (12), the average number of bits R allocated to each frame of the video is determined fra :

[0181] R fra =R tar / N fra (12)

[0182] (7) Normalized data

[0183] The existing minimum-maximum normalization method is used to normalize the video motion speed T, the video texture complexity S, and the texture complexity S of the initial frame of the video. ini , the bits per pixel (bpp) of the video, the average number of bits allocated to each frame of the video R fra Perform normalization to obtain the normalized video motion speed Texture complexity of the video Texture complexity of the initial frame of the video Video bits per pixel The average number of bits allocated to each frame of the video

[0184] (8) Determine the minimum Euclidean distance

[0185] Determine according to formula (13) and The Euclidean distance L between K,P :

[0186]

[0187] In the formula, represents the motion speed obtained by normalizing the motion speed of the Pth video in the Kth category of videos using the existing minimum-maximum normalization method, The texture complexity of the Pth video in the Kth video is normalized. When it is 0, it means that the video belongs to the large area motion video category. When K is 1, it means that the video belongs to the small area motion video category. When K is 2, it means that the video belongs to the scene switching video category. P∈

[0188] {1,…,Z}, Z represents the number of videos in the category to which the video belongs, and Z is a finite positive integer. The minimum Euclidean distance is L opt .

[0189] (9) Determine the minimum absolute value difference

[0190] Determine according to formula (14) and The absolute value difference A y :

[0191]

[0192] In the formula, Indicates that L opt The corresponding video obtains the normalized bits per pixel at the yth target bit rate, y∈{0,1,…,U}, U represents the number of quantization parameter values ​​allowed by the actual video coding standard, and U is 51-63. In this embodiment, U is 63, and the minimum absolute value difference is A opt .

[0193] (10) Determine weight factors and parameters

[0194] According to formula (15), determine the difference A with the minimum absolute value opt The corresponding intra-reference dependency weight factor a opt :

[0195]

[0196] In the formula, l 0 , o 0 、p 0 represents the parameters corresponding to the large area motion video category, l 0 ∈[1,126.58], in this embodiment, l 0 The value is 63.29, o 0∈[1,18.86], o in this embodiment 0 The value is 9.43, p 0 ∈[0,0.05292], p in this example 0 The value is 0.02646; l 1 , o 1 、p 1 represents the parameters corresponding to the small area motion video category, l 1 ∈[0.0048,0.0378], in this embodiment, 1 The value is 0.0213, o 1 ∈[1,702.2], o in this embodiment 1 The value is 351.1, p 1 ∈[0,3.179], p in this embodiment 1 The value is 0.1895; f 2 , l 2 , o 2 、p 2 Indicates the parameters corresponding to the scene switching video category, f 2 ∈[3.875e+07,4.112e+07], f in this embodiment 2 The value is 3.962e+07, l 2 ∈[-6.589e+05,-6.315e+05], in this embodiment, 2 The value is -6.514e+05, o 2 ∈[3000,3832], o in this embodiment 2 The value is 3416, p 2 ∈[-100,100], p in this example 2 The value is -5.683.

[0197] The existing minimum-maximum normalization method is used to normalize a opt Perform normalization to obtain the normalized difference A from the minimum absolute value opt The corresponding intra-reference dependency weight factor

[0198] Determine the difference A with the minimum absolute value according to formula (16) opt The corresponding intra-frame reference dependency weight factor b opt :

[0199]

[0200] Determine the difference A with the minimum absolute value according to formula (17) opt The corresponding parameter c opt :

[0201]

[0202] In the formula, e 0 、s 0 ,t 0 represents the parameters corresponding to the large area motion video category, e 0 ∈[0,0.001034], e in this embodiment 0 The value is 0.000517, s 0 ∈[1,11.08], s in this embodiment 0 The value is 5.54, t 0 ∈[1,18.82], t in this embodiment 0 The value is 9.41; 1 、e 1 、s 1 ,t 1 represents the parameter corresponding to the small area motion video category, ε 1 ∈[1.1456e+10,1.1684e+10], ε in this embodiment 1 The value is 1.157e+10, e 1 ∈[-1.633e+08,-1.433e+08], e in this embodiment 1 The value is -1.533e+08, s 1 ∈[5.645e+05,5.696e+05], s in this embodiment 1 The value is 5.684e+05, t 1 ∈[-592,-500], t in this embodiment 1 The value is -546; e 2 、s 2 ,t 2 Indicates the parameters corresponding to the scene switching video category, e 2 ∈[2.699e+05,2.839e+05], e in this embodiment 2 The value is 2.769e+05, s 2 ∈[-4598,-2188], s in this embodiment 2 The value is -3393, t 2 ∈[6.797,13.3], t in this embodiment 2 The value is 10.05.

[0203] According to formula (18), determine the difference A with the minimum absolute value opt The corresponding parameter k opt :

[0204]

[0205] In the formula, ι 0 , η 0 ,Ω 0 Indicates the parameters corresponding to the large area motion video category, This embodiment of The value is 1.146e+06, ι 0 ∈[-1.323e+05,-0.723e+05], ι in this embodiment 0 The value is -1.023e+05, η 0 ∈[2320,3000], η in this embodiment 0 The value is 2660,Ω 0 ∈[-20,-17.78], Ω in this embodiment 0 The value is -18.89; 1 , η 1 ,Ω 1 represents the parameters corresponding to the small area motion video category, ι 1 ∈[-5.568e+06,2.863e+06], ι in this embodiment 1 The value is -1.353e+06, η 1 ∈[-3.03e+04,5.081e+04], η in this embodiment 1 The value is 1.026e+04, Ω 1 ∈[-80.77,82.62], Ω in this embodiment 1 The value is 0.9264; ι 2 , η 2 ,Ω 2 Indicates the parameters corresponding to the scene switching video category. This embodiment of The value is -2.984e+09, ι 2 ∈[4.915e+07,4.937e+07], ι in this embodiment 2 The value is 4.926e+07, η 2 ∈[-2.618e+05,-2.600e+05], η in this embodiment 2 The value is -2.609e+05,Ω 2 ∈[440.2,450.2], Ω in this embodiment 2 The value is 445.2.

[0206] According to formula (19), determine the difference A with the minimum absolute value opt The corresponding parameter z opt :

[0207]

[0208] In the formula, C 0 、E 0 、F 0 , G 0 represents the parameters corresponding to the large area motion video category, C 0 ∈[-1.569e+06,-1.419e+06], C in this embodiment 0 The value is -1.494e+06, E 0 ∈[1.117e+05,1.359e+05], E in this embodiment 0 The value is 1.238e+05, F 0 ∈[-3059,-2997], F in this embodiment 0 The value is -3028, G 0 ∈[24,28.6], C in this embodiment 0 The value is 26.3; C 1 、E 1 、F 1 , G 1 represents the parameters corresponding to the small area motion video category, G 1 ∈[4.975e+08,4.981e+08], C in this embodiment 1 The value is 4.978e+08, E 1 ∈[-6.213e+06,-6.173e+06], E in this embodiment 1 The value is -6.193e+06, F 1 ∈[2.121e+04,2.151e+04], F in this embodiment 1 The value is 2.136e+04, G 1 ∈[-24,-20.68], G in this embodiment 1 The value is -22.34; C 2 、E 2 、F 2 , G 2 Indicates the parameters corresponding to the scene switching video category, C 2 ∈[-2.325e+09,2.005e+09], C in this embodiment 2 The value is -2.165e+09, E 2 ∈[3.265e+07,3.843e+07], E in this embodiment 2 The value is 3.554e+07, F 2 ∈[-2.112e+05,-1.598e+05], F in this embodiment 2 The value is -1.855e+05, G2 ∈[301,317.2], G in this embodiment 2 The value is 309.1.

[0209] (11) Determine the initial frame normalized bit

[0210] According to formula (20), the normalized bits allocated to the initial frame are determined

[0211]

[0212] In the formula, g, I, V, χ, i, j represent parameters, g∈[1,10], the g value of this embodiment is 4, V∈[0.1,1], the V value of this embodiment is 0.5582, χ∈[1,100], the χ value of this embodiment is 40, i∈[0.1,1], the i value of this embodiment is 0.3, j∈[0.1,1], the j value of this embodiment is 0.25.

[0213] (12) Determine the initial frame bit

[0214] According to the existing minimum-maximum normalization method, Perform denormalization to obtain the initial frame bit R ini .

[0215] (13) Determine the Lagrangian factor

[0216] Determine the Lagrangian factor λ of the initial frame according to formula (21): ini :

[0217] λ ini =(α / κ)×(spp ini ξ / bpp ini ) β (twenty one)

[0218] spp ini =S ini / (hw)

[0219] bpp ini =R ini / (hw)

[0220] In the formula, α, κ, ξ, and β are model parameters, α∈[1,13.5084], the value of α in this embodiment is 6.7542, κ∈[1,512], the value of κ in this embodiment is 256, ξ∈[1,5], the value of ξ in this embodiment is 1.2517, β∈[1,3.5720], the value of β in this embodiment is 1.7860. spp ini Indicates the initial frame texture complexity per pixel, bpp iniIndicates the original frame bits per pixel.

[0221] (14) Limiting the Lagrangian factor

[0222] According to formula (22), the Lagrangian factor λ ini Make a restriction and obtain the Lagrangian factor after restriction

[0223] In the formula, λ max ∈[10,μ], μ is an intermediate variable, λ min ∈[0.01,0.2].

[0224] (15) Determine the initial frame quantization parameter

[0225] Determine the quantization parameter Q of the initial frame according to formula (23): ini :

[0226]

[0227] In the formula, With ψ as parameter, This embodiment of The value is 4.2005, ψ∈[1,20]. In this embodiment, the value of ψ is 13.7122. lum Indicates the bit depth scaling factor of the luma component, d lum is an intermediate variable.

[0228] (16) Limiting the initial frame quantization parameters

[0229] According to formula (24), Q ini Limit to obtain the quantitative parameters after limitation

[0230]

[0231] (17) Encoding

[0232] Use the quantization parameters after limiting The initial video frame is encoded according to the encoding software recommended by the video encoding standard.

[0233] A method for selecting initial quantization parameters for screen content video bit rate control is completed.

[0234] Example 2

[0235] The method for selecting the initial quantization parameter for the screen content video rate control in this embodiment comprises the following steps:

[0236] (1) Determine the video category

[0237] 1) Determine block difference and frame difference

[0238] Divide each frame in the video into pieces of size Non-overlapping pixel blocks, τ∈{1,2,…,h}, h represents the height of the video, w represents the width of the video, and the value of h is 768-2160. In this embodiment, the value of h is 768, and the value of w is 1024-3840. In this embodiment, the value of w is 1024.

[0239] The expression of formula (1) is the same as that of Example 1.

[0240] In formula (1), the meanings and values ​​of the parameters are the same as those in Example 1.

[0241] The other steps of this step are the same as those in Example 1.

[0242] (2) Determine the speed of video motion

[0243] This step is the same as in Example 1.

[0244] (3) Determine the video texture complexity

[0245] This step is the same as in Example 1.

[0246] (4) Determine the initial frame texture complexity

[0247] This step is the same as in Example 1.

[0248] (5) Determine video bits per pixel

[0249] This step is the same as in Example 1.

[0250] (6) Determine the average bit per frame of the video

[0251] This step is the same as in Example 1.

[0252] (7) Normalized data

[0253] This step is the same as in Example 1.

[0254] (8) Determine the minimum Euclidean distance

[0255] This step is the same as in Example 1.

[0256] (9) Determine the minimum absolute value difference

[0257] Determine according to formula (14) and The absolute value difference A y :

[0258] The expression of formula (14) is the same as that of Example 1.

[0259] In formula (14), Indicates that L opt The corresponding video obtains the normalized bits per pixel at the yth target bit rate, y∈{0,1,…,U}, U represents the number of quantization parameter values ​​allowed by the actual video coding standard, and U is 51-63. In this embodiment, U is 63, and the minimum absolute value difference is A opt .

[0260] (10) Determine weight factors and parameters

[0261] According to formula (15), determine the difference A with the minimum absolute value opt The corresponding intra-reference dependency weight factor a opt :

[0262] The expression of formula (15) is the same as that of Example 1.

[0263] In formula (15), l 0 , o 0 、p 0 represents the parameters corresponding to the large area motion video category, l 0 ∈[1,126.58], in this embodiment, l 0 The value is 1, o 0 ∈[1,18.86], o in this embodiment 0 The value is 1, p 0 ∈[0,0.05292] p in this example 0 The value is 0; l 1 , o 1 、p 1 represents the parameters corresponding to the small area motion video category, l 1 ∈[0.0048,0.0378], in this embodiment, 1 The value is 0.0048, 1 ∈[1,702.2], o in this embodiment 1 The value is 1, p 1 ∈[0,3.179], p in this embodiment 1 The value is 0; f 2 , l 2 , o 2 、p 2 Indicates the parameters corresponding to the scene switching video category, f 2 ∈[3.875e+07,4.112e+07], f in this embodiment 2 The value is 3.875e+07, l 2 ∈[-6.589e+05,-6.315e+05], in this embodiment, 2The value is -6.589e+05, o 2 ∈[3000,3832], o in this embodiment 2 The value is 3000, p 2 ∈[-100,100], p in this example 2 The value is -100.

[0264] Determine the difference A with the minimum absolute value according to formula (17) opt The corresponding model parameter c opt :

[0265] The expression of formula (17) is the same as that of Example 1.

[0266] In formula (17), e 0 、s 0 ,t 0 represents the parameters corresponding to the large area motion video category, e 0 ∈[0,0.001034], e in this embodiment 0 The value is 0, s 0 ∈[1,11.08], s in this embodiment 0 The value is 1, t 0 ∈[1,18.82], t in this embodiment 0 The value is 1; 1 、e 1 、s 1 ,t 1 represents the parameter corresponding to the small area motion video category, ε 1 ∈[1.1456e+

[0267] 10,1.1684e+10], ε in this embodiment 1 The value is 1.1456e+10, e 1 ∈[-1.633e+08,-1.433e+08], e in this embodiment 1 The value is -1.633e+08, s 1 ∈[5.645e+05,5.696e+05], s in this embodiment 1 The value is 5.645e+05, t 1 ∈[-592,-500], t in this embodiment 1 The value is -592; e 2 、s 2 ,t 2 Indicates the parameters corresponding to the scene switching video category, e 2 ∈[2.699e+05,2.839e+05], e in this embodiment 2 The value is 2.699e+05, s 2∈[-4598,-2188], s in this embodiment 2 The value is -4598, t 2 ∈[6.797,13.3], t in this embodiment 2 The value is 6.797.

[0268] According to formula (18), determine the difference A with the minimum absolute value opt The corresponding parameter k opt :

[0269] The expression of formula (18) is the same as that of Example 1.

[0270] In formula (18), l 0 , η 0 ,Ω 0 Indicates the parameters corresponding to the large area motion video category, This embodiment of The value is 1.1359e+06, l 0 ∈[-1.323e+05,-0.723e+05], ι in this embodiment 0 The value is -1.323e+05, η 0 ∈[2320,3000], η in this embodiment 0 The value is 2320,Ω 0 ∈[-20,-17.78], Ω in this embodiment 0 The value is -20; l 1 , η 1 ,Ω 1 represents the parameters corresponding to the small area motion video category, l 1 ∈[-5.568e+06,2.863e+06], ι in this embodiment 1 The value is -5.568e+06, η 1 ∈[-3.03e+04,5.081e+04], η in this embodiment 1 The value is -3.03e+04,Ω 1 ∈[-80.77,82.62], Ω in this embodiment 1 The value is -80.77; ι 2 , η 2 ,Ω 2 Indicates the parameters corresponding to the scene switching video category. This embodiment of The value is -2.994e+09, ι 2 ∈[4.915e+07,4.937e+07], ι in this embodiment2 The value is 4.915e+07, η 2 ∈[-2.618e+05,-2.600e+05], η in this embodiment 2 The value is -2.618e+05,Ω 2 ∈[440.2,450.2], Ω in this embodiment 2 The value is 440.2.

[0271] According to formula (19), determine the difference A with the minimum absolute value opt The corresponding parameter z opt :

[0272] The expression of formula (19) is the same as that of Example 1.

[0273] In formula (19), C 0 、E 0 、F 0 , G 0 represents the parameters corresponding to the large area motion video category, C 0 ∈[-1.569e+06,-1.419e+06], C in this embodiment 0 The value is -1.569e+06, E 0 ∈[1.117e+05,1.359e+05], E in this embodiment 0 The value is 1.117e+05, F 0 ∈[-3059,-2997], F in this embodiment 0 The value is -3059, G 0 ∈[24,28.6], C in this embodiment 0 The value is 24; C 1 、E 1 、F 1 , G 1 represents the parameters corresponding to the small area motion video category, C 1 ∈[4.975e+08,4.981e+08], C in this embodiment 1 The value is 4.975e+08, E 1 ∈[-6.213e+06,-6.173e+06], E in this embodiment 1 The value is -6.213e+06, F 1 ∈[2.121e+04,2.151e+04], F in this embodiment 1 The value is 2.121e+04, G 1 ∈[-24,-20.68], G in this embodiment 1 The value is -24; C 2 、E 2、F 2 , G 2 Indicates the parameters corresponding to the scene switching video category, C 2 ∈[-2.325e+09,2.005e+09], c in this embodiment 2 The value is -2.325e+09, e 2 ∈[3.265E+07,3.843E+07], e in this embodiment 2 The value is 3.265e+07, F 2 ∈[-2.112E+05,-1.598E+05], F in this embodiment 2 The value is -2.112e+05, G 2 ∈[301,317.2], G in this embodiment 2 The value is 301.

[0274] (11) Determine the initial frame normalized bit

[0275] According to formula (20), the normalized bits allocated to the initial frame are determined

[0276] The expression of formula (20) is the same as that of Example 1.

[0277] In formula (20), g, i, v, χ, I, j represent parameters, g∈[1,10], the value of g in this embodiment is 1, V∈[0.1,1], the value of V in this embodiment is 0.1, χ∈[1,100], the value of χ in this embodiment is 1, i∈[0.1,1], the value of i in this embodiment is 0.1, j∈[0.1,1], the value of j in this embodiment is 0.1.

[0278] (12) Determine the initial frame bit

[0279] This step is the same as in Example 1.

[0280] (13) Determine the Lagrangian factor

[0281] Determine the Lagrangian factor λ of the initial frame according to formula (21): ini :

[0282] The expression of formula (21) is the same as that of Example 1.

[0283] In formula (21), α, κ, ξ, and β are model parameters, α∈[1,13.5084], the value of α in this embodiment is 1, κ∈[1,512], the value of κ in this embodiment is 1, ξ∈[1,5], the value of ξ in this embodiment is 1, and β∈[1,3.5720], the value of β in this embodiment is 1. spp ini Indicates the initial frame texture complexity per pixel, bppini Indicates the original frame bits per pixel.

[0284] (14) Limiting the Lagrangian factor

[0285] This step is the same as in Example 1.

[0286] (15) Determine the initial frame quantization parameter

[0287] Determine the quantization parameter Q of the initial frame according to formula (23): ini :

[0288] The expression of formula (23) is the same as that of Example 1.

[0289] In formula (23), With ψ as parameter, This embodiment of The value is 1, ψ∈[1,20]. In this embodiment, the value of ψ is 1, d lum Indicates the bit depth scaling factor of the luma component, d lum is an intermediate variable.

[0290] The other steps are the same as those in Example 1.

[0291] A method for selecting initial quantization parameters for screen content video bit rate control is completed.

[0292] Example 3

[0293] The method for selecting the initial quantization parameter for the screen content video rate control in this embodiment comprises the following steps:

[0294] (1) Determine the video category

[0295] 1) Determine block difference and frame difference

[0296] Divide each frame in the video into pieces of size Non-overlapping pixel blocks, τ∈{1,2,…,h}, h represents the height of the video, w represents the width of the video, and the value of h ranges from 768 to 2160. In this embodiment, the value of h ranges from 768 to 3840. In this embodiment, the value of w ranges from 1024 to 3840. In this embodiment, the value of w is 1024.

[0297] The expression of formula (1) is the same as that of Example 1.

[0298] In formula (1), the meanings and values ​​of the parameters are the same as those in Example 1.

[0299] The other steps of this step are the same as those in Example 1.

[0300] (2) Determine the speed of video motion

[0301] This step is the same as in Example 1.

[0302] (3) Determine the video texture complexity

[0303] This step is the same as in Example 1.

[0304] (4) Determine the initial frame texture complexity

[0305] This step is the same as in Example 1.

[0306] (5) Determine video bits per pixel

[0307] This step is the same as in Example 1.

[0308] (6) Determine the average bit per frame of the video

[0309] This step is the same as in Example 1.

[0310] (7) Normalized data

[0311] This step is the same as in Example 1.

[0312] (8) Determine the minimum Euclidean distance

[0313] This step is the same as in Example 1.

[0314] (9) Determine the minimum absolute value difference

[0315] Determine according to formula (14) and The absolute value difference A y :

[0316] The expression of formula (14) is the same as that of Example 1.

[0317] In formula (14), Indicates that L opt The corresponding video obtains the normalized bits per pixel at the yth target bit rate, y∈{0,1,…,U}, U represents the number of quantization parameter values ​​allowed by the actual video coding standard, and U is 51-63. In this embodiment, U is 63, and the minimum absolute value difference is A opt .

[0318] (10) Determine weight factors and parameters

[0319] According to formula (15), determine the difference A with the minimum absolute value opt The corresponding intra-reference dependency weight factor a opt :

[0320] The expression of formula (15) is the same as that of Example 1.

[0321] In formula (15), l0 , o 0 、p 0 represents the parameters corresponding to the large area motion video category, l 0 ∈[1,126.58], in this embodiment, l 0 The value is 126.58, 0 ∈[1,18.86], o in this embodiment 0 The value is 18.86, p 0 ∈[0,0.05292], p in this example 0 The value is 0.05292; l 1 , o 1 、p 1 represents the parameters corresponding to the small area motion video category, l 1 ∈[0.0048,0.0378], in this embodiment, 1 The value is 0.0378, 1 ∈[1,702.2], o in this embodiment 1 The value is 702.2, p 1 ∈[0,3.179], p in this embodiment 1 The value is 3.179; f 2 , l 2 , o 2 、p 2 Indicates the parameters corresponding to the scene switching video category, f 2 ∈[3.875e+07,4.112e+07], f in this embodiment 2 The value is 4.112e+07, l 2 ∈[-6.589e+05,-6.315e+05], in this embodiment, 2 The value is -6.315e+05, o 2 ∈[3000,3832], o in this embodiment 2 The value is 3832, p 2 ∈[-100,100], p in this example 2 The value is 100.

[0322] Determine the difference A with the minimum absolute value according to formula (17) opt The corresponding model parameter c opt :

[0323] The expression of formula (17) is the same as that of Example 1.

[0324] In formula (17), e 0 、s 0 ,t 0 represents the parameters corresponding to the large area motion video category, e0 ∈[0,0.001034], e in this embodiment 0 The value is 0.001034, s 0 ∈[1,11.08], s in this embodiment 0 The value is 11.08, t 0 ∈[1,18.82], t in this embodiment 0 The value is 18.82; 1 、e 1 、s 1 ,t 1 represents the parameter corresponding to the small area motion video category, ε 1 ∈[1.1456e+10,1.1684e+10], ε in this embodiment 1 The value is 1.1684e+10, e 1 ∈[-1.633e+08,-1.433e+08], e in this embodiment 1 The value is -1.433e+08, s 1 ∈[5.645e+05,5.696e+05], s in this embodiment 1 The value is 5.696e+05, t 1 ∈[-592,-500], t in this embodiment 1 The value is -500; 2 、s 2 ,t 2 Indicates the parameters corresponding to the scene switching video category, e 2 ∈[2.699e+05,2.839e+05], e in this embodiment 2 The value is 2.839e+05, s 2 ∈[-4598,-2188], s in this embodiment 2 The value is -2188, t 2 ∈[6.797,13.3], t in this embodiment 2 The value is 13.3.

[0325] According to formula (18), determine the difference A with the minimum absolute value opt The corresponding parameter k opt :

[0326] The expression of formula (18) is the same as that of Example 1.

[0327] In formula (18), ι 0 , η 0 ,Ω 0 Indicates the parameters corresponding to the large area motion video category, This embodiment of The value is 1.1561e+06, ι 0 ∈[-1.323e+05,-0.723e+05], ι in this embodiment 0 The value is -0.723e+05, η 0 ∈[2320,3000], η in this embodiment 0 The value is 3000, Ω 0 ∈[-20,-17.78], Ω in this embodiment 0 The value is -17.28; 1 , η 1 ,Ω 1 represents the parameters corresponding to the small area motion video category, ι 1 ∈[-5.568e+06,2.863e+06], in this embodiment, 1 The value is 2.863e+06, η 1 ∈[-3.03e+04,5.081e+04], η in this embodiment 1 The value is 5.081e+04, Ω 1 ∈[-80.77,82.62], Ω in this embodiment 1 The value is 82.62; l 2 , η 2 ,Ω 2 Indicates the parameters corresponding to the scene switching video category. This embodiment of The value is -2.974e+09, l 2 ∈[4.915e+07,4.937e+07], in this embodiment, 2 The value is 4.937e+07, η 2 ∈[-2.618e+05,-2.600e+05], η in this embodiment 2 The value is -2.600e+05,Ω 2 ∈[440.2,450.2], Ω in this embodiment 2 The value is 450.2.

[0328] According to formula (19), determine the difference A with the minimum absolute value opt The corresponding parameter z opt :

[0329] The expression of formula (19) is the same as that of Example 1.

[0330] In formula (19), C 0 、E 0 、F 0, G 0 represents the parameters corresponding to the large area motion video category, C 0 ∈[-1.569e+06,-1.419e+06], C in this embodiment 0 The value is -1.419e+06, E 0 ∈[1.117e+05,1.359e+05], E in this embodiment 0 The value is 1.359e+05, F 0 ∈[-3059,-2997], F in this embodiment 0 The value is -2997, G 0 ∈[24,28.6], C in this embodiment 0 The value is 28.6; C 1 、E 1 、F 1 , G 1 represents the parameters corresponding to the small area motion video category, C 1 ∈[4.975e+08,4.981e+08], C in this embodiment 1 The value is 4.981e+08, E 1 ∈[-6.213e+06,-6.173e+06], E in this embodiment 1 The value is -6.173e+06, F 1 ∈[2.121e+04,2.151e+04], F in this embodiment 1 The value is 2.151e+04, G 1 ∈[-24,-20.68], G in this embodiment 1 The value is -20.68; C 2 、E 2 、F 2 , G 2 Indicates the parameters corresponding to the scene switching video category, C 2 ∈[-2.325e+09,2.005e+09], C in this embodiment 2 The value is 2.005e+09, E 2 ∈[3.265e+07,3.843e+07], E in this embodiment 2 The value is 3.843e+07, F 2 ∈[-2.112e+05,-1.598e+05], F in this embodiment 2 The value is -1.598e+05, G 2 ∈[301,317.2], G in this embodiment 2 The value is 317.2.

[0331] (11) Determine the initial frame normalized bit

[0332] According to formula (20), the normalized bits allocated to the initial frame are determined

[0333] The expression of formula (20) is the same as that of Example 1.

[0334] In formula (20), g, I, V, χ, i, and j represent parameters, g∈[1,10], the value of g in this embodiment is 10, V∈[0.1,1], the value of V in this embodiment is 1, χ∈[1,100], the value of χ in this embodiment is 100, i∈[0.1,1], the value of i in this embodiment is 1, and j∈[0.1,1], the value of j in this embodiment is 1.

[0335] (12) Determine the initial frame bit

[0336] This step is the same as in Example 1.

[0337] (13) Determine the Lagrangian factor

[0338] Determine the Lagrangian factor λ of the initial frame according to formula (21): ini :

[0339] The expression of formula (21) is the same as that of Example 1.

[0340] In formula (21), α, κ, ξ, and β are model parameters, α∈[1,13.5084], the value of α in this embodiment is 13.5084, κ∈[1,512], the value of κ in this embodiment is 512, ξ∈[1,5], the value of ξ in this embodiment is 5, and β∈[1,3.5720], the value of β in this embodiment is 3.5720. ini Indicates the initial frame texture complexity per pixel, bpp ini Indicates the original frame bits per pixel.

[0341] (14) Limiting the Lagrangian factor

[0342] This step is the same as in Example 1.

[0343] (15) Determine the initial frame quantization parameter

[0344] Determine the quantization parameter Q of the initial frame according to formula (23): ini :

[0345] The expression of formula (23) is the same as that of Example 1.

[0346] In formula (23), With ψ as parameter, This embodiment of The value is 20, ψ∈[1,20]. In this embodiment, the value of ψ is 20, d lum Indicates the bit depth scaling factor of the luma component, d lum is an intermediate variable.

[0347] The other steps are the same as those in Example 1.

[0348] A method for selecting initial quantization parameters for screen content video bit rate control is completed.

[0349] In order to verify the beneficial effects of the present invention, the inventors conducted a simulation experiment on the test video using the method of Embodiment 1 of the present invention. The experiment is as follows:

[0350] Five videos in the screen content video data set recommended by the International Organization for Standardization are selected as test sequences. The specific video information is as follows: the resolution of the cartoon robot sequence is 1280×720, and the frame rate is 30 frames / second; the resolution of the cartoon Chinese speed sequence is 1024×768, and the frame rate is 30 frames / second; the resolution of the console sequence is 1920×1080, and the frame rate is 60 frames / second; the resolution of the web page switching sequence is 1280×720, and the frame rate is 30 frames / second; the resolution of the data analysis sequence is 1920×1080, and the frame rate is 30 frames / second. A comparative experiment was conducted using Example 1 of the present invention (hereinafter referred to as the method of the present invention) and the rate control initial quantization parameter selection method in the encoder VTM19.0 recommended by the general video coding standard formulation organization (hereinafter referred to as the comparison method). The bit error B err The bit control accuracy of the two methods is calculated as follows:

[0351] B err =|R act -R tar | / R tar ×100%

[0352] In the formula, R act Indicates the actual bit rate of the encoded video. err The smaller the bit control accuracy, the higher the bit control accuracy. The method of determining the difference between the method of the present invention and the comparative method Rate (i.e. B del ) measures the rate-distortion performance of the two methods. del The smaller the value, the greater the improvement in rate-distortion performance.

[0353] Using the test model VTM19.0 recommended by the International Standards Organization, given R tar Next, the code rate control is enabled according to the method of the present invention and the comparative method to encode each test sequence, and the encoding structure is obtained, as shown in Table 1.

[0354] Table 1 Coding results of the present invention and the comparative method

[0355]

[0356] As can be seen from Table 1, for all test sequences, the average B err The average B err The average B err The experimental results show that the present invention has better bit control accuracy. Based on the comparison method, the average B del The value is -3.81%, which means that the present invention can save 3.81% of the encoding bit rate under the premise that the present invention and the comparison method can obtain the same quality of reconstructed video by encoding. The present invention has better rate-distortion performance.

[0357] Taking the console sequence as an example, Figure 2 The performance comparison between the method of the present invention and the comparative method is intuitively demonstrated. Figure 2 As shown, the coding bit rate of the method of the present invention is closer to R tar , whose rate-distortion curve is significantly higher than that of the comparison method.

[0358] In summary, the method of the present invention has better bit rate control performance.

Claims

1. A method for selecting an initial quantization parameter for screen content video rate control, characterized in that It consists of the following steps: (1) Determine the video category 1) Determine block difference and frame difference Divide each frame in the video into non-overlapping pixel blocks of size τ×θ, τ∈{1,2,…,h}, θ∈{1,2,…,w}, h represents the height of the video, w represents the width of the video, and h and w are finite positive integers; According to formula (1), the average value of the sum of pixel differences of the δth block in the xth frame is determined, that is, the block difference σ of the δth block in the xth frame x,δ : Where x∈{1,2,…,N fra },N fra Indicates the total number of video frames, N fra The value is a finite positive integer, δ∈{1,2,…,N blo },N blo Indicates the total number of pixel blocks per frame of the video, N blo The value is a finite positive integer, m and n represent the position coordinates of the pixel in the pixel block, m∈{1,2,…,τ}, n∈{1,2,…,θ}, u x,δ,m,n represents the pixel value of the mth row and nth column in the δth block of the xth frame of the video, u x,δ,m,n ∈{0,1,…,Γ},u x-1,δ,m,n represents the pixel value of the mth row and nth column in the δth block of the x-1th frame of the video, u x-1,δ,m,n ∈{0,1,…,Γ}, Γ is a finite positive integer; According to formula (2), the average value of all block differences in the xth frame is determined, that is, the frame difference σ x : 2) Determine the video scene switching frame ① Determine the difference between adjacent frames Determine the frame difference σ of the x-1th frame according to formula (3): Υ : Where Υ∈{1,2,…,N fra -1},σ Υ,δ represents the block difference of the δth block in the γth frame; Determine the frame difference σ of the x+1th frame according to formula (4): Ψ : Where Ψ∈{2,…,N fra },σ Ψ,δ represents the block difference of the δth block in the Ψth frame; ②Determine the scene switching frame Arrange all frame differences in ascending order to obtain the set H, and determine the average value σ of the first φ elements of the set H according to formula (5): ele : Where φ∈{1,2,...,N fra -1},σ ρ represents the frame difference corresponding to the ρth element of the set H, σ x -σ Υ >σ ele And σ x -σ Ψ >σ ele , determine the xth frame as the scene switching frame; 3) Determine the category to which the video belongs If the video contains scene switching frames, execute steps ①-⑥ to determine the category to which the video belongs. If the video does not contain scene switching frames, directly execute steps ⑤-⑥ to determine the category to which the video belongs. ①Determine the average value of the remaining frame difference Delete all scene switching frames and determine the average value σ of the remaining frame differences according to formula (6): rem : Where N fra,rem Indicates the number of frames remaining after deleting all scene switching frames from the video, N fra,rem ∈{1,2,…,N fra -1},r∈{1,2,…,N fra,rem },σ r represents the frame difference of the rth frame, σ r,δ represents the block difference of the δth block in the rth frame, u r,δ,m,n represents the pixel value of the mth row and nth column of the δth block of the rth frame, u r,δ,m,n ∈{0,1,…,Γ},u r-1,δ,m,n represents the pixel value of the mth row and nth column of the δth block in the r-1th frame, u r-1,δ,m,n ∈{0,1,…,Γ}; ②Determine the average value of all frame differences Determine the average value σ of all frame differences according to formula (7): vid : ③Determine the ratio of frame difference to block difference According to formula (8), the frame difference of all scene switching frames in the video is determined as σ vid The ratio θ: θ=|σ vid -s rem | / s vid ×100% (8) According to formula (9), the ratio ω of the number of pixel blocks with a block difference of 0 in all scene switching frames of the video to the total number of pixel blocks is determined as follows: ω=N SC,blo,0 / N SC,blo ×100% (9)Where N SC,blo,0 Indicates the number of pixel blocks with a block difference of 0 in all scene switching frames, N SC,blo,0 The value is a finite non-negative integer, N SC,blo Represents the total number of pixel blocks in all scene switching frames, N SC,blo The value is a finite positive integer; ④ Initially determine the category of the video If θ>Θ and ω<Θ, the video is determined to be a scene switching video, and steps ⑤ and ⑥ are no longer executed; if θ>Θ and ω<Θ are not satisfied, the video is determined to be a large-area motion video or a small-area motion video, Θ is the video category classification threshold, Θ∈(0,1); ⑤ Determine the category of the video again The video does not contain frames with a frame difference greater than 0, the video is in a static state, which is an extreme case of small-area motion, and the video is determined to be a small-area motion video, and step ⑥ is no longer executed; the video contains frames with a frame difference greater than 0, and the video is determined to be a large-area motion video or a small-area motion video; ⑥Finally determine the category to which the video belongs According to formula (10), the ratio of the number of pixel blocks with a block difference of 0 in all frames with a frame difference greater than 0 to the total number of pixel blocks is determined. Where N γ,blo,0 Indicates the number of pixel blocks with a block difference of 0 in all frames with a frame difference greater than 0, N γ,blo,0 The value is a finite non-negative integer, N γ,blo Indicates the total number of pixel blocks of all frames whose frame difference is greater than 0, N γ,blo The value is a finite positive integer. Determine that the video is a large-area motion video. Determine that the video is a small-area motion video; (2) Determine the speed of video motion The existing method of taking the maximum value of the standard deviation of the pixel value difference between adjacent frames is used to determine the speed T of the video motion; (3) Determine the video texture complexity The texture complexity S of the video is determined by using the existing minimum absolute error in the transform domain and the mean value method; (4) Determine the initial frame texture complexity The texture complexity S of the initial frame of the video is determined by using the existing minimum transform domain absolute error and method. ini ; (5) Determine video bits per pixel Determine the video bit per pixel (bpp) according to formula (11): In the formula, R tar Indicates the target bitrate of the video. Indicates the frame rate of the video. The value is a finite positive integer; (6) Determine the average bit per frame of the video According to formula (12), the average number of bits R allocated to each frame of the video is determined fra : R fra =R tar / N fra (12) (7) Normalized data The existing minimum-maximum normalization method is used to normalize the video motion speed T, the video texture complexity S, and the texture complexity S of the initial frame of the video. ini , the bits per pixel (bpp) of the video, the average number of bits allocated to each frame of the video R fra Perform normalization to obtain the normalized video motion speed Texture complexity of the video Texture complexity of the initial frame of the video Video bits per pixel The average number of bits allocated to each frame of the video (8) Determine the minimum Euclidean distance Determine according to formula (13) and The Euclidean distance L between K,P : In the formula, represents the motion speed obtained by normalizing the motion speed of the Pth video in the Kth category of videos using the existing minimum-maximum normalization method, The texture complexity of the Pth video in the Kth category is normalized. When K is 0, it means that the video belongs to the large area motion video category. When K is 1, it means that the video belongs to the small area motion video category. When K is 2, it means that the video belongs to the scene switching video category. P∈{1,…,Z}, Z represents the number of videos in the category to which the video belongs. Z is a finite positive integer. The minimum Euclidean distance is L opt ; (9) Determine the minimum absolute value difference Determine according to formula (14) and The absolute value difference A y : In the formula, Indicates that L opt The corresponding video obtains the normalized bits per pixel at the yth target bit rate, y∈{0,1,…,U}, where U represents the number of quantization parameter values ​​allowed by the actual video coding standard, and U is a finite positive integer. The minimum absolute value difference is A opt ; (10) Determine weight factors and parameters According to formula (15), determine the difference A with the minimum absolute value opt The corresponding intra-reference dependency weight factor a opt : Wherein, l0, o0, p0 represent parameters corresponding to the large-area motion video category, l0 takes a finite positive real number, o0 takes a finite positive real number, p0 takes a finite non-negative real number, l1, o1, p1 represent parameters corresponding to the small-area motion video category, l1 takes a finite positive real number, o1 takes a finite positive real number, p1 takes a finite non-negative real number, f2, l2, o2, p2 represent parameters corresponding to the scene switching video category, f2 takes a finite positive real number, l2 takes a finite negative real number, o2 takes a finite positive real number, o2 takes a finite real number; The existing minimum-maximum normalization method is used to normalize a opt Perform normalization to obtain the normalized difference A from the minimum absolute value opt The corresponding intra-reference dependency weight factor Determine the difference A with the minimum absolute value according to formula (16) opt The corresponding intra-frame reference dependency weight factor b opt : Determine the difference A with the minimum absolute value according to formula (17) opt The corresponding parameter c opt : Wherein, e0, s0, t0 represent parameters corresponding to the large-area motion video category, e0 takes a finite non-negative real number, s0 takes a finite positive real number, t0 takes a finite positive real number, ε1, e1, s1, t1 represent parameters corresponding to the small-area motion video category, ε1 takes a finite positive real number, e1 takes a finite negative real number, s1 takes a finite positive real number, t1 takes a finite negative real number, e2, s2, t2 represent parameters corresponding to the scene switching video category, e2 takes a finite positive real number, s2 takes a finite negative real number, t2 takes a finite positive real number; According to formula (18), determine the difference A with the minimum absolute value opt The corresponding parameter k opt : In the formula, ι0, η0, Ω0 represent parameters corresponding to the large-area motion video category, The value is a finite positive real number, ι0 is a finite negative real number, η0 is a finite positive real number, Ω0 is a finite negative real number, ι1, η1, Ω1 represent parameters corresponding to the small area motion video category, ι1 is a finite real number, η1 is a finite real number, Ω1 is a finite real number, ζ2, ι2, η2, Ω2 represent parameters corresponding to the scene switching video category, ζ2 is a finite negative real number, ι2 is a finite positive real number, η2 is a finite negative real number, Ω2 is a finite positive real number; According to formula (19), determine the difference A with the minimum absolute value opt The corresponding parameter z opt : Wherein, C0, E0, F0, G0 represent parameters corresponding to the large-area motion video category, C0 takes a finite negative real number, E0 takes a finite positive real number, F0 takes a finite negative real number, G0 takes a finite positive real number, C1, E1, F1, G1 represent parameters corresponding to the small-area motion video category, C1 takes a finite positive real number, E1 takes a finite negative real number, F1 takes a finite positive real number, G1 takes a finite negative real number, C2, E2, F2, G2 represent parameters corresponding to the scene switching video category, C2 takes a finite real number, E2 takes a finite positive real number, F2 takes a finite negative real number, G2 takes a finite positive real number; (11) Determine the initial frame normalized bit According to formula (20), the normalized bits allocated to the initial frame are determined In the formula, g, I, V, χ, i, j represent parameters, and their values ​​are finite positive real numbers; (12) Determine the initial frame bit According to the existing minimum-maximum normalization method, Perform denormalization to obtain the initial frame bit R ini ; (13) Determine the Lagrangian factor Determine the Lagrangian factor λ of the initial frame according to formula (21): ini : l ini =(a / k)×(spp ini ξ / bpp ini ) β (21) spp ini =S ini / (hw) bpp ini =R ini / (hw) Where α, κ, ξ, and β are model parameters, which are finite positive real numbers. ini Indicates the initial frame texture complexity per pixel, bpp ini Indicates the initial frame bits per pixel; (14) Limiting the Lagrangian factor According to formula (22), the Lagrangian factor λ ini Make a restriction and obtain the Lagrangian factor after restriction In the formula, λ max ∈[10,μ], μ is an intermediate variable, λ min ∈[0.01,0.2]; (15) Determine the initial frame quantization parameter Determine the quantization parameter Q of the initial frame according to formula (23): ini : In the formula, and ψ as parameters, which are finite positive real numbers, d lum Indicates the bit depth scaling factor of the luma component, d lum is an intermediate variable; (16) Limiting the initial frame quantization parameters According to formula (24), Q ini Limit to obtain the quantitative parameters after limitation (17) Encoding Use the quantization parameters after limiting The initial video frame is encoded according to the encoding software recommended by the video encoding standard.

2. The method for selecting initial quantization parameters for screen content video rate control according to claim 1, characterized in that: In the formula (1) for determining the video category in step (1), m and n represent the position coordinates of the pixel in the pixel block, m∈{1,2,…,τ}, τ∈{1,2,…,h}, h takes a value of 768 to 2160, n∈{1,2,…,θ}, θ∈{1,2,…,w}, and w takes a value of 1024 to 3840.

3. The method for selecting initial quantization parameters for screen content video rate control according to claim 1, characterized in that: In the formula (14) for determining the minimum absolute value difference in step (9), the y∈{0,1,…,U}, U represents the number of quantization parameter values ​​allowed by the video coding standard actually used, and U is 51-63.

4. The method for selecting initial quantization parameters for screen content video rate control according to claim 1, characterized in that: In the formula (15) for determining the weight factors and parameters in step (10), the l0, o0, and p0 represent parameters corresponding to the large-area motion video category, l0∈[1,126.58], o0∈[1,18.86], and p0∈[0,0.05292]; l1, o1, and p1 represent parameters corresponding to the small-area motion video category, l1∈[0.0048,0.0378], o1∈[1,702.2], and p1∈[0,3.179]; f2, l2, o2, and p2 represent parameters corresponding to the scene switching video category, f2∈[3.875e+07,4.112e+07], l2∈[-6.589e+05,-6.315e+05], o2∈[3000,3832], and p2∈[-100,100]; In the formula (17) for determining the weight factor and parameter in step (10), the e0, s0, t0 represent parameters corresponding to the large area motion video category, e0∈[0,0.001034], s0∈[1,11.08], t0∈[1,18.82], ε1, e1, s1, t1 represent parameters corresponding to the small area motion video category, ε1∈[1.1456e+10,1.1684e+10], e1∈[-1.633e+08,-1.433e+08], s1∈[5.645e+05,5.696e+05], t1∈[-592,-500], e2, s2, t2 represent the parameters corresponding to the scene switching video category, ee2∈[2.699e+05,2.839e+05], s2∈[-4598,-2188], t2∈[6.797,13.3]; In the formula (18) for determining the weight factors and parameters in step (10), the ζ0, ι0, η0, and Ω0 represent parameters corresponding to the large-area motion video category, ζ0∈[1.1359e+06,1.1561e+06], ι0∈[-1.323e+05,-0.723e+05], η0∈[2320,3000], Ω0∈[-20,-17.78], l1, η1, and Ω1 represent parameters corresponding to the small-area motion video category, ι1∈[-5.568e+06,2.863e+06], η1∈[-3.03e+04,5.081e+04], Ω1∈[-80.77,82.62], ι2, η2, Ω2 represent parameters corresponding to the scene switching video category, ι2∈[4.915e+07,4.937e+07], η2∈[-2.618e+05,-2.600e+05], Ω2∈[440.2,450.2]; In the formula (19) for determining the weight factors and parameters in step (10), C0, E0, F0, and G0 represent parameters corresponding to the large-area motion video category, C0∈[-1.569e+06,-1.419e+06], E0∈[1.117e+05,1.359e+05], F0∈[-3059,-2997], and G0∈[24,28.6], and C1, E1, F1, and G1 represent parameters corresponding to the small-area motion video category, and C1∈[4.975e+08,4.981e+08 ], E1∈[-6.213e+06,-6.173e+06], F1∈[2.121e+04,2.151e+04], G1∈[-24,-20.68], C2, E2, F2, G2 represent the parameters corresponding to the scene switching video category, C2∈[-2.325e+09,2.005e+09], E2∈[3.265e+07,3.843e+07], F2∈[-2.112e+05,-1.598e+05], G2∈[301,317.2].

5. The method for selecting initial quantization parameters for screen content video rate control according to claim 1, characterized in that: In the formula (20) for determining the normalized bits of the initial frame in step (11), the g, V, χ, i, and j represent parameters, g∈[1,10], V∈[0.1,1], χ∈[1,100], i∈[0.1,1], and j∈[0.1,1].

6. The method for selecting initial quantization parameters for screen content video rate control according to claim 1, characterized in that: In formula (21) for determining the Lagrangian factor in step (13), the α, κ, ξ, and β are parameters, α∈[1,13.5084], κ∈[1,512], ξ∈[1,5], and β∈[1,3.5720].

7. The method for selecting initial quantization parameters for screen content video rate control according to claim 1, characterized in that: In the formula (23) for determining the initial frame quantization parameter in step (15), With ψ as parameter, ψ∈[1,20].

Images