CN107147911A - Method and device for fast inter-frame coding mode selection based on local brightness compensation LIC - Google Patents

Abstract

The invention discloses a kind of quick interframe coding mode selection method and device that LIC is compensated based on local luminance, the brightness variation characteristic and temporal correlation that the program obtains video sequence by the current coded unit CU of the video histograms that CU is matched with reference frame shift to an earlier date anticipation and go out that LIC encoding block need not be carried out, skip unnecessary and time-consuming LIC selection courses, the computation complexity of coding is reduced, the scramble time is reduced；The present invention is simple and easy to apply, is conducive to the Industry Promotion of video encoding standard of new generation.

Description

Method and device for fast inter-frame coding mode selection based on local brightness compensation LIC

technical field

The invention belongs to the field of video coding, in particular to a fast inter-frame coding mode selection method and device based on local brightness compensation LIC.

Background technique

LIC (Local Illumination Compensation) is a linear model for image brightness changes, using scaling factor a and offset b to compensate the brightness of the original pixel, and adaptively enable or disable for each inter coding unit (CU) . Recently, the establishment of a new generation of video coding standards has introduced a large number of new coding tools, and LIC technology is one of them.

In recent years, with the application of high-definition and ultra-high-definition video (resolution up to 4K×2K, 8K×4K) gradually entering people’s field of vision, video compression technology has been greatly challenged, and the video compression coding standard system has also been developed rapidly. In addition, various video applications are also emerging with the development of network and storage technologies. Today, digital video broadcasting, mobile wireless video, remote detection, medical imaging and portable photography have all entered people's lives. The requirements for video quality are also getting higher and higher. Therefore, the diversification of video applications and the trend of high-definition have put forward stronger requirements for the next-generation video coding standard with higher coding efficiency than H.265/HEVC. It is against this background that the VCEG (Video Coding Experts Group) of ITU-T VCEG and the MPEG (Motion Picture Experts Group) of ISO/IEC established the JVET (Joint Video Exploration Team) in 2016. The purpose is to explore the development and formulation of a new generation of video coding standards.

The new generation of video coding standards still uses a hybrid coding framework, including modules such as transform, quantization, entropy coding, intra-frame prediction, inter-frame prediction, and loop filtering. However, in order to improve the video compression rate, the standard uses QTBT (Quadtreeplus binary tree ) partition structure, replacing the quadtree partition of HEVC. Under the QTBT structure, a variety of division types such as CU, PU and TU separation concepts are removed, and more flexible CU division types are supported to better match the local characteristics of video data. At the same time, a series of time-consuming procedures are introduced in each module. New coding tools, such as Affine-based merge technology, FRUC-based merge technology, local brightness compensation technology, referred to as LIC (Local Illumination Compensation), etc., these improvements to improve the compression rate have greatly increased the computational complexity of the encoder, which is not conducive to new The industrial promotion of a generation of video coding standards. Therefore, optimizing the encoder and reducing the encoding time while ensuring that the subjective quality of the video is negligible is one of the problems to be studied and solved in the field of video encoding and decoding.

The LIC technology introduced by the new generation of video coding standards mainly judges whether the current coding block performs layout brightness compensation through the image brightness change characteristics and the rate-distortion function. It is mainly divided into two steps. In the first step, if there is no obvious brightness change between the current frame and its reference frame, no LIC is performed on the entire frame. In order to identify this situation, the encoder calculates the histogram difference between the current frame and the reference frame before encoding the image. If the sum of the differences of all pixel values is less than a given threshold, there is no need to perform LIC on the current frame; otherwise , perform LIC on the current frame;

During the inter-frame prediction process, the encoder will perform two operations on the three modes of 2Nx2N, FRUC merge, and IMV 2Nx2N without LIC and with LIC, and select the LIC mode of the current encoding block by comparing the rate-distortion function RD Cost (yes or no ), if so, the parameters a and b are calculated by using the least squares method by using the adjacent samples of the current CU block and its reference block. More specifically, as shown in FIG. 1 , the reference block and the 1/2 downsampled points of the adjacent pixels of the current block are used for calculation, and the LIC parameters are obtained and all pixels of the current block are compensated for inter-frame prediction. Otherwise, inter prediction can be performed directly without compensation.

Through the test and analysis of the reference software JEM of the new generation of video coding standards, it is found that under the Lowdelay configuration, the coding time of LIC technology accounts for 30% to 35% of the total coding time. mode, so as to avoid unnecessary judgment and selection process will greatly improve the coding efficiency of the new generation of video coding standards.

Contents of the invention

The object of the present invention is to provide a fast inter-frame coding mode selection method and device based on local brightness compensation LIC, which can predict in advance the coding blocks that do not need to perform LIC through the brightness change characteristics of the video and the temporal-spatial correlation, and skip unnecessary coding blocks. And the time-consuming LIC selection process reduces the computational complexity of encoding and reduces encoding time.

A fast inter-frame coding mode selection method based on local brightness compensation LIC. The coding units in all video frames sequentially perform Affine Merge mode without LIC, normal Merge mode, FRUC Merge mode and 2N×2N mode to obtain The rate-distortion cost of the coding unit, select the mode corresponding to the minimum rate-distortion cost as the coding mode of the coding unit; if the LIC mode flag bit of the frame where the current coding unit is located is true, the current coding unit continues to perform FRUC Merge mode with LIC and 2N ×2N mode, to obtain the sum of the luminance value residual between the current coding unit CU and the reference frame matching CU and the difference between SumPixel _SAD and the luminance histogram SumHist _SAD , if the luminance value residual rate in the current coding unit CU Less than the set first threshold or the histogram difference rate in the current coding unit CU is less than the set second threshold, the current coding unit skips the LIC mode;

The reference frame matching coding unit CU is a coding unit obtained by performing motion compensation on the co-located block of the reference frame by using the motion vector MV of the current coding unit in the selected coding mode. A coding unit at the same position as the current coding unit in the current frame;

The luminance value residual and SumPixel _SAD between the current coding unit CU and the reference frame matching coding unit CU are obtained by calculating the difference between the pixel values of the corresponding pixel points between the current coding unit CU and the reference frame matching coding unit CU and get;

The sum of the difference of the brightness histogram between the current coding unit CU and the reference frame matching coding unit CU is obtained by matching the histogram value between the corresponding gray levels between the current coding unit CU and the reference frame matching coding unit CU The sum of the differences is obtained;

W _CU is the number of pixels in the current coding unit CU, the value of the first threshold is 1-10, and the value of the second threshold is 0-2.

Further, before performing the LIC mode, first calculate the total number of true LIC flags of the adjacent coding units of the current coding unit CU Then judge whether it is satisfied: If it is satisfied, the current coding unit CU skips the LIC mode, otherwise, the current coding unit continues to perform the FRUC Merge mode and 2N×2N mode with LIC;

Among them, CU _x represents the adjacent unit of the current coding unit, x∈{Left, AboveLeft, Above, Col}, CU _Left , CU _AboveLeft , CU _Above , and CU _Col respectively represent the left adjacent block, upper left adjacent block, Upper adjacent block and co-located block, the co-located block refers to the coding unit that is in the reference frame and has the same position as the current coding unit in the current frame; f(CU _x ) represents the LIC of the adjacent unit of the current coding unit The flag, if true, takes the value 1, otherwise, takes the value 0.

Furthermore, after the current coding unit performs the FRUC Merge mode with LIC and the 2N×2N mode, it then performs the IMV_2N×2N mode without LIC to obtain the corresponding rate-distortion costs in all modes, and use the mode corresponding to the minimum rate-distortion cost As the best mode; if the best mode is the mode processed by LIC, and the LIC flag bit of the adjacent coding unit is true, the total number If it is not 0, the current coding unit continues the IMV_2N×2N mode with LIC; otherwise, the current coding unit skips the IMV_2N×2N mode with LIC.

A fast inter-frame coding mode selection device based on local brightness compensation LIC, comprising:

The rate-distortion cost calculation unit is used to sequentially perform LIC-free AffineMerge mode, normal Merge mode, FRUC Merge mode and 2N×2N mode on the coding units in all video frames, and obtain the rate-distortion cost of the coding units in each mode;

Selecting the unit by selecting the mode corresponding to the minimum rate-distortion cost as the coding mode of the coding unit;

Skip the unit, if the LIC mode flag of the frame where the current coding unit is located is true, judge whether the current coding unit skips the LIC mode according to the following rules:

If the luminance value residual rate in the current coding unit CU Less than the set first threshold or the histogram difference rate in the current coding unit CU is less than the set second threshold, the current coding unit skips the LIC mode;

Among them, SumPixel _SAD and SumHist _SAD respectively represent the sum of the difference between the brightness value residual and the brightness histogram between the current coding unit CU and the reference frame matching coding unit CU;

The reference frame matching coding unit CU is a coding unit obtained by performing motion compensation on the co-located block of the reference frame by using the motion vector MV of the current coding unit in the selected coding mode. A coding unit at the same position as the current coding unit in the current frame;

The luminance value residual and SumPixel _SAD between the current coding unit CU and the reference frame matching coding unit CU are obtained by calculating the difference between the pixel values of the corresponding pixel points between the current coding unit CU and the reference frame matching coding unit CU and get;

The sum of the difference SumHist _SAD of the luminance histogram between the current coding unit CU and the reference frame matching coding unit CU is obtained by matching the histogram between the corresponding gray levels between the current coding unit CU and the reference frame matching coding unit CU The sum of the difference between the graph values is obtained;

W _CU is the number of pixels in the current coding unit CU, the value of the first threshold is 1-10, and the value of the second threshold is 0-2.

Further, before performing the LIC mode, by counting the total number of true LIC flags of the adjacent coding units of the current coding unit CU and judge Whether it is 0, if it is 0, the current coding unit CU skips the LIC mode, otherwise, the current coding unit continues to perform the FRUC Merge mode and 2N×2N mode with LIC.

Further, it also includes the best mode selection unit, if the selected best mode belongs to the mode processed by LIC, and the total number of adjacent coding unit LIC flags is true If it is not 0, the current coding unit continues to perform the IMV_2N×2N mode with LIC, otherwise, skip the IMV_2N×2N mode with LIC;

The optimal mode selection unit performs the FRUC Merge mode with LIC and the 2N×2N mode in the current coding unit, and then performs the IMV_2N×2N mode without LIC to obtain the corresponding rate-distortion costs in all modes, and use the minimum The mode corresponding to the rate-distortion cost is used to select the best mode.

Beneficial effect

The present invention provides a fast inter-frame coding mode selection method and device based on local brightness compensation LIC. The scheme predicts in advance the coding blocks that do not need to perform LIC through the brightness change characteristics of the video and the temporal-spatial correlation, and skips unnecessary and The time-consuming LIC selection process reduces the computational complexity of encoding and reduces encoding time; specifically reflected in the following points:

1. Use the histogram of the current coding unit CU and the matching block of the reference frame to obtain the brightness change characteristics of the video sequence, avoid the time-consuming LIC mode selection process in the coding process, and reduce the coding time;

2. Utilize the temporal and spatial correlation of video sequences to avoid the time-consuming LIC mode selection process in the encoding process, reduce the computational complexity of the new generation of video encoding, and greatly improve the encoding efficiency;

3. Use the best LIC mode information obtained in non-IMV mode to avoid unnecessary LIC mode selection process in IMV_2N×2N mode and improve coding efficiency;

4. The present invention is simple and easy to implement, and is beneficial to the industrial promotion of the new generation of video coding standards.

Description of drawings

Fig. 1 is the adjacent pixel schematic diagram of obtaining LIC parameter;

Fig. 2 is the overall flow chart of the present invention.

detailed description

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

In order to reduce encoding time and improve work efficiency, the technical solution specifically adopted in the present invention is: firstly judge whether the image frame where the current block is located needs to select the LIC mode; The SumHist _SAD of the difference and the SumPixel _SAD of the luminance value residual, if SumHist _SAD /W _CU <λ ₀ or SumPixel _SAD /W _CU <λ ₁ , it is considered that the brightness change of the current block and the reference frame matching block is not obvious, and the advance It is determined that brightness compensation is not required for the current block, and the LIC mode selection process is skipped. Second, if the LIC mode that is not the best mode in IMV is No, the LIC mode selection process of the IMV mode is skipped. In addition, according to the temporal-spatial correlation of the video sequence, the adjacent blocks and co-located blocks of the current block are obtained to predict the LIC mode of the current block. Also for No, directly skip the time-consuming LIC mode selection process.

A fast inter-frame coding mode selection device based on local brightness compensation LIC, comprising:

The rate-distortion cost calculation unit is used to sequentially perform LIC-free AffineMerge mode, normal Merge mode, FRUC Merge mode and 2N×2N mode on the coding units in all video frames, and obtain the rate-distortion cost of the coding units in each mode;

Selecting the unit by selecting the mode corresponding to the minimum rate-distortion cost as the coding mode of the coding unit;

Skip the unit, if the LIC mode flag of the frame where the current coding unit is located is true, judge whether the current coding unit skips the LIC mode according to the following rules:

If the luminance value residual rate in the current coding unit CU Less than the set first threshold or the histogram difference rate in the current coding unit CU is less than the set second threshold, the current coding unit skips the LIC mode;

Among them, SumPixel _SAD and SumHist _SAD respectively represent the sum of the difference between the brightness value residual and the brightness histogram between the current coding unit CU and the reference frame matching coding unit CU;

The reference frame matching coding unit CU is a coding unit obtained by performing motion compensation on the co-located block of the reference frame by using the motion vector MV of the current coding unit CU in the selected coding mode, the co-located block refers to being in the reference frame, and A coding unit whose position is the same as that of the current coding unit in the current frame;

The luminance value residual and SumPixel _SAD between the current coding unit CU and the reference frame matching coding unit CU are obtained by calculating the difference between the pixel values of the corresponding pixel points between the current coding unit CU and the reference frame matching coding unit CU and get;

The sum of the difference SumHist _SAD of the luminance histogram between the current coding unit CU and the reference frame matching coding unit CU is obtained by matching the histogram between the corresponding gray levels between the current coding unit CU and the reference frame matching coding unit CU The sum of the difference between the graph values is obtained;

W _CU is the number of pixels in the current coding unit CU, the value of the first threshold is 1-10, and the value of the second threshold is 0-2.

Before performing the LIC mode, by counting the total number of LIC flags of the adjacent coding units of the current coding unit CU that are true and judge Whether it is 0, if it is 0, the current coding unit CU skips the LIC mode, otherwise, the current coding unit continues to perform the FRUC Merge mode and 2N×2N mode with LIC.

Also includes the best mode selection unit, if the selected best mode belongs to the mode processed by LIC, and the total number of adjacent coding unit LIC flags is true If it is not 0, the current coding unit continues to perform the IMV_2N×2N mode with LIC, otherwise, skip the IMV_2N×2N mode with LIC;

The optimal mode selection unit performs the FRUC Merge mode with LIC and the 2N×2N mode in the current coding unit, and then performs the IMV_2N×2N mode without LIC to obtain the corresponding rate-distortion costs in all modes, and use the minimum The mode corresponding to the rate-distortion cost is used to select the best mode.

Embodiment one

A fast inter-frame coding mode selection method based on local brightness compensation LIC, comprising the following steps:

Step 1: In inter-frame prediction, the JEM encoder sequentially performs Affine Merge, normal Merge, FRUC Merge and 2N×2N modes without LIC (no LIC: no local brightness compensation), and then decides the best value according to the rate-distortion cost. best mode;

Step 2: First obtain the LIC mode flag bit of the image frame where the current coding unit CU is located, so as to initialize the LIC flag bit Cur_IC _Flag of the current block, if Cur_IC _Flag =true, then continue to step 3; otherwise, go to step 8;

Step 3: Use the best mode obtained in step 1 to obtain the motion vector MV, and then perform motion compensation on the co-located block of the reference frame, that is, translate the co-located block according to the MV to obtain the matching block RefCU of the current CU;

Step 4: Use the matching block RefCU obtained in Step 3 to obtain the brightness values of each pixel in the current block and the matching block. According to this, the brightness histograms CurrHist[N _Pel ] and RefHist[N _Pel ] are generated, where N _Pel is the maximum value of the brightness in the video sequence. Since the range of brightness in the test sequence is 0-255, the value in the algorithm is for 255. Additionally, by

Calculate the SumPixel _SAD of the luminance value residuals between the current block and the matching block, where CurrCU[i] and RefCU[i] are the pixel values of the current block and the matching block respectively, and W _CU is the CU size, that is, the number of pixels.

In addition, the histogram is generated as follows

Hist[i]=∑f(x,y,i)(x,y)∈CU,i∈[0,N _Pel ]

Wherein, Pixel_value(x, y) is the pixel value at the coordinate point (x, y) in the current coding unit CU.

Step 5: Using the histograms CurrHist[N _Pel ] and RefHist[N _Pel ] generated in step 4, pass

Compute the SumHist _SAD of the differences between the brightness histograms of the current block and the matching block.

Step 6: Use the SumPixel _SAD and SumHist _SAD calculated in Step 4 and Step 5 to make the following judgments. If any one of the two formulas is satisfied, it is considered that the brightness change between the current coding unit CU and the matching block is not obvious, and local brightness compensation is not required. Assign Cur_IC _Flag = false and go to step eight; otherwise, Cur_IC _Flag maintains the original value, which is true, and continues to the next step

SumHist _SAD /W _CU <λ ₀

SumPixel _SAD /W _CU <λ ₁

In the formula, λ ₀ and λ ₁ are the thresholds (through a large number of experimental statistics, when λ ₀ =0.5, λ ₁ =5, the encoding effect is the best), and W _CU is the current CU size, that is, the number of pixels.

Step 7: The encoder performs FRUC Merge mode and 2N×2N mode with LIC (with LIC: for local brightness compensation), and decides the best mode according to the rate-distortion cost.

Step 8: The encoder performs an integer-pixel IMV_2N×2N mode without LIC, still determines the best mode according to the rate-distortion cost, and obtains the LIC information IC_Flag_temp of the best mode. Then select the follow-up operation according to Cur_IC _Flag , if Cur_IC _Flag =true, proceed to the next step, otherwise, go to step eleven.

Step 9: According to the IC_Flag_temp obtained in step 8, if it is true, go to the next step directly, otherwise, go to step 11, consider that the current coding unit CU does not need to perform the LIC mode selection process, and assign Cur_IC _Flag = false,

Step 10: The encoder performs IMV_2N×2N mode with LIC.

Step 11: End the current algorithm and make subsequent judgments on other modes.

Embodiment two

A fast inter-frame coding mode selection method based on local brightness compensation LIC, comprising the following steps:

Step 1: In inter-frame prediction, the JEM encoder sequentially performs Affine Merge, normal Merge, FRUC Merge and 2N×2N modes without LIC (no LIC: no local brightness compensation), and then decides the best value according to the rate-distortion cost. best mode;

Step 2: first obtain the LIC mode flag bit of the current coding unit CU, so as to initialize the LIC flag bit Cur_IC _Flag of the current block, if Cur_IC _Flag =true, continue to step 3; otherwise, go to step 9;

Step 3: Use the best mode obtained in step 1 to obtain the motion vector MV, and then perform motion compensation on the co-located block of the reference frame, that is, translate the co-located block according to the MV to obtain the matching block RefCU of the current coding unit CU;

Step 4: Use the matching block RefCU obtained in Step 3 to obtain the brightness value of each pixel in the current coding unit and the matching block. According to this, the brightness histograms CurrHist[N _Pel ] and RefHist[N _Pel ] are generated, where N _Pel is the maximum value of the brightness in the video sequence. Since the range of brightness in the test sequence is 0-255, the value in the algorithm is for 255. Additionally, by

Calculate the SumPixel _SAD of the residual difference between the luminance value of the current coding unit CU and the matching block, where CurrCU[i] and RefCU[i] are the pixel values of the current coding unit CU and the matching block respectively, and W _CU is the CU size, that is, the number of pixels .

In addition, the histogram is generated as follows

Hist[i]=∑f(x,y,i)(x,y)∈CU,i∈[0,N _Pel ]

Wherein, Pixel_value(x, y) is the pixel value at the coordinate point (x, y) in the current coding unit CU.

Step 5: Using the histograms CurrHist[N _Pel ] and RefHist[N _Pel ] generated in step 4, pass

Calculate the sum SumHist _SAD of the differences between the current coding unit CU and the brightness histogram of the matching block.

Step 6: Use the SumPixel _SAD and SumHist _SAD calculated in Step 4 and Step 5 to make the following judgments. If any of the following two formulas is satisfied, it is considered that the brightness change between the current coding unit CU and the matching block is not obvious, and local brightness compensation is not required. , assign value Cur_IC _Flag =false and go to step 9; otherwise, Cur_IC _Flag maintains the original value, which is true, and continues to the next step;

SumHist _SAD /W _CU <λ ₀

SumPixel _SAD /W _CU <λ ₁

In the formula, λ ₀ and λ ₁ are the thresholds (through a large number of experimental statistics, when λ ₀ =0.5, λ ₁ =5, the encoding effect is the best), and W _CU is the current CU size, that is, the number of pixels.

Step 7: If none of the formulas in step 6 are satisfied, obtain the top, left, and top left of the current coding unit CU, and the co-located blocks CU _Above , CU _Left , CU _AboveLeft , and CU _Col of the reference frame. These adjacent blocks and co-located blocks have been encoded, and their LIC flag bit IC _Flag (indicating whether a CU block performs local brightness compensation) can be obtained, according to

Wherein, CU _x ∈ {CU _Above , CU _Left , CU _AboveLeft , CU _Col }, the LIC mode flag Cur_IC _Flag of the current coding unit CU is calculated. If Cur_IC _Flag = true, proceed to the next step, otherwise, go to step nine.

Step 8: The encoder performs FRUC Merge mode and 2N×2N mode with LIC (with LIC: for local brightness compensation), and determines the best mode according to the rate-distortion cost.

Step 9: The encoder performs an integer-pixel IMV_2N×2N mode without LIC, still determines the best mode according to the rate-distortion cost, and obtains the LIC information IC_Flag_temp of the best mode. Then select the follow-up operation according to Cur_IC _Flag , if Cur_IC _Flag =true, proceed to the next step, otherwise, go to step eleven.

Step 10: The encoder performs IMV_2N×2N mode with LIC.

Step 11: End the current algorithm and make subsequent judgments on other modes.

Embodiment Three

A fast inter-frame coding mode selection method based on local brightness compensation LIC, the specific process is as shown in Figure 2, including the following steps:

Step 1: In inter-frame prediction, the JEM encoder sequentially performs Affine Merge, normal Merge, FRUC Merge and 2N×2N modes without LIC (no LIC: no local brightness compensation), and then decides the best value according to the rate-distortion cost. best mode;

Step 2: first obtain the LIC mode flag of the current CU, so as to initialize the LIC flag Cur_IC _Flag of the current block, if Cur_IC _Flag = true, then continue to step 3; otherwise, go to step 9;

Step 3: Use the best mode obtained in step 1 to obtain the motion vector MV, and then perform motion compensation on the co-located block of the reference frame, that is, translate the co-located block according to the MV to obtain the matching block RefCU of the current CU;

Step 4: Use the matching block RefCU obtained in Step 3 to obtain the brightness values of each pixel in the current block and the matching block. According to this, the brightness histograms CurrHist[N _Pel ] and RefHist[N _Pel ] are generated, where N _Pel is the maximum value of the brightness in the video sequence. for 255. Additionally, by

Calculate the SumPixel _SAD of the luminance value residuals between the current block and the matching block, where CurrCU[i] and RefCU[i] are the pixel values of the current block and the matching block respectively, and W _CU is the CU size, that is, the number of pixels.

In addition, the histogram is generated as follows

Hist[i]=∑f(x,y,i)(x,y)∈CU,i∈[0,N _Pel ]

Wherein, Pixel_value(x, y) is the pixel value at the coordinate point (x, y) in the current CU.

Step 5: Using the histograms CurrHist[N _Pel ] and RefHist[N _Pel ] generated in step 4, pass

Compute the SumHist _SAD of the differences between the brightness histograms of the current block and the matching block.

Step 6: Use the SumPixel _SAD and SumHist _SAD calculated in Step 4 and Step 5 to make the following judgments. If any one of the two formulas is satisfied, it is considered that the brightness change between the current coding unit CU and the matching block is not obvious, and local brightness compensation is not required. Assign value Cur_IC _Flag =false and go to step nine; otherwise, Cur_IC _Flag maintains the original value, which is true, and continues to the next step;

SumHist _SAD /W _CU <λ ₀

SumPixel _SAD /W _CU <λ ₁

In the formula, λ ₀ and λ ₁ are the thresholds (after a lot of experimental statistics, when λ ₀ =0.5 and λ ₁ =5, the encoding effect is the best), and W _CU is the current CU size, that is, the number of pixels.

Step 7: If none of the formulas in step 6 is satisfied, obtain the upper, left, upper left of the current CU, and the co-located blocks CU _Above , CU _Left , CU _AboveLeft , and CU _Col of the reference frame. These adjacent blocks and co-located blocks have been encoded, and their LIC flag bit IC _Flag (indicating whether a CU block performs local brightness compensation) can be obtained, according to

Wherein, CU _x ∈ {CU _Above , CU _Left , CU _AboveLeft , CU _Col }, the LIC mode flag Cur_IC _Flag of the current coding unit CU is calculated. If Cur_IC _Flag = true, proceed to the next step, otherwise, go to step nine.

Step 8: The encoder performs FRUC Merge mode and 2N×2N mode with LIC (with LIC: for local brightness compensation), and determines the best mode according to the rate-distortion cost.

Step 9: The encoder performs an integer-pixel IMV_2N×2N mode without LIC, still determines the best mode according to the rate-distortion cost, and obtains the LIC information IC_Flag_temp of the best mode. Then select the follow-up operation according to Cur_IC _Flag , if Cur_IC _Flag =true, proceed to the next step, otherwise, go to step 12.

Step 10: According to the IC_Flag_temp obtained in step 9, if it is true, go to the next step directly, otherwise, go to step 12, consider that the current coding unit CU does not need to perform the LIC mode selection process, and assign Cur_IC _Flag = false;

Step 11: The encoder performs IMV_2N×2N mode with LIC.

Step 12: End the current algorithm and make subsequent judgments on other modes.

The blocks described in the examples are coding units.

In order to verify the correctness and effectiveness of the present invention, the method described in Embodiment 3 is implemented on the Visual Studio 2015 software based on the reference software JEM4.0. The configuration of the specific encoding parameters of all experiments uses the JEM standard configuration file: encoder_lowdelay_jvet10.cfg and the standard configuration file corresponding to the test sequence.

In order to verify the performance of the algorithm, two indicators, BDBR (Bjotegaard Delta Bit rate) and ΔT, are used for evaluation. Among them, BDBR is used to evaluate the impact of the algorithm on the video quality. The larger the BDBR, the greater the impact of the algorithm on the video quality, that is, the worse the performance of the algorithm. It is mainly obtained by setting four sets of different quantization parameters QP. Different Bits and PSNR are used for calculation. ΔT reflects the improvement of the encoder efficiency by the current algorithm, and its calculation formula is as follows:

Among them, T _org represents the encoding time of the original encoder without any fast algorithm, T _new represents the time required for encoding after the fast algorithm is accelerated, and ΔT represents the percentage of encoder efficiency improvement after the fast algorithm is accelerated.

Through experimental simulation, the experimental results are shown in Table 1, ΔBits% is the percentage change of the bit rate compared with the traditional encoder, and ΔPSNR/dB is the change of the peak signal-to-noise ratio compared with the traditional encoder.

Table 1 Experimental results

According to Table 1 of the experimental simulation results, it can be seen that after applying the method of the present invention, the encoding time is reduced by 24%, while the increase of BDBR is only 0.49. From the experimental results, it can be seen that the present invention greatly improves the coding efficiency under the premise of ensuring the subjective quality of the video, and achieves the purpose of the present invention.

The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which the present invention belongs can make various modifications or supplements to the described specific embodiments or adopt similar methods to replace them, but they will not deviate from the spirit of the present invention or go beyond the definition of the appended claims range.

Claims (6)

1. the coding list in a kind of quick interframe coding mode selection method that LIC is compensated based on local luminance, all frame of video Member carries out no LIC Affine Merge patterns, common Merge patterns, FRUC Merge patterns and 2N × 2N patterns successively, obtains The rate distortion costs of coding unit under each pattern are obtained, the corresponding pattern of minimum rate distortion costs are selected as the volume of coding unit Pattern；If the LIC mode flags position of frame where current coded unit is true, current coded unit proceeds have LIC's FRUC Merge patterns and 2N × 2N patterns, it is characterised in that obtain current coded unit CU and match coding unit with reference frame Brightness value residual sum SumPixel between CU_SADWith the difference sum SumHist of brightness histogram_SADIf, current coded unit Brightness value residual error rate, in CULess than the histogram difference rate in setting first threshold or current coded unit CULess than Second Threshold is set, then current coded unit skips LIC patterns；

Reference frame matching coding unit CU is the motion vector MV under selected coding mode using current coded unit, right Reference frame carries out the obtained coding unit of motion compensation with position block, and the same position block refers to be in reference frame, and position and works as The same coding unit in the position of preceding coding unit in the current frame；

The current coded unit CU matches the brightness value residual sum SumPixel between coding unit CU with reference frame_SADIt is logical Cross and the difference summation of the pixel value between coding unit CU between corresponding pixel points is matched to current coded unit CU and reference frame obtain ；

The difference sum that the current coded unit CU matches the brightness histogram between coding unit CU with reference frame is to pass through The difference summation that the histogram value between coding unit CU between corresponding grey scale level is matched to current coded unit CU and reference frame is obtained ；

W_CUFor the pixel number in current coded unit CU, first threshold value is 1-10, and Second Threshold value is 0-2.

2. according to the method described in claim 1, it is characterised in that before LIC patterns are carried out, first calculate current coded unit CU Adjacent encoder unit LIC flag bits be genuine total quantityJudge whether to meet again：If full Foot, then current coded unit CU skips LIC patterns, otherwise, and current coded unit proceeds the FRUCMerge patterns for having LIC With 2N × 2N patterns；

<mrow> <msub> <mi>Cur</mi> <mrow> <msub> <mi>IC</mi> <mrow> <mi>F</mi> <mi>l</mi> <mi>a</mi> <mi>g</mi> </mrow> </msub> </mrow> </msub> <mo>=</mo> <mi>&amp;Sigma;</mi> <mi>f</mi> <mrow> <mo>(</mo> <msub> <mi>CU</mi> <mi>x</mi> </msub> <mo>)</mo> </mrow> </mrow>

Wherein, CU_xRepresent the adjacent cells of current coded unit, x ∈ { Left, AboveLeft, Above, Col }, CU_Left、 CU_AboveLeft、CU_Above、CU_ColThe left adjacent block of current coded unit, upper left adjacent block, upper adjacent block are represented respectively and with position block, institute Same position block is stated to refer to be in reference frame, and the coding unit as the position of position and current coded unit in the current frame；f (CU_x) represent current coded unit adjacent cells LIC flag bits, if very, value is 1, otherwise, value is 0.

3. method according to claim 2, it is characterised in that current coded unit carries out the FRUC Merge moulds for having LIC After formula and 2N × 2N patterns, followed by IMV_2N × 2N patterns without LIC of progress, corresponding rate distortion generation under all patterns is obtained Valency, and optimal mode is used as using the corresponding pattern of minimum rate distortion costs；If optimal mode is the pattern handled by LIC, and Adjacent encoder unit LIC flag bits are genuine total quantityIt is not 0, then current coded unit proceeds have LIC's IMV_2N × 2N patterns；Otherwise, current coded unit skips LIC IMV_2N × 2N patterns.

4. a kind of quick interframe encoding mode selection device that LIC is compensated based on local luminance, it is characterised in that including：

Rate distortion costs computing unit, the Affine for carrying out no LIC successively to the coding unit in all frame of video Merge patterns, common Merge patterns, FRUC Merge patterns and 2N × 2N patterns, obtain the rate of coding unit under each pattern Distortion cost；

Selecting unit, by selecting the corresponding pattern of minimum rate distortion costs as the coding mode of coding unit；

Unit is skipped, if the LIC mode flags position of frame where current coded unit is true, is currently compiled according to following rule judgment Whether code unit skips LIC patterns：

If the brightness value residual error rate, in current coded unit CULess than setting first threshold or current coded unit Histogram difference rate in CULess than Second Threshold is set, then current coded unit skips LIC patterns；

Wherein, SumPixel_SADAnd SumHist_SADRespectively represent current coded unit CU matched with reference frame coding unit CU it Between brightness value residual sum and brightness histogram difference sum；

Reference frame matching coding unit CU is the motion vector MV under selected coding mode using current coded unit, right Reference frame carries out the obtained coding unit of motion compensation with position block, and the same position block refers to be in reference frame, and position and works as The same coding unit in the position of preceding coding unit in the current frame；

The current coded unit CU matches the brightness value residual sum SumPixel between coding unit CU with reference frame_SADIt is logical Cross and the difference summation of the pixel value between coding unit CU between corresponding pixel points is matched to current coded unit CU and reference frame obtain ；

The current coded unit CU matches the difference sum of the brightness histogram between coding unit CU with reference frame SumHist_SADIt is by matching the Nogata between coding unit CU between corresponding grey scale level with reference frame to current coded unit CU The difference summation of map values is obtained；

W_CUFor the pixel number in current coded unit CU, first threshold value is 1-10, and Second Threshold value is 0-2.

5. device according to claim 4, it is characterised in that before LIC patterns are carried out, by counting current coded unit The LIC flag bits of CU adjacent encoder unit are genuine total quantityAnd judgeWhether it is 0, if 0, then Current coded unit CU skips LIC patterns, otherwise, and current coded unit proceeds FRUC Merge patterns and the 2N for having LIC × 2N patterns.

6. device according to claim 5, it is characterised in that also choose unit including optimal mode, if that chooses is optimal Pattern belongs to the pattern handled by LIC, and adjacent encoder unit LIC flag bits are genuine total quantityIt is not 0, then Current coded unit proceeds the IMV_2N × 2N patterns for having LIC, otherwise, skips LIC IMV_2N × 2N patterns；

It is the FRUC Merge patterns and 2N × 2N patterns for having LIC in current coded unit progress that the optimal mode, which chooses unit, Afterwards, followed by IMV_2N × 2N patterns without LIC of progress, corresponding rate distortion costs under all patterns are obtained, and with minimum rate The corresponding pattern of distortion cost selects optimal mode.