• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Unified FPGA Design for the HEVC Dequantization and Inverse Transform Modules

    2022-08-23 02:15:38TurkiAlanaziandAhmedBenAtitallah
    Computers Materials&Continua 2022年6期

    Turki M.Alanazi and Ahmed Ben Atitallah

    Department of Electrical Engineering,Jouf University,Sakaka,Aljouf,2014,Saudi Arabia

    Abstract:As the newest standard,the High Efficiency Video Coding(HEVC)is specially designed to minimize the bitrate for video data transfer and to support High Definition (HD) and ULTRA HD video resolutions at the cost of increasing computational complexity relative to earlier standards like the H.264.Therefore,real-time video decoding with HEVC decoder becomes a challenging task.However, the Dequantization and Inverse Transform(DE/IT) are one of the computationally intensive modules in the HEVC decoder which are used to reconstruct the residual block.Thus,in this paper,a unified hardware architecture is proposed to implement the HEVC DE/IT module for all Transform Unit(TU)block size,including 4×4,8×8,16×16 and 32×32.This architecture is designed using the High-Level Synthesis(HLS)and the Low-Level Synthesis(LLS)methods in order to compare and determine the best method to implement in real-time the DE/IT module.In fact, the C/C++ programming language is used to generate an optimized hardware design for DE/IT module through the Xilinx Vivado HLS tool.On the other hand, the LLS hardware architecture is designed by the VHSIC Hardware Description language(VHDL)and using the pipeline technique to decrease the processing time.The experimental results on the Xilinx XC7Z020 FPGA show that the LLS design increases the throughput in term of frame rate by 80% relative to HLS design with a 4.4% increase in the number of Look-Up Tables(LUTs).Compared with existing related works in literature,the proposed architectures demonstrate significant advantages in hardware cost and performance improvement.

    Keywords: HEVC decoder; dequantization; IDCT/IDST; LLS design; HLS design;FPGA

    1 Introduction

    Nowadays, several consumer electronic devices such as television [1], smartphone [2], tablet [3],etc use video standard codec for video compressing and transmitting data with minimum bitrate.In this context,a High Efficiency Video Coding(HEVC)video standard is developed[4,5].The HEVC is a video compression standard that can provide a good performance than the previous standard,i.e.,H.264/AVC [6].In fact, it can support a high video resolution which can reach to 8K (7680×4320)and reduces the bitrate approximately by 50%relative to H.264/AVC[7,8]with same video quality.

    In video standard codec the Dequantization and Inverse Transform(DE/IT)play a very important role to reconstruct the compressed video sequences [9].Nevertheless, in HEVC video decoder, the DE/IT module allows to reconstruct the residual block which can be with several size 4×4, 8×8,16×16, and 32×32.In fact, the largest coding unit in HEVC can be up to 64×64 in size, and the Transform Unit(TU)sizes can be 4×4,8×8,16×16,and 32×32[10].This multiple TU sizes improve the compression performance but increase the computational complexity to reach a real-time execution [11,12].In this context, based on the complexity analysis of the HEVC decoder modules for all-intra configuration performed in[13],we can notice that the entropy decoding(ED),the intra prediction(IP)and the DE/IT modules consume on average 38%,32%and 20%of the total decoding time,respectively.However,according to the analysis given in[14]for the ED module,it is obviously that the computational complexity presented in the regular CABAC algorithm makes it difficult to be pipelined and parallelized in such hardware context.This complexity is caused by the critical bin-tobin data dependencies and the use of feedback between entropy decoding steps.On the other hand,we have proposed in[13]a hardware architecture to implement the IP block.For this reason,we focus in this work to propose an efficient hardware design to reduce the computational complexity of the HEVC DE/IT module.

    Recently, field-programmable gate arrays (FPGAs) have been gaining popularity for image and video processing.Indeed, modern FPGAs have sufficient resources to implement a complex application[15,16].The traditional approach used to design and implement any algorithm in FPGA is Low-Level Synthesis(LLS)using hardware description language(HDL)such as VHSIC hardware description language (VHDL or Verilog).With such low-level design, it is possible to adjust the Register Transfer Level (RTL) description to generate an optimized hardware architecture [17–19].But,this kind of design requires a lot of time and effort specially for complex algorithm.To alleviate this problem,the High-Level Synthesis(HLS)is introduced[20–22].Nevertheless,the HLS flow uses the high-level programming language like C/C++, systemC, etc to generate automatically the RTL design.This approach makes the code more readable, shortens design and verification times, and increases the design reusability over those of handwritten HDL equivalents.

    In literature, many architectures are proposed to implement the dequantization and inverse transform modules for HEVC decoder.In fact, the design outlined in [23] presents an FPGA implementation of the HEVC 2D integer inverse discrete cosine transform(2D-IDCT)using different HLS tools.The developed design can decode 54 frames/sec(FPS)for 1080p video sequences.In[24],the authors provide a System-On-Chip FPGA platform based on Xilinx Zynq to integrate the DCT coding block as an accelerator with HLS tool.The proposed design is capable to perform the coding of 1080@30fps.In [25], a LLS hardware architecture for 2D-IDCT is proposed.This architecture can process 4K@28fps at 135 MHz under XC7Z045 Xilinx FPGA.In addition, Chen et al.[26]design a 2D-IDCT architecture that supports all TU sizes.This architecture can calculate two rows in parallel during the 1D-IDCT instead of only one.In this case,the maximum throughput achieved is about 4K@30fps with the Xilinx Zynq platform.But Liang et al.[27] present an architecture that supports the 2D integer inverse discrete sine transform (2D-IDST) and 2D-IDCT using two 1DIDCT/IDST units and memory block.This architecture calculates four residual pixels in parallel in each clock cycle.This design can decode 7680×4320@30fps.In [28] a hardware implementation of the 2D dequantization,IDCT and IDST(2D-DE/IDCT/IDST)for HEVC decoder is described.The proposed architecture can perform the decoding of 4K@30fps at 200 MHz clock frequency in TSMC 40 nm technology.However, such implementation presents a good compromise in terms of energy efficiency,resources utilization and performance but miss flexibility in term of programmability.On the other,in[29] an efficient OpenCL implementation of the HEVC 2D-DE/IDCT/IDST module is proposed.This software implementation is realized based on an exploitation of the parallel processing offered by modern Graphic Processing Unit (GPUs).The proposed solution can decode 4K@15fps on GeForce GTX 780Ti@1046 MHz.This implementation preserves the flexibility and accuracy of software but with an increase in power consumption.

    When he reached the first court of the castle he saw before him a flight of agate20 steps, and went up them, and passed through several splendidly furnished rooms

    Hence, the aim of this paper is to provide a unified and optimized hardware architecture to implement the 2D-DE/IDCT/IDST module for HEVC decoder.This architecture should support 4×4, 8×8, 16×16 and 32×32 HEVC TU block size and offer a trade-off between performance,hardware cost and processing time.For this, the LLS and HLS design flow are used and explored to design a hardware architecture for HEVC 2D-DE/IDCT/IDST module.However, the HLS 2DDE/IDCT/IDST design is explored through the Xilinx Vivado HLS 2018.1 tool by adding specific directives (e.g., PIPILINE, RESSOURCE, etc) to the high level programming language such as C/C++code.But,the LLS 2D-DE/IDCT/IDST design is developed by using the pipeline technique through the VHDL language.The hardware architectures for both methods are mapped and evaluated on Xilinx XC7Z020 FPGA for processing time and hardware cost in order to determine which design method(LLS or HLS)provides better design productivity when facing a complex algorithm like the 2D-DE/IDCT/IDST module for HEVC decoder.

    He nodded to them, saying, It must be rather dull standing28 on the stairs; I would rather go inside! The halls blazed with lights; councillors and ambassadors were walking about in noiseless shoes carrying gold dishes

    The remainder of the paper is structured as follows.Section 2 introduces an overview of HEVC 2D-DE/IDCT/IDST module.Section 3 describes the hardware architecture designed for 2D-DE/IDCT/IDST module using HLS and LLS design flow.The implementation results and performance evaluation are reported in Section 4.Finally,Section 5 concludes the paper.

    In order to improve the design performances, several directives are added incrementally to the HEVC 2D-DE/IDCT/IDST C code.A part of the C code developed and given as input to Xilinx Vivado HLS tool 18.1 is shown in Figs.4 and 5 used for dequantization and 1D-IDCT8,respectively.Thus, several designs are generated to implement the 2D-DE/IDCT/IDST module.In fact, Design 1 is generated without adding any directive.The synthesis results on the Xilinx XC7Z020 FPGA shows that this design uses 21466 (40.3%) of Look-Up Tables (LUTs), 10252 (19.3%) of Flip-Flops(FFs), 40 (18.2%) of DSP blocks and 30 (21.4%) of BRAM as illustrated in Fig.6 and needs in worst case 15562 clock cycles to reconstruct the 32×32 residual block (Fig.7).From these results,we notice that the number of clock cycles is high.For that, the PIPELINE directive should be used to decrease the number of clock cycles.In fact,from Figs.4 and 5,we can see that the dequantization and IDCT C codes use loop iteration to reconstruct the transom and residual coefficients.The length of the loop iterations depends of the TU size.So, the loop can reach 1024 iterations.However, the PIPLINE directive is applied to loop iterations with interval equal to 1 to reduce the cycles number of latencies required for an input data to reach its output.Consequently, the Design 2 is created.According to Figs.6 and 7,Design 2 allows to decrease by 91%the number of clock cycles but with an important increase in the hardware cost by 45%of LUTs,62.3%of FFs,47.4%of DSP blocks and 44.5%of BRAMs relative to Design 1.Therefore,to reduce the hardware cost,the Design 3 is generated by using the ALLOCATION and RESSOURCE directives.Indeed,the ALLOCATION directive is added to process the multiplication operations in the level of dequantization equation which allows to share the hardware resources between several operations.Furthermore,the RESOURCE directive is used to implement the residual and dequantization arrays by a specific memory block (BRAMs).This optimization has allowed to decrease the hardware cost by 42%of LUTs,30.8%of FFs,42.1%of DSP blocks and 55.5% of BRAMs but with an increase by 61% in the number of clock cycles compared to Design 2 as reported in Figs.6 and 7.On the other hand,we can constate that the Design 3 provides a decrease in number of clock cycles by about 67%in worst case but with an increase by 5%of LUTs,45%of FFs and 9%of DSP blocks relative to Design 1.Afterward,the Design 3 is used for the comparative study with LLS 2D-DE/IDCT/IDST because it offers a trade-off between hardware cost and processing time.

    2 Dequantization and Inverse Transform in HEVC Decoder

    In HEVC, each frame is partitioned into coding tree block structure involving different sizes of large coding units(LCUs)up to 64×64.As illustrated in Fig.1,each LCU can be recursively split into several sizes of coding units(CUs).Starting from a defined partition of CUs,it is possible to further generate two other types of units corresponding to prediction unit(PU),which is used for inter/intra prediction processes,and TU,which is the elementary unit of dequantization and inverse transform.However,the size of the TU block is defined by HEVC encoder and can be 4×4,8×8,16×16 and 32×32.

    The hardware architecture depicted in Fig.8 describes the LLS design of the HEVC 2DDE/IDCT/IDST module.In fact, this architecture is composed by four dequantization units(Unit_DQ0, Unit_DQ1, Unit_DQ2 and Unit_DQ3), 1D-IDCT/IDST unit, transpose memory unit and control unit.It receives as input four 16-bit coefficients(Coeff0,Coeff1,Coeff2 and Coeff3)each three clock cycles from the entropy decoder,size of TU block(selsignal),QP value(QPsignal)andstartsignal and generates as output the residual block andDone_DQ/IDCTsignal which indicates that the residual block is ready.However,the designed architecture contains the dequantization and 1D-IDCT/IDST units in one design in order to benefit of the pipeline technique which can decrease the processing time.

    If you do not come, and if I do not hear from you, then I shall know that you could never be happy so far from the people and the country which you love

    Figure 1:Coding structure in HEVC standard

    Table 1: Definition of f(QP%6)

    After dequantization,the 2D-IDCT is performed.In fact,the IDCT module takes dequantized coefficient and performs as two separate 1D-IDCT to obtain is outputs the residual block.The HEVC decoder supports two types of inverse transform which are IDCT and IDST.The IDST is applied only to the 4×4 TU block.During decoding,the transformed coefficients are converted back to the spatial domain via an inverse transform.According the HEVC,the 2D-IDCT/IDST can be expressed by Eq.(3).

    When the kids came in, he took them for walks along the pier21 near their office. Often she went along and watched Eric, who was becoming a master of sign language, talk and laugh with her boys as no one else had before.

    where SRC is the transformed block, DST is the residual block and M is a N×N IDCT/IDST coefficient matrix which defined by HEVC standard.The value N is depending of the TU size.For each TU size,it has its own transform matrix.But,the 32×32 TU’s transform matrix includes other sizes TU’s transform matrix.In HEVC,the 2D-IDCT/IDST is computed by two 1D-IDCT/IDST where a column inverse transform is followed by a row inverse transform.

    To decrease the implementation complexity of 2D-IDCT/IDST,Chen et al.[30]was proposed a fast algorithm for transform by decomposing the transform matrixMNto some zero matrixes.This decomposition technique is presented by Eq.(4).

    wherePNis the permutation matrix.BNis theN-point butterfly structure.MN,oddandMN,evenare the odd and even parts ofMN, respectively.For example, Eq.(5) illustrates the 8×8 coefficient matrix in HEVC and Eq.(6)the decomposition of this matrix.Fig.2 presents the flowchart of 8-points 1DIDCT based on Chen’s algorithm.

    where:

    Figure 2:Flowchart of 8-points 1D-IDCT based on chen’s algorithm

    3 2D-DE/IDCT/IDST Hardware Architecture

    In this section,we describe the HLS and LLS hardware architectures designed to implement the HEVC 2D-DE/IDCT/IDST algorithm on Xilinx XC7Z020 FPGA.In this work,the HEVC test Model(HM16.0)[31]is used as reference software.

    3.2.2 1D-IDCT/IDST

    3.1 HLS 2D-DE/IDCT/IDST Hardware Architecture

    HLS is gaining more and more popularity specially when using FPGA circuit.Nevertheless,with HLS,it becomes possible to reduce the conception and validation time of the hardware design.Therefore,the exploration and the simulation of multiple hardware architectures can be done in the shortest time.But,HLS requests designers to restructure programs,change a source code and add a specific directive to get a good result.In this context,Xilinx developed the Vivado HLS tool.This tool accepts as input a high-level programming language such as C/C++and generates automatically as output an RTL hardware description.Through this tool,it is possible to add several directives(such as LOOP unrolling,ALLOCATION,RESOURCE,etc)in order to generate an optimize RTL design in terms of hardware cost and processing time.

    3.2.1 Unit_DQ

    Figure 3:Dataflow of the HLS 2D-DE/IDCT/IDST module for HEVC decoder

    However, in the beginning, the HLS architecture receives as input the TU size, the correspond coefficients(maximum 1024 coefficients)and the QP value.Then,these coefficients are dequantized to generate the transform coefficients.After that,if the TU size is egal to 4×4,thus in the first step,the 1D-IDCT4/IDST4 will be applied to the columns of TU to generate the 1D-transfrom coefficients.In the second step,these coefficients are stored in transpose memory to be used for 2D-transfrom.In the last step,the 1D-IDCT4/IDST4 will be applied to the row of TU to reconstruct the residual block.But,if the TU size is equal to 8×8,16×16 or 32×32,so the 4-point odd,8-point odd and 16-point odd are used with 4-point even and butterfly module to produce 1D/2D-IDCT8/16/32 coefficients,respectively.

    “Dance you shall,” said he, “dance in your red shoes till you are pale and cold, till your skin shrivels up and you are a skeleton! Dance you shall, from door to door, and where proud and wicked children live you shall knock, so that they may hear you and fear you! Dance you shall, dance—!”

    Woe82 to him who chooses the middle path! if he had a thousand lives he would not save one; it is very hazardous83; it leads to the Caucasus, and is an endless road

    Figure 4:Dequantization Vivado HLS C code

    Figure 5:8-point IDCT Vivado HLS C code

    Figure 6:HLS synthesis results of HEVC 2D-DE/IDCT/IDST module

    3.2 LLS 2D-DE/IDCT/IDST Hardware Architecture

    Thus,the 2D-DE/IDCT/IDST module receives the coefficients of the TU block from the entropy decoder and applies the dequantization to restore the original Transform coefficients.The dequantization scheme as specified be HEVC is given by Eq.(1).

    Figure 7:Number of clock cycles to reconstruct 4/8/16/32 residual block

    Figure 8:LLS hardware architecture for HEVC 2D-DE/IDCT/IDST module

    For the HLS implementation of the HEVC 2D-DE/IDCT/IDST module, the C code of this module is extracted from HM16.0.The 2D-DE/IDCT/IDST algorithm is implemented with HLS based on the algorithm proposed in Fig.3.In fact, some characteristics are taken account in this algorithm to reduce the hardware cost and complexity and generate a performant HLS design for 2DDE/IDCT/IDST module.However,the HLS design supports 4/8/16/32 TU size.The dequantization and 2D-IDCT/IDST are integrated in one design to minimize the processing time.Moreover,the evenodd 1D-IDCT/IDST algorithm is used as shown in Fig.2.In addition,8-point 1D-IDCT is computed by using 4-point IDCT (even part), 4-point odd and butterfly module.Even for 16-point IDCT is calculated based-on 8-point 1D-IDCT, 8-point odd and butterfly module.Also, 32-point IDCT is determined from 16-point 1D-IDCT,16-point odd and butterfly module.Further,all multiplication operations in 1D-IDCT/IDST are replaced by shift and addition based on the study realized in[25].Furthermore,1D-IDCT/IDST module is called two time and the intermediate memory block is used to save and transpose coefficients to generate 2D-IDCT/IDST.All these characteristics allow to reduce the hardware cost and hardware complexity.

    Fig.9 presents the hardware architecture of Unit_DQ.This architecture is developed based on Eq.(1).Indeed, two Read-Only-Memory (ROM1 and ROM2) are used to store the precalculated values proportional to QP%6 and QP/6,respectively.These memories are addressed based on QP value through the control unit.However, the 16-bit level signal receives the quantized coefficient which is multiplied by the value collected for ROM1.Then,the obtained result is shifted by the value collected from ROM2, added to the offset value and shifted by offset1.In the end, the 16-bit dequantization coefficient is obtained in two clock cycles through CoeffQ signal.

    Figure 9:Hardware architecture of Unit_DQ

    When the Emperor heard from the Prince how he had gained possession of his fair prize, he at once recognized that he had been helped by some magic art, and on the spot gave up all claim to the beautiful mermaid

    Fig.10 depicts the 1D-IDCT/IDST hardware architecture.This architecture is designed to support 4×4, 8×8, 16×16 and 32×32 TU size and use even-odd IDCT decomposition.In fact, the 32-point 1D-IDCT is constructed by using a 4-point even, a 4-point odd, an 8-point odd and a 16-point odd.Moreover,to reduce the hardware cost,the multiplication operations are replaced by shift and addition.Also, when analyzing the IDCT equation, we constate that there are some repetitive coefficients (X2, X4, X9, X18, X36, X64, X90).The Xcoeff components is used to compute these coefficients which are used to deduce the other cofficients(i.e.,:X75=X64+X9+X2).However,1DIDCT/IDST architecture receives dequantized coefficients through 32 SRC signals and produces the transformed coefficients through 32 DST signals in 12 clock cycles, 24 clock cycles, 80 clock cycles and 160 clock cycles for 4/8/16/32 TU size,respectively.

    where coeffDQ is the dequantized coefficient, level is the quantized DCT coefficient, QP is the quantization parameter ranged from 0 to 51 (every TU has its own QP value), N is the size of TU block,B is the bit depth and the function f(QP%6)is determined from Tab.1.The coeffDQ is clipped to the range[-32768,32768]to guarantee that the dequantized coefficient is computed with 16-bit.

    Figure 10:1D-IDCT/IDST hardware architecture

    3.2.3 Transpose Memory

    The transpose memory is used to store the intermediate coefficients between column and row of the inverse transform.It can store the coefficients of all TU size.The access to memory is optimized by concatenation eight 16-bit coefficients.In fact,in one clock cycles,it is possible to write and read 128-bit which mean eight coefficients in same time.

    3.2.4 Control Unit

    The control unit serves to share and synchronize data between all units in our design as shown in Fig.11.Accordingly,for 4×4 TU size,in the first step the control unit send four 16-bit coefficients(1stcolumn)to the dequantized units.Then,in the second step,the dequantized units receive the 2sdcolumn after 3 clock cycles and the 1D-IDCT4/IDST4 process the 1stcolumn in 3 clock cycles.After that,in the third step,1stcolumn is concatenated and stored in transpose memory in one clock cycle,the 2sdcolumn is processed by inverse transform in 3 clock cycles and the 3thcolumn is treated by the dequantized units in 2 clock cycles.Thus,the pipeline technique is used between all units to optimize the processing time.So,for 4×4 TU size, firstly the TU is processed column by column by dequantized units and 1D-IDCT4/IDST4 and the output coefficients for each column are stored in transpose memory.This step needs 16 clock cycles.Then,the 1D-IDCT4/IDST4 is performed again row by row from transpose memory.In the end,the DE/IDCT of 4×4 TU size is obtained in 29 clock cycles.All these steps are used for 8/16/32 TU size and need 77 clock cycles,280 clock cycles and 938 clock cycles,respectively as shown in Fig.11.

    Figure 11:Timing diagram for HEVC 2D-DE/IDCT/IDST module

    4 Implementation Results and Performance Evaluation

    Tab.2 conducts a comparison of the synthesis results,number of clock cycles and clock frequency for HLS and LLS 2D-DE/IDCT/IDST designs under XC7Z020 FPGA for 4×4,8×8,16×16 and 32×32 TU size.It can be seen from this table that the HLS design uses less LUTs by 4.4%but more FFs,RAMs and DSPs by 9.5%,5.7%and 18.2%,respectively,relative to LLS design.Moreover,the LLS design allows to reduce the number of clock cycles by 64%compared to HLS design.

    Table 2: Synthesis results of HLS and LLS flow for 2D-DE/IDCT/IDST under XC7Z020 FPGA

    On the other hand,the performance of HLS and LLS design for HEVC 2D-DE/IDCT/IDST is measured for several class of video sequences such as Class A(2560×1600),Class B(1920×1080),Class C(1280×720)and Class D(832×480).So,from Fig.12,we can see that the frame rate of LLS design can reach 339 fps for class D and 33 fps for class A instead of 66 fps for class D and 6 fps for class A using HLS design.However, with LLS design the frame rate is increased by 80% relative to HLS design in worst case as shown in Fig.13.

    Figure 12:Frame rate comparison between HLS and LLS designs for A,B,C and D video class

    Figure 13:Gain in frame rate for LLS relative to HLS for A,B,C and D video class

    Comparing our HLS 2D-DE/IDCT/IDST design with the HLS design proposed in[23]and[24]used to implement only the 2D-IDCT and 2D-DCT,respectively,we can constate from Tab.3 that our design allows a gain in LUTs and DSP blocks reaching approximately 55% and 59% relative to [23]and[24],respectively.Moreover,the maximal throughput of our design could perform the decoding of 1080p@13fps at 100 MHz instead of the 1080p@54fps at 208 MHz and 1080p@30fps found in[23] for 2D-IDCT and [24] 2D-DCT only, respectively.On the other hand, the designs [25] and [26]propose the LLS design for the 2D-DCT and 2D-IDCT/IDST only which can achieve 4K@28fps and 4K@30fps,respectively.But,our proposed LLS 2D-DE/IDCT/IDST design can reach 1080p@65fps.Hence,our design has better performance than[25]and[26]in term of frame rate and also use lower LUTs and DSP blocks count than[25]and[26]by 28%and 96%,respectively,as reported in Tab.3.Further,comparing the LLS 2D-DE/IDCT/DST design with[21],we can constate that our design can process 4K@16fps at 145 MHz instead of 4K@15fps on GeForce GTX 780Ti@1046 MHz in [29].So, our design is more performant and can achieves lower power consumption.Besides, our design permits more flexibility than[27]and[28]with the almost same performance.

    Table 3: Literature comparison of the dequantization and inverse transform designs

    Table 3:Continued

    5 Conclusion

    In this work, a unified hardware architecture is proposed to implement the HEVC 2DDE/IDCT/IDST module for 4/8/16/32 TU block size.However,two design methods are used to design this hardware architecture which are the HLS and the LLS design flow.Our goal was to compare these two methods and to select the best architecture to implement the HEVC 2D-DE/IDCT/IDST module.It is clear from experimental results under Xilinx XC7Z020 FPGA that the LLS design is more performant than HLS design in terms of processing time and hardware cost.But, the performance of HLS design depends on the selected directives, and the algorithm complexity and can be a good solution to speed up the design time and time to market(TTM).

    Funding Statement:This work was funded by the Deanship of Scientific Research at Jouf University(Kingdom of Saudi Arabia)under grant No.DSR-2021-02-0391.

    The spring sun had chased away the last snow from its hiding place under the hedges; the fields were full of flowers; nightingales sang in the trees, and all the world was gay

    But his mother told him that it was an unhappy man who had lost everything, and had come all this way to consult him, and bade the young man not to be afraid, but to come forward and show himself

    Conflicts of Interest:The authors declare that they have no conflicts of interest to report regarding the present study.

    99久久99久久久精品蜜桃| www.自偷自拍.com| 国产三级在线视频| 国产欧美日韩精品亚洲av| 狠狠狠狠99中文字幕| 国产极品粉嫩免费观看在线| 人人澡人人妻人| 亚洲一区中文字幕在线| 午夜福利18| 老汉色av国产亚洲站长工具| 欧美乱妇无乱码| 久9热在线精品视频| 午夜久久久久精精品| 成人国产综合亚洲| 亚洲色图av天堂| 亚洲精品一区av在线观看| 久久天堂一区二区三区四区| 国产亚洲欧美在线一区二区| 男人操女人黄网站| 久久人妻av系列| 九色亚洲精品在线播放| 国产xxxxx性猛交| 久久中文字幕一级| 一级作爱视频免费观看| 一级毛片女人18水好多| 国产精品亚洲一级av第二区| 两个人免费观看高清视频| 18禁美女被吸乳视频| 女人精品久久久久毛片| 精品卡一卡二卡四卡免费| 悠悠久久av| 在线免费观看的www视频| 亚洲av熟女| 成人国语在线视频| 一本综合久久免费| 99国产极品粉嫩在线观看| 国产三级黄色录像| 国产极品粉嫩免费观看在线| 丁香六月欧美| 国产精品综合久久久久久久免费 | 欧美激情高清一区二区三区| 日韩一卡2卡3卡4卡2021年| 国产成人免费无遮挡视频| 亚洲一区二区三区色噜噜| 国产精品,欧美在线| 在线观看午夜福利视频| 国产精品永久免费网站| 亚洲欧美精品综合一区二区三区| 国产又色又爽无遮挡免费看| av视频在线观看入口| 国产成人欧美在线观看| 欧美 亚洲 国产 日韩一| 国产av在哪里看| 女性生殖器流出的白浆| bbb黄色大片| 国产精品电影一区二区三区| 身体一侧抽搐| 亚洲精品美女久久久久99蜜臀| 桃红色精品国产亚洲av| 午夜久久久久精精品| 午夜两性在线视频| 久久中文看片网| 禁无遮挡网站| 精品久久久久久久久久免费视频| 夜夜看夜夜爽夜夜摸| 宅男免费午夜| 国产色视频综合| 日韩大尺度精品在线看网址 | 日韩一卡2卡3卡4卡2021年| 精品卡一卡二卡四卡免费| av天堂在线播放| 国产精品爽爽va在线观看网站 | 精品日产1卡2卡| 日韩国内少妇激情av| 亚洲国产毛片av蜜桃av| 无限看片的www在线观看| 一区二区三区高清视频在线| 淫秽高清视频在线观看| 国内精品久久久久久久电影| 咕卡用的链子| 亚洲国产精品sss在线观看| 免费女性裸体啪啪无遮挡网站| 久久久国产成人精品二区| 欧美性长视频在线观看| 一a级毛片在线观看| 狠狠狠狠99中文字幕| 亚洲精品国产一区二区精华液| 亚洲国产精品合色在线| 久久亚洲精品不卡| 视频在线观看一区二区三区| 九色国产91popny在线| 久久精品91蜜桃| 国产精品一区二区三区四区久久 | 精品国产一区二区久久| 国产亚洲精品av在线| 亚洲成人精品中文字幕电影| 亚洲aⅴ乱码一区二区在线播放 | 性欧美人与动物交配| 日韩国内少妇激情av| 日韩精品免费视频一区二区三区| 在线观看午夜福利视频| 一级毛片女人18水好多| 神马国产精品三级电影在线观看 | 久久久久久国产a免费观看| 国产成人一区二区三区免费视频网站| 国产熟女xx| 亚洲第一av免费看| 久久亚洲真实| 久久午夜综合久久蜜桃| 亚洲精品在线美女| av有码第一页| 国产97色在线日韩免费| 老熟妇乱子伦视频在线观看| 欧美午夜高清在线| 制服诱惑二区| 亚洲色图综合在线观看| 最好的美女福利视频网| av视频在线观看入口| 亚洲第一电影网av| 国产一区二区三区视频了| 欧美日韩中文字幕国产精品一区二区三区 | 好男人在线观看高清免费视频 | 91字幕亚洲| 国产精品九九99| 亚洲欧美激情综合另类| 在线十欧美十亚洲十日本专区| 日韩精品免费视频一区二区三区| 黄片大片在线免费观看| 国产激情欧美一区二区| 最好的美女福利视频网| 国产精品 国内视频| av天堂久久9| 色哟哟哟哟哟哟| videosex国产| 精品欧美一区二区三区在线| 国产亚洲av高清不卡| 91精品三级在线观看| 国产麻豆成人av免费视频| 好看av亚洲va欧美ⅴa在| 国产不卡一卡二| 日本 欧美在线| 精品第一国产精品| 自线自在国产av| 久久人妻福利社区极品人妻图片| 成人三级做爰电影| 亚洲五月色婷婷综合| 深夜精品福利| 老司机午夜十八禁免费视频| 少妇熟女aⅴ在线视频| 久热这里只有精品99| 亚洲色图 男人天堂 中文字幕| 国产成人免费无遮挡视频| 亚洲成人精品中文字幕电影| 一级,二级,三级黄色视频| 怎么达到女性高潮| 国产亚洲精品第一综合不卡| 免费搜索国产男女视频| 正在播放国产对白刺激| 在线观看舔阴道视频| 日本五十路高清| 男女下面插进去视频免费观看| 亚洲国产中文字幕在线视频| 亚洲国产日韩欧美精品在线观看 | 国产精品影院久久| 人成视频在线观看免费观看| 成人18禁高潮啪啪吃奶动态图| 悠悠久久av| 久久中文字幕人妻熟女| 大码成人一级视频| 欧美成人午夜精品| 亚洲中文av在线| 欧美一级毛片孕妇| 一边摸一边做爽爽视频免费| av有码第一页| 成人三级做爰电影| 欧美亚洲日本最大视频资源| 免费观看人在逋| 国产精品 欧美亚洲| 麻豆久久精品国产亚洲av| 女性生殖器流出的白浆| 成人亚洲精品av一区二区| 女同久久另类99精品国产91| 成人免费观看视频高清| 成人三级做爰电影| 国产成人欧美| 精品国产乱码久久久久久男人| 久久伊人香网站| 12—13女人毛片做爰片一| 制服丝袜大香蕉在线| 成人18禁在线播放| 大型黄色视频在线免费观看| 免费搜索国产男女视频| 中文字幕av电影在线播放| 亚洲第一av免费看| 精品欧美国产一区二区三| 国产亚洲精品av在线| 黄色女人牲交| 人人妻人人澡人人看| 国产精品乱码一区二三区的特点 | 色综合婷婷激情| 精品国产一区二区三区四区第35| 真人一进一出gif抽搐免费| 亚洲午夜理论影院| 午夜福利一区二区在线看| 麻豆国产av国片精品| 午夜福利影视在线免费观看| 亚洲av美国av| 欧美丝袜亚洲另类 | 国产精品香港三级国产av潘金莲| 制服诱惑二区| 女性被躁到高潮视频| 亚洲男人天堂网一区| 精品无人区乱码1区二区| 9热在线视频观看99| 国产午夜精品久久久久久| 成人亚洲精品av一区二区| 少妇粗大呻吟视频| 亚洲aⅴ乱码一区二区在线播放 | 国产日韩一区二区三区精品不卡| 久久久国产欧美日韩av| 久久午夜亚洲精品久久| 欧美 亚洲 国产 日韩一| 男女床上黄色一级片免费看| 久久精品国产99精品国产亚洲性色 | 久久国产精品男人的天堂亚洲| 嫁个100分男人电影在线观看| 久久久久久久久中文| 午夜福利一区二区在线看| 久久久精品国产亚洲av高清涩受| 一边摸一边抽搐一进一出视频| 亚洲精品美女久久av网站| 精品久久蜜臀av无| 亚洲成人久久性| 高清黄色对白视频在线免费看| 亚洲欧美日韩高清在线视频| 美女午夜性视频免费| 欧美人与性动交α欧美精品济南到| 亚洲最大成人中文| 色综合站精品国产| 久久午夜综合久久蜜桃| 国产在线观看jvid| 日韩有码中文字幕| 少妇 在线观看| or卡值多少钱| 久久午夜综合久久蜜桃| 搡老岳熟女国产| 午夜精品久久久久久毛片777| 久久久久国产精品人妻aⅴ院| 长腿黑丝高跟| www.精华液| 自线自在国产av| 十八禁人妻一区二区| 国产亚洲精品一区二区www| 久久精品aⅴ一区二区三区四区| 日韩有码中文字幕| 久久香蕉精品热| 久久人妻熟女aⅴ| 欧美最黄视频在线播放免费| 日韩欧美一区二区三区在线观看| 久久久国产成人精品二区| 麻豆久久精品国产亚洲av| 亚洲黑人精品在线| 99国产精品99久久久久| 亚洲男人天堂网一区| 精品午夜福利视频在线观看一区| 韩国精品一区二区三区| 成年人黄色毛片网站| 激情在线观看视频在线高清| av在线天堂中文字幕| 欧美久久黑人一区二区| 99re在线观看精品视频| 欧美色欧美亚洲另类二区 | 国产熟女午夜一区二区三区| 看黄色毛片网站| 国产亚洲精品第一综合不卡| 亚洲色图 男人天堂 中文字幕| 精品日产1卡2卡| 两个人看的免费小视频| 一边摸一边抽搐一进一出视频| 在线永久观看黄色视频| 亚洲黑人精品在线| 精品久久久久久成人av| 成人特级黄色片久久久久久久| АⅤ资源中文在线天堂| 18禁黄网站禁片午夜丰满| 中文字幕久久专区| 亚洲三区欧美一区| 国产一区在线观看成人免费| 午夜福利视频1000在线观看 | 老司机在亚洲福利影院| 天天躁狠狠躁夜夜躁狠狠躁| 亚洲av日韩精品久久久久久密| 大型av网站在线播放| 久9热在线精品视频| 久久热在线av| 女同久久另类99精品国产91| 97人妻精品一区二区三区麻豆 | 精品乱码久久久久久99久播| 最近最新免费中文字幕在线| 精品欧美一区二区三区在线| 国产熟女xx| 亚洲成人精品中文字幕电影| 淫秽高清视频在线观看| 国产成人欧美在线观看| 亚洲最大成人中文| 国产精品久久久人人做人人爽| 亚洲性夜色夜夜综合| 村上凉子中文字幕在线| 亚洲精华国产精华精| 精品午夜福利视频在线观看一区| 黑人巨大精品欧美一区二区mp4| 黄色视频不卡| 欧美色欧美亚洲另类二区 | 久久精品影院6| 国产亚洲欧美98| 国产亚洲av高清不卡| 国产精品精品国产色婷婷| 性欧美人与动物交配| 激情在线观看视频在线高清| 久久久久久大精品| 老司机福利观看| 身体一侧抽搐| 久久久久久国产a免费观看| 亚洲成国产人片在线观看| 黄片播放在线免费| 欧美亚洲日本最大视频资源| 国产91精品成人一区二区三区| 久久九九热精品免费| 波多野结衣巨乳人妻| 亚洲美女黄片视频| 少妇裸体淫交视频免费看高清 | 亚洲国产精品999在线| 国产伦人伦偷精品视频| 美女扒开内裤让男人捅视频| 亚洲中文字幕日韩| 久久久久久国产a免费观看| 大码成人一级视频| 99久久久亚洲精品蜜臀av| 18禁裸乳无遮挡免费网站照片 | 一区二区三区高清视频在线| 亚洲一卡2卡3卡4卡5卡精品中文| 丝袜在线中文字幕| 69精品国产乱码久久久| 激情视频va一区二区三区| a在线观看视频网站| 麻豆av在线久日| 欧美乱码精品一区二区三区| 国产一区二区三区在线臀色熟女| 日日爽夜夜爽网站| 日本精品一区二区三区蜜桃| 9色porny在线观看| 午夜免费观看网址| xxx96com| 亚洲熟妇中文字幕五十中出| 亚洲在线自拍视频| 老汉色av国产亚洲站长工具| 99精品在免费线老司机午夜| 久久久久久国产a免费观看| 国产精品野战在线观看| 亚洲av日韩精品久久久久久密| 可以在线观看毛片的网站| 亚洲久久久国产精品| 黄色a级毛片大全视频| 日本精品一区二区三区蜜桃| 免费观看精品视频网站| 国产精品九九99| 一级毛片精品| 亚洲无线在线观看| 黑人操中国人逼视频| 免费高清在线观看日韩| 国产亚洲精品一区二区www| 亚洲性夜色夜夜综合| 搡老熟女国产l中国老女人| 亚洲狠狠婷婷综合久久图片| 久久精品国产综合久久久| 久久久精品欧美日韩精品| 精品国产国语对白av| 大陆偷拍与自拍| 香蕉久久夜色| 国产高清有码在线观看视频 | 国产精品久久电影中文字幕| 国产主播在线观看一区二区| 亚洲欧洲精品一区二区精品久久久| 婷婷六月久久综合丁香| 久久久久精品国产欧美久久久| 国产欧美日韩一区二区三| 欧美大码av| 亚洲激情在线av| 欧美久久黑人一区二区| 变态另类丝袜制服| 麻豆一二三区av精品| 国产精品日韩av在线免费观看 | 亚洲黑人精品在线| 亚洲五月色婷婷综合| 男人的好看免费观看在线视频 | 欧美日韩精品网址| 18禁裸乳无遮挡免费网站照片 | 91成人精品电影| 桃色一区二区三区在线观看| 日本欧美视频一区| 丰满人妻熟妇乱又伦精品不卡| avwww免费| 国内久久婷婷六月综合欲色啪| 少妇被粗大的猛进出69影院| 91老司机精品| 久久精品影院6| 看片在线看免费视频| 成人免费观看视频高清| 亚洲专区字幕在线| 久久亚洲真实| 精品国产超薄肉色丝袜足j| 黄片大片在线免费观看| ponron亚洲| 免费在线观看亚洲国产| 日韩成人在线观看一区二区三区| 美女扒开内裤让男人捅视频| 精品乱码久久久久久99久播| 99精品在免费线老司机午夜| 成在线人永久免费视频| 国产精品国产高清国产av| 麻豆av在线久日| 精品一区二区三区视频在线观看免费| 操出白浆在线播放| 亚洲欧美日韩无卡精品| 国内久久婷婷六月综合欲色啪| 精品乱码久久久久久99久播| 国产1区2区3区精品| 国产精品精品国产色婷婷| 亚洲自偷自拍图片 自拍| 亚洲aⅴ乱码一区二区在线播放 | 欧美日韩精品网址| 两性夫妻黄色片| 999久久久国产精品视频| 国产精品秋霞免费鲁丝片| 精品久久久久久,| 国产精品久久电影中文字幕| 男女床上黄色一级片免费看| 国产亚洲精品综合一区在线观看 | 亚洲精品国产精品久久久不卡| 日韩欧美国产在线观看| 日韩欧美国产在线观看| 国产精品亚洲av一区麻豆| 黄网站色视频无遮挡免费观看| 色播亚洲综合网| 久久精品亚洲精品国产色婷小说| 国产黄a三级三级三级人| 制服人妻中文乱码| 黑人欧美特级aaaaaa片| 国产精品久久久久久亚洲av鲁大| 禁无遮挡网站| 一边摸一边做爽爽视频免费| 欧美在线一区亚洲| 色播在线永久视频| 国产精品九九99| 淫妇啪啪啪对白视频| 两个人免费观看高清视频| 国产片内射在线| 美女高潮喷水抽搐中文字幕| 99久久99久久久精品蜜桃| 久久人妻熟女aⅴ| 一二三四在线观看免费中文在| 香蕉国产在线看| 18禁观看日本| 久久草成人影院| 久热这里只有精品99| 50天的宝宝边吃奶边哭怎么回事| 大陆偷拍与自拍| 亚洲av五月六月丁香网| 欧美乱妇无乱码| 久久久久亚洲av毛片大全| 欧美一级毛片孕妇| 亚洲男人天堂网一区| 在线国产一区二区在线| 啪啪无遮挡十八禁网站| 亚洲国产欧美日韩在线播放| 国产区一区二久久| 欧美午夜高清在线| 一区二区三区国产精品乱码| 两个人视频免费观看高清| 亚洲中文字幕日韩| 性少妇av在线| 国产高清激情床上av| 人妻丰满熟妇av一区二区三区| 大陆偷拍与自拍| av免费在线观看网站| 如日韩欧美国产精品一区二区三区| 人人妻,人人澡人人爽秒播| 精品久久久久久成人av| 久久久水蜜桃国产精品网| 中文字幕精品免费在线观看视频| 久久久久九九精品影院| 亚洲伊人色综图| 国产精品久久电影中文字幕| 亚洲熟妇熟女久久| 亚洲自偷自拍图片 自拍| 日韩大码丰满熟妇| 亚洲av成人不卡在线观看播放网| 法律面前人人平等表现在哪些方面| 亚洲精品美女久久久久99蜜臀| av电影中文网址| 亚洲国产欧美一区二区综合| 18禁观看日本| 国产成人啪精品午夜网站| 国产精品一区二区三区四区久久 | 真人做人爱边吃奶动态| cao死你这个sao货| 99久久久亚洲精品蜜臀av| 可以免费在线观看a视频的电影网站| 国产精品久久久人人做人人爽| 国产精品野战在线观看| 国产精品久久久久久人妻精品电影| 国产成人欧美| 91麻豆av在线| 久久精品影院6| 欧美性长视频在线观看| 精品欧美国产一区二区三| 久久国产乱子伦精品免费另类| 91大片在线观看| 久久精品人人爽人人爽视色| 免费无遮挡裸体视频| 黄色视频不卡| 日韩精品中文字幕看吧| 真人一进一出gif抽搐免费| 欧美av亚洲av综合av国产av| 国产成人av激情在线播放| 91av网站免费观看| 精品国产超薄肉色丝袜足j| 国产亚洲欧美98| 宅男免费午夜| 午夜福利影视在线免费观看| 禁无遮挡网站| 国产精品久久久av美女十八| 国内精品久久久久精免费| 男女之事视频高清在线观看| 久久婷婷成人综合色麻豆| 91麻豆av在线| 变态另类丝袜制服| 成年版毛片免费区| 国产成人av教育| 久久婷婷人人爽人人干人人爱 | 亚洲国产毛片av蜜桃av| 99精品久久久久人妻精品| av天堂在线播放| 国产精品秋霞免费鲁丝片| 老司机靠b影院| 麻豆国产av国片精品| 久久婷婷成人综合色麻豆| 久久九九热精品免费| 免费在线观看影片大全网站| 99精品久久久久人妻精品| 日韩精品青青久久久久久| 亚洲成a人片在线一区二区| 在线视频色国产色| 午夜久久久在线观看| av在线播放免费不卡| 欧美国产日韩亚洲一区| 国产精品,欧美在线| 国产激情欧美一区二区| 嫩草影院精品99| 亚洲最大成人中文| 免费看十八禁软件| 欧美亚洲日本最大视频资源| 女警被强在线播放| 人人妻人人澡欧美一区二区 | 波多野结衣av一区二区av| 老司机福利观看| 丰满的人妻完整版| 欧美最黄视频在线播放免费| 国产精品久久久久久人妻精品电影| 91字幕亚洲| 亚洲欧洲精品一区二区精品久久久| av在线天堂中文字幕| 国产一级毛片七仙女欲春2 | 欧美日韩精品网址| x7x7x7水蜜桃| 深夜精品福利| 曰老女人黄片| 极品人妻少妇av视频| 99热只有精品国产| 久久久久国产精品人妻aⅴ院| 久99久视频精品免费| 日韩欧美一区二区三区在线观看| 国产熟女xx| 午夜精品在线福利| 亚洲片人在线观看| 咕卡用的链子| 久9热在线精品视频| 婷婷六月久久综合丁香| 精品久久久精品久久久| 国产欧美日韩一区二区精品| 日韩欧美一区视频在线观看| 麻豆一二三区av精品| 一本久久中文字幕| 欧美av亚洲av综合av国产av| 可以在线观看毛片的网站| 老鸭窝网址在线观看| 久久久久亚洲av毛片大全| 久久热在线av| 免费人成视频x8x8入口观看| 亚洲精品在线观看二区| 高潮久久久久久久久久久不卡| 亚洲国产精品成人综合色| av中文乱码字幕在线| 日韩大尺度精品在线看网址 | 亚洲专区中文字幕在线| 在线播放国产精品三级| 日韩欧美免费精品| 日本免费一区二区三区高清不卡 | 女性生殖器流出的白浆| 黄网站色视频无遮挡免费观看| 精品日产1卡2卡| 国产精品自产拍在线观看55亚洲| 日韩中文字幕欧美一区二区| 欧美另类亚洲清纯唯美| av在线天堂中文字幕|