When inverting the S-wave velocity and azimuthal anisotropy from ambient noise data, it is always to obtain the partial overlapped inversion results in contiguous different regions. Merging different data to achieve a...When inverting the S-wave velocity and azimuthal anisotropy from ambient noise data, it is always to obtain the partial overlapped inversion results in contiguous different regions. Merging different data to achieve a consistent model becomes an essential requirement. Based on the S-wave velocity and azimuthal anisotropy obtained from different contiguous regions, this paper introduces three kinds of methods for merging data. For data from different regions with partial overlapping areas, the merged results could be calculated by direct average weighting(DAW), linear dynamic weighting(LDW), and Gaussian function weighting(GFW), respectively. Data tests demonstrate that the LDW and GFW methods can effectively merge data by reasonably allocating data weights to capitalize on the data quality advantages in each zone. In particular, they can resolve the data smoothness at the boundaries of data areas, resulting in a consistent data model in larger regions. This paper presents the effective methods and valuable experiences that can be referred to as advancing data merging technology.展开更多
A fast motion estimation algorithm for variable block-size using the ＂line scan and block merge procedure＂ is proposed for airborne image compression modules.Full hardware implementation via FPGA is discussed in det...A fast motion estimation algorithm for variable block-size using the ＂line scan and block merge procedure＂ is proposed for airborne image compression modules.Full hardware implementation via FPGA is discussed in detail.The proposed pipelined architecture based on the line scan algorithm is capable of calculating the required 41 motion vectors of various size blocks supported by H.264 within a 16 × 16 block in parallel.An adaptive rate distortion cost function is used for various size block decision.The motion vectors of adjacent small blocks are merged to predict the motion vectors of larger blocks for reducing computation.Experimental results show that our proposed method has lower computational complexity than full search algorithm with slight quality decrease and little bit rate increase.Due to the high real-time processing speed it can be easily realized in hardware.展开更多
基金supported by the National Natural Science Foundation of China (Project 42330311)the Central Public-interest Scientific Institution Basal Research Fund (No. 2021IEF0103)the National Key R&D Project of China (2017YFC1500304)。
文摘When inverting the S-wave velocity and azimuthal anisotropy from ambient noise data, it is always to obtain the partial overlapped inversion results in contiguous different regions. Merging different data to achieve a consistent model becomes an essential requirement. Based on the S-wave velocity and azimuthal anisotropy obtained from different contiguous regions, this paper introduces three kinds of methods for merging data. For data from different regions with partial overlapping areas, the merged results could be calculated by direct average weighting(DAW), linear dynamic weighting(LDW), and Gaussian function weighting(GFW), respectively. Data tests demonstrate that the LDW and GFW methods can effectively merge data by reasonably allocating data weights to capitalize on the data quality advantages in each zone. In particular, they can resolve the data smoothness at the boundaries of data areas, resulting in a consistent data model in larger regions. This paper presents the effective methods and valuable experiences that can be referred to as advancing data merging technology.
基金Supported by the Aviation Science Fund of China(2009ZC15001)
文摘A fast motion estimation algorithm for variable block-size using the ＂line scan and block merge procedure＂ is proposed for airborne image compression modules.Full hardware implementation via FPGA is discussed in detail.The proposed pipelined architecture based on the line scan algorithm is capable of calculating the required 41 motion vectors of various size blocks supported by H.264 within a 16 × 16 block in parallel.An adaptive rate distortion cost function is used for various size block decision.The motion vectors of adjacent small blocks are merged to predict the motion vectors of larger blocks for reducing computation.Experimental results show that our proposed method has lower computational complexity than full search algorithm with slight quality decrease and little bit rate increase.Due to the high real-time processing speed it can be easily realized in hardware.
