论文部分内容阅读
为降低加权平方误差测度下的矢量量化运算量,针对加权因子固定与不固定2种情况,分别提出了快速搜索算法。加权因子固定时,对等均值最近临搜索算法做了相应改动即可应用;加权因子随输入矢量变化时,提出了一种分裂多级等均值最近临搜索算法,算法提出了3个新的排除准则,在不同的场合下选用部分或者全部,从而有效降低码字搜索运算量。测试结果表明:分裂多级等均值最近临搜索算法能够有效降低加权平方误差测度下矢量量化的运算量,比全搜索算法能够节省约69%的运算量。
In order to reduce the vector quantization operation under the weighted square error measure, aiming at the two cases of fixed and unfixed weighting factors, a fast search algorithm is proposed respectively. When the weighting factor is fixed, the nearest-neighbor search algorithm of the peer-to-peer average value can be applied correspondingly. When the weighting factor changes with the input vector, a new nearest-neighbor search algorithm with split-level mean is proposed. The algorithm proposes three new exclusions Guidelines, in different occasions, select some or all, thus effectively reducing the amount of codeword search. The test results show that the nearest-neighbor search algorithm such as splitting multistage equal-mean can effectively reduce the computational complexity of vector quantization under the weighted square error measure, which can save about 69% of the computations compared with the full search algorithm.