Radial variation in sap flux density (SFD) as a function of sapwood thickness is of importance in accurately estimating sap flux through sapwood area which, in turn, decides the precision of heat pulse application. Ho...Radial variation in sap flux density (SFD) as a function of sapwood thickness is of importance in accurately estimating sap flux through sapwood area which, in turn, decides the precision of heat pulse application. However, until now, only a few studies have evaluated the magnitude and significance of sampling errors associated with radial gradients in SFD, which were based on the small monitoring measurement data from a few trees. Based on one year of heat pulse observation of two 3 - 4 years old Eucalyptus urophylla S. T.,P Blake plantations in Leizhou Peninsula, Guangdong Province, China, a way of data processing was developed to treat with the lots of SFD data measured from 39 trees. It was found that the radial variation in SFD as a function of sapwood thickness in the two eucalyptus plantation sites could be expressed as y = 3. 667 5x(3) - 7.295 5x(2) + 3.682 6x + 0. 567 4 (R-2 = 0. 939 1, n = 80, P = 0.01), where y is the ratio of SFD of a sensor to the average of four data in different depths, x is the ratio of a sensor depth to tire radial sapwood thickness. It was the same (as in the following equation) in Jijia site, y = 5.006 2x(3) - 9.116 1x(2) + 4. 454 4x + 0.463 4 (R-2 = 0. 806 9, n = 72, P = 0.01) in Hetou site. From cambium to heartwood, SFD showed some increases at first and then decreases continuously. However, because die trees were very young, the maximum SFD was only 0. 33 - 0. 36 times more than the minimum.展开更多
代数码书作为一种流行的固定码书结构,其搜索方法直接影响解码语音质量及计算复杂度。本文提出了一种代数码书分级分段优化搜索GSOS(Graded and Subsection Optimization Search)方法。GSOS方法融合了脉冲替代法、分段搜索及分级优化方...代数码书作为一种流行的固定码书结构,其搜索方法直接影响解码语音质量及计算复杂度。本文提出了一种代数码书分级分段优化搜索GSOS(Graded and Subsection Optimization Search)方法。GSOS方法融合了脉冲替代法、分段搜索及分级优化方法的优点,通过创建优质的初始码书,达到提高初始码书质量的目的,并将码书矢量的乘法运算用分段子码书的加法运算替代,同时将脉冲按贡献不同进行分级,由此提高脉冲替代优化的搜索效率;将该方法用于自适应多速率宽带语音编码器AMR-WB固定码书搜索阶段,实验结果表明,所提出的搜索方法可使固定码书搜索计算量降为深度优先树搜索方法的13.75%,但解码语音质量只降低了4.01%,主观听觉感受基本感觉不出差异。展开更多
文摘图像特征是基于内容的图像检索(Content-based image retrieval,CBIR)的关键,大部分使用的手工特征难以有效地表示乳腺肿块的特征,底层特征与高层语义之间存在语义鸿沟。为了提高CBIR的检索性能,本文采用深度学习来提取图像的高层语义特征。由于乳腺X线图像的深度卷积特征在空间和特征维度上存在一定的冗余和噪声,本文在词汇树和倒排文件的基础上,对深度特征的空间和语义进行优化,构建了两种不同的深度语义树。为了充分发挥深度卷积特征的识别能力,根据乳腺图像深度特征的局部特性对树节点的权重进行细化,提出了两种节点加权方法,得到了更好的检索结果。本文从乳腺X线图像数据库(Digital database for screening mammography,DDSM)中提取了2200个感兴趣区域(Region of interest,ROIs)作为数据集,实验结果表明,该方法能够有效提高感兴趣肿块区域的检索精度和分类准确率,并且具有良好的可扩展性。
文摘Radial variation in sap flux density (SFD) as a function of sapwood thickness is of importance in accurately estimating sap flux through sapwood area which, in turn, decides the precision of heat pulse application. However, until now, only a few studies have evaluated the magnitude and significance of sampling errors associated with radial gradients in SFD, which were based on the small monitoring measurement data from a few trees. Based on one year of heat pulse observation of two 3 - 4 years old Eucalyptus urophylla S. T.,P Blake plantations in Leizhou Peninsula, Guangdong Province, China, a way of data processing was developed to treat with the lots of SFD data measured from 39 trees. It was found that the radial variation in SFD as a function of sapwood thickness in the two eucalyptus plantation sites could be expressed as y = 3. 667 5x(3) - 7.295 5x(2) + 3.682 6x + 0. 567 4 (R-2 = 0. 939 1, n = 80, P = 0.01), where y is the ratio of SFD of a sensor to the average of four data in different depths, x is the ratio of a sensor depth to tire radial sapwood thickness. It was the same (as in the following equation) in Jijia site, y = 5.006 2x(3) - 9.116 1x(2) + 4. 454 4x + 0.463 4 (R-2 = 0. 806 9, n = 72, P = 0.01) in Hetou site. From cambium to heartwood, SFD showed some increases at first and then decreases continuously. However, because die trees were very young, the maximum SFD was only 0. 33 - 0. 36 times more than the minimum.
文摘代数码书作为一种流行的固定码书结构,其搜索方法直接影响解码语音质量及计算复杂度。本文提出了一种代数码书分级分段优化搜索GSOS(Graded and Subsection Optimization Search)方法。GSOS方法融合了脉冲替代法、分段搜索及分级优化方法的优点,通过创建优质的初始码书,达到提高初始码书质量的目的,并将码书矢量的乘法运算用分段子码书的加法运算替代,同时将脉冲按贡献不同进行分级,由此提高脉冲替代优化的搜索效率;将该方法用于自适应多速率宽带语音编码器AMR-WB固定码书搜索阶段,实验结果表明,所提出的搜索方法可使固定码书搜索计算量降为深度优先树搜索方法的13.75%,但解码语音质量只降低了4.01%,主观听觉感受基本感觉不出差异。
