A Normalization Method of Moment Invariants for 3D Objects on Different Manifolds

3D objects can be stored in computer of different describing ways, such as point set, polyline, polygonal surface and Euclidean distance map. Moment invariants of different orders may have the different magnitude. A method for normalizing moments of 3D objects is proposed, which can set the values of moments of different orders roughly in the same range and be applied to different 3D data formats universally. Then accurate computation of moments for several objects is presented and experiments show that this kind of normalization is very useful for moment invariants in 3D objects analysis and recognition.     

黑白棋盘网格中的直线检测方法研究

黑白棋盘网格的直线检测在摄像机标定中具有重要的应用价值。针对黑白棋盘网格图像的特点及传统Hough变换直线检测方法中参数选择敏感的不足,在Hough变换的基础上进行了后续处理,结合最小二乘法获得黑白棋盘网格图像中的直线信息。最后,通过大量真实图像实验验证了方法的可行性,并与传统Hough变换方法及其改进方法进行了比较。     

基于TBB的Hough变换并行检测直线

Hough变换存在着运算时间长的缺点,用了并行处理这种解决海量数据计算的有效方法来减少其运行时间。该文主要研究了：利用TBB（Threading Building Blocks）这种线程构建模块在多核机上对Hough变换中可并行的部分进行并行化;实验表明这种方法对Hough变换的并行化都有很好的加速效果。     

Diffuse reflectance fourier transform infrared spectroscopy of phospholipid adsorption onto silicic acid

Diffuse reflectance Fourier transform infrared spectroscopy was used to study the mode of adsorption of phosphatidylcholine (PC) in hexane onto silicic acid (SA). PC adsorption was mainly through the charged phosphate group with minimal binding through the ester carbonyl. When the SA surface with adsorbed PC is washed with hexane, containing a small concentration of isopropanol, the desorbed PC is recovered without structural change, i.e., there is no evidence of PC hydrolysis in the adsorption process. Adsorbent misture probably promotes PC adsorption due to the increased availability of surface water hydroxyl groups for interaction with the PC phosphate groups. Isopropanol promoted PC binding by destabilizing PC reverse miscelles in solution, thus promoting its adsorption.     

Fatty acid composition of spanish shortenings with special emphasis on <Emphasis Type="Italic">trans</Emphasis> unsaturation content as determined by fourier transform infrared spectroscopy and gas chromatography

A Fourier transform infrared spectroscopic procedure was used to analyze 34 edible fats (22 shortenings and 12 vegetable margarines) as neat fats (IR_NF) to determine their total trans fatty acid (TFA) content. The sloping baseline was corrected with a reference spectrum based on a nonprocessed olive oil. The calibration was done using seven partially hydrogenated fats with an individual TFA content previously determined by the combination of gas chromatography (GC) with argentation thin-layer chromatography. Taking into account the different absorptivities of various trans isomers, different correction factors were calculated using the calibration standards (0.83 and 1.71 for single trans bonds in both diethylene and triethylene and for trans, trans-diethylene fatty acids, respectively) and applied to calculate the total TFA of samples. Moreover, the samples were converted to their methyl esters and reanalyzed following the same procedure (IR_FAME). Differences in TFA content of fats were not found when a t-test was used to compare the results obtained by IR_NF vs. either IR_FAME or GC, suggesting that IR of neat fats could be used, thus avoiding the need to prepare sample solutions in organic solvents and to prepare fatty acid methyl esters. The mean TFA content (determined by IR_NF) of a representative group of Spanish shortenings (22 samples) that varied widely in terms of fat sources, processes, and purposes (bakery, sandwiches, ice cream, coatings, chocolate coverings) was 6.55±11.40%, although more than 54% contained <3% of TFA. Fatty acid composition of shortenings by direct GC using a 100-m polar cyanopolysiloxane capillary column indicated that the mean trans-18∶2 isomer content was 0.58%, ranging from 0.9 to 3.4%. Small amounts of trans-18∶3 isomers (<0.3%) were observed in 18 of the 22 shortenings studied; the maximal value was <2%. The mean value of the fraction saturated+TFA of shortenings was high (59.95±12.73%), including two values higher than 83%.     

密集标志点集的拓扑保持变换方法*

为了解决图像变换后拓扑结构不保持的问题,提出了一种密集标志点集情况下的拓扑保持变换方法.该方法首先确定拓扑不保持的局部区域;然后确定拓扑不保持区域的方向角;最后根据该方向角确定并删除密集标志点集中引起拓扑关系不保持的标志点,从而得到拓扑关系保持的变换结果.实验证明该方法能较好地解决密集标志点情况下的拓扑不保持问题,不仅效率高,而且配准效果好.     

一种基于小波系数边信息的鲁棒水印算法

为了实施图像的版权保护,提出了一种基于小波系数幅值关系的盲水印算法。原始图像通过三级小波分解得到小波系数,算法在密钥的控制下,伪随机地选择中频子带中两块共8个系数组成一组。通过小幅度量化来加强每组数据中两块系数幅值的大小关系特征,利用这组系数的特征作为边信息和水印信息生成密钥、嵌入水印。保存密钥,并在密钥的控制下提取水印。实验结果表明,该算法不仅能有效抵抗JPEG压缩、低通滤波、高斯噪声和几何攻击,而且水印图像的PSNR可以保持在50 dB以上。     

求解缓变波导中反问题的一种步进方法

对于声学和电磁学中的大尺度反问题,基于直接方法(例如有限元或者无网格方法)的数值计算往往会导致一个巨大的线性系统,因而在计算速度和存储上并不现实.基于求解Helmholtz方程正问题的Dirichlet-to-Neuman(DtN)算子步进方法,提出了一种用于求解波在大尺度缓变波导中传播的反问题的反向基本解算子步进方法(IFOMM).     

基于概率神经网络的人群密度估计

研究图像在频域中的纹理特性,实现了在频域中的人群密度估计.首先,对视频的每一帧进行离散余弦变换(DCT),将视频图像的空域性质转化到频域中;其次,统计分析视频图像DCT变换矩阵的纹理特性,提取能描述不同密度等级的纹理描述子作为特征向量;最后,用提取出的特征向量训练概率神经网络,并进行人群密度等级估计.实验表明,该方法可以很好地进行密度估计,获得了较高识别效率.