[This corrects the article DOI: 10.3389/fchem.2023.1156891.].

In this paper, a general method for calculating the axial resolution (delta(z)) of optical coherence tomography (OCT) system based on the Wiener-Khinchin theorem was proposed. The power spectral density (PSD) distribution of a light source was subject to inverse Fourier transform to obtain the self-coherence function of the source, and the full width at half maximum of the function was employed to calculate the delta(z). The delta(z) values of the OCT systems illuminated by the Gaussian and non-Gaussian spectrum sources were calculated by this method, respectively, and its validity for the above sources was verified by comparing the calculated values with the nominal values of products given by the manufacturers. Furthermore, an ultra-broadband white light source was taken as an example to form a non-Gaussian spectrum by edge filtering with two filters. An experimental setup was built to measure the delta(z), and the result was very close to the calculated one with the proposed method. In conclusion, the method proposed in this paper can be applied to the delta(z) calculation of OCT systems illuminated by non-Gaussian spectrum sources, providing a basis for the parameter consideration and component selection during system design.

面向高性能计算领域的多核、众核处理器飞速发展,为了降低并行编程的难度,提高并行计算效率,数据驱动的并行编程模型成为高性能计算领域的研究热点.AceMesh是数据流驱动的、支持多核和众核异构平台的任务并行编程模型,能自动发掘结构化网格应用中存在的数据驱动的任务图并行性.但如果任务粒度划分较细,其构图过程会造成很大开销.本研究结合"申威26010"异构众核处理器的结构特点,从主、从核通信优化、内存池、无后继任务收集等方面对AceMesh构图过程进行优化,并采用航天飞行器应用中的7个热点子程序对优化效果进行测试.测试数据表明以上优化取得5倍的加速.为验证构图优化对AceMesh整体性能的提升,对航天飞行器应用分别在Acemesh和神威OpenACC的加速效果进行了测试,优化后的AceMesh加速效果约为神威OpenACC的1.5倍.

With the development of high speed communication system and the improvement of the transmission speed, Phase Lock Loop(PLL) to be the core circuit of providing precision clock has been put forward higher request, not only required to produce low jitter and low noise clock, but also demanded wide frequency range and multi-protocol supportment. As the core module of PLL, the performance of the phase noise and frequency range of voltage control oscillation(VCO) directly influence the quality of transmission clock. To reach the different transmission frequency requirement of multi-protocol, a 6.5-to-11 GHz LCVCO with wide-frequency-range and low-phase-noise is designed. Using 6-bits frequency –band-selected signal to control tuning capacitor array for relalizing the division and adjustment of the output frequency, and design optimal Kvco within each band to cover all frequency points with low phase noise. This chip is fabricated in 40 nm CMOS process, the simulation results shown that the output frequency is from 6.5 GHz to 11 GHz and the phase noise is below 107.1 dBc@1 MHz.

大气环流模式是研究全球气候变化及其成因的主要工具之一,在大规模异构众核的并行计算系统上高效地并行运行复杂的大气环流模式是一个具有挑战性的课题.Tend_lin是中国科学院大气物理研究所研发的第4代大气环流模式IAP AGCM-4中动力框架的热点过程,具有计算/通信比低的特点.面向国产大规模异构众核超算平台"神威·太湖之光",用OpenACC和AceMesh 2种不同的并行编程接口对Tend_lin进行优化.重点介绍了如何用数据驱动的任务并行编程接口AceMesh对其进行加速,介绍了计算循环和通信代码的任务并行方法,讨论了如何放松通信资源共享,对比了单层任务图和嵌套任务图下的任务映射等优化问题.测试结果表明,相比OpenACC,AceMesh在16～1024进程的不同并行配置下获得了平均2倍左右的性能提升,最后详细分析了性能收益的来源.

A method to correct the longitudinal chromatic aberration (LCA) between the imaging beam (IB) and beacon beam of ocular adaptive optics (AO) imaging system is presented, by inserting trial spherical lens, i.e. chromatic aberration corrector (CAC), on the imaging light path. The LCA of the optical system, and the quantitative relation between the CAC diopter and the IB focus shift, are determined experimentally. Then the CAC diopter required for human eye imaging can be determined by using above results combined the existing ocular LCA data. The feasibility of this method was proved by imaging results of different samples with a model eye. Finally, the method was demonstrated to significantly improve the imaging effect in in vivo retinal AO optical coherence tomography imaging of volunteers. It does not require additional burdens, is easy-to-use and suitable for various ocular AO imaging systems.

近年来硬件设计呈现出异构化的趋势,如何有效开发并行程序成为制约异构系统发展的瓶颈之一已成为业界共识.我国自主研制的"神威·太湖之光"超级计算机,采用了国产片上异构众核处理器SW26010,为了降低程序员的编程难度,同时提高软件的移植效率,作者设计并实现了支持国产SW26010众核处理器的OpenCL编译系统.该编译系统实现了OpenCL平台模型、内存模型和执行模型到SW26010众核处理器的映射与优化机制,同时生成性能良好的可执行文件.最后通过实验验证了该编译系统的正确性和有效性,典型OpenCL应用经该编译系统编译后,在中小输入规模下,性能显著优于Intel Xeon Phi,与NVIDIA GPU可比;在较大输入规模下,受限于局存SPM的容量限制,性能略低于NVIDIA GPU.

In view of the problems of low recognition accuracy and slow speed in current emotion recognition model,we designed an emotion recognition model based on kernel correlation analysis algorithm.Firstly,the current research status of emotion recognition was analyzed,and the causes of low recognition accuracy were found out.Secondly,the characteristics of emotion recognition was extracted,and the feature subset of emotion recognition was selected by kernel correlation analysis algorithm to reduce the number of feature vectors of emotion recognition.Finally,the Gauss mixture model was used to model the training set of emotion recognition,and the simulation experiments were carried out by the specific emotional data set. The experimental results show that the kernel correlation analysis algorithm can effectively remove the disadvantageous features of emotion recognition,accelerate the speed of emotion recognition, and improve the accuracy of emotion recognition.

In this paper, the author proposes to identify Land Surface Phenology from remote sensing data by using segmented Morlet wavelet transform. Land Surface Phenology is a necessary parameter for human understanding the Earth's ecological system, and the essential basis for protecting animals and plants, farming and other activities. The author finds that there are some defects in the existing methods, such as inaccurate in identifying phenology, poor at removing noise, while Morlet wavelet performs very good in cycle identification and noise removal. Therefore, the Morlet wavelet transform is used to deal with NDVI of Qinghai Lake Basin from 2003 to 2014. Then it is found that there is a case where the transformed curve is not fit with the original NDVI or the phenological period is shifted. So the author proposes an improvement method: segmenting Morlet wavelet transform, which means to divide each NDVI cycle into two sections according to the NDVI maximum, and then use Morlet wavelet transform on two segments respectively, and finally select appropriate parameter automatically. With this method, the phenology identification will be more reasonable and accurate. The authors extract the LSP parameters of Qinghai Lake Basin by segmenting Morlet wavelet transform and maximum slope, analysis of LSP parameters on time, space, and special year scales, and reveal the characteristics of Land Surface Phenology in the Qinghai Lake Basin. At the same time, it is proved that the Land Surface Phenology remote sensing recognition method based on segmented Morlet wavelet transform has improved both its accuracy and efficiency.

With the rapid development of heterogeneous system, it’s important to enhance data locality and fully utilize on-chip cache via compiler. However, classic reuse distance criteria exhibites platform-sensitive attribute in heterogeneous systems, therefore a unified reused distance calculation framework is needed for compiler to describe and optimize data locality. This paper proposes relaxed reuse distance with a unified calculation method in OpenCL programs as criteria for data layout optimization. Relaxed reuse distance is calculated with heterogeneous execution models and statistical approximation. Experiments are conducted on Intel Xeon Phi, AMD Opteron CPU, and Tilera Tile-GX36, and results show that this optimization can achieve at least 1.23x speedup on average.

地表物候是人类了解地球生态系统的必要参数,也是动植物保护、农耕等活动的重要依据.使用遥感数据进行地表物候提取为大尺度的地表物候变化监测提供了一种有效途径.研究发现目前应用最广泛的非对称性高斯函数拟合法和双Logistic函数拟合法存在一定的缺陷,尤其是在提取NDVI峰值谷值和物候周期不明显区域(如干旱地区、沙地)的物候参数时存在严重的误差,而morlet小波在分析地表年际变化的周期特征方面表现良好.本文使用morlet小波对青海湖流域2003-2014年的MODIS数据进行分析,得到地表物候在该时间段内的年度变化与趋势,进行不同区域、不同时间的差异性分析,发现青海湖流域的地表物候期整体都略有提前,但生长季的长度呈延长趋势,认为青海湖流域的地表物候整体上响应全球变暖的趋势.中部地区的生长季长度大于高海拔、高纬度区域和沙地区域,认为青海湖流域的中部地区是最适合动植物生长、活动的范围.

主流的多核/众核并行系统的计算能力并非与其资源成正比,其可扩展性与架构、资源部署等均有一定联系.除此之外,来自核心软件——操作系统以及并行运行时库的影响也是不可忽视的,硬件结构与核心软件将共同影响并行系统性能的充分发挥.因此,对这些因素的深入研究将有益于未来多核/众核结构的发展,促使其扬长避短.本文首先给出了三种真实多核/众核平台的可扩展性表现,然后采用量化的方法,分别从并行处理器结构、操作系统的角度对影响系统可扩展性的若干问题进行了深入分析.

With the recent development on heterogeneous hardware, heterogeneous parallel programming model has been widely used with the intension of simplifying programming and improving efficiency. This paper analyses latest achievements in heterogeneous parallel programming interfaces and runtime supporting systems, and solutions to new problems brought by heterogeneous architectures and various applications. In the end, some future trends in this area are discussed.

The wave aberration theory of non-symmetrical optical systems is useful for understanding the misalignment in symmetrical systems and designing of off-axis mirror systems. A theory about the third-order aberrations for sub-aperture plane symmetric optical system is developed by using the aberration theory for full-aperture axially symmetric spherical systems. This paper proves the nodal aberration theory, namely, the points in the field with zero third-order aberration will deviate from the field center except for spherical aberration. It also reveals that the nodal aberrations arise from the transformation of the aberrations in full-aperture systems. This theory can be efficiently used in non-symmetric optical system designing process.

Owing to the advantage of DSP in signal processing and function implementing,the underwater acoustic beacon can take DSP as its core part to achieve the system function.This paper mainly describes those function designs based on DSP,the operation principle of the beacon and the composition of DSP board are also briefly introduced.

A hybrid method for Chinese-English machine translation is presented. A rule-based analysis is combined with statistical data. The rule-based lexical analyzer and syntactic analyzer leave some amount of ambiguity that is resolved using a statistical approach. A hidden Markov model (HMM) is used to return a score for each part of speech. An improved probabilistic context free grammar (PCFG) is used for syntactic scoring, and a probabilistic score function is used for the semantic aspect. In addition, parameters are usually estimated poorly when the training data is sparse. Smoothing the parameters is thus important in the estimation process. A back-off procedure is used to smooth the parameters