一种混合负二进制编码的光学矩阵乘法系统

将数学中的编码方法与光学原理相结合,利用光折变晶体的四波混频效应,提出了一种可实现矩阵-矩阵的混合负二进制编码的光学系统.介绍了该系统的工作原理,分析了其工作性能.研究表明,此系统的矩阵面编码信息简单,便于扩展,可实现大规模矩阵乘法运算,对光计算的研究有一定的指导作用.     

光学全混洗网络互连函数的矩阵描述及其应用

采用自由空间光学互连技术实现的全混洗网络是数字光计算系统的基本互连网络,本文针对全混洗互连函数现有描述方法的不足,引入了描述全混洗互连函数的互连矩阵,并给出了全混洗网络互连矩阵的具体形式,为模拟或验证全混洗网络实验结果提供了理论依据。     

用细菌视紫红质膜实现多进制数字光学运算

细菌视紫红质(bR)膜对黄光和蓝光的透射具有互补调制特性，由解黄光、蓝光透射光强的速 率方程而得到黄光和蓝光的互补抑制透射特性的解析表达式和图示结果.提出了无进位和无退 位的多进制数字光计算的矩阵加、减运算模型，构造了新型光计算的基本处理模式和计算结 构，设计了多进制矩阵的加法和减法的最佳实验方案，并用bR膜实现了多进制数字光矩阵的 减法运算实验操作. 关键词： 细菌视紫红质膜 数字光计算 运算结构 光调制     

实时编码的光学逻辑处理器

本文提出了一种能实时完成二进制逻辑运算的光学并行处理系统,并给出了作为半加法器的实验结果。     

基于布尔偏振编码逻辑代数的高效冗余二进制加法的光学实现

冗余二进制法适合数字光学混合计算机.利用该法可以在常数时间内实现无进位的并行算数和,并建立内在的并行算数单元.本文在分析该方法的基础上,利用布尔偏振编码逻辑代数设计了高效光学冗余二进制加法操作系统.     

基于混合光模式阵列的自由空间编码通信

光学涡旋的产生、传输与应用是当前光学领域的研究热点之一.光学涡旋具有轨道角动量,作为一种全新的自由度,丰富了目前光通信的方式.利用面向目标的共轭对称延拓傅里叶计算全息技术,基于空间光调制器,用单束激光直接产生混合光模式阵列进行编码通信.采用由单光涡和复合光涡构成的4种易于识别的模式组成2×2混合光模式阵列,进行灰度图像的编码传输.在接收端提取混合光模式阵列图的信息并进行解码,实现零误码的灰度图像再现.以传输一幅Lena图像为例,使用2×2混合光模式阵列进行编码通信,相对于传统单光涡编码通信,其信息容量可增加4倍.该方法光路简单易行,可扩展性强,进一步拓展使用4×4混合光模式阵列进行编码通信,信息容量提升16倍.提出的混合光模式阵列编码通信方法对于提高信息传输容量具有重要价值.     

波导阵列与光纤阵列的遗传算法自动对接

设计了一种新的阵列对接方法,即将遗传算法导入光纤、光波导分支耦合器、光纤阵列的自动对接。数值仿真结果表明,常规的八通道波导分支耦合器的对接耦合能实现各通道小于9.1dB的插入损耗,最大值与最小值之差小于0.1dB,即自动对接方案具有良好的性能。     

波导阵列与光纤阵列的智能算法自动对接

开发了一种遗传算法和中心寻优算法相结合的阵列自动对接方法,试制了调芯系统,实现了光纤-光波导1×8分支耦合器-光纤阵列系统的自动对接.实验中8个通道的插入损耗平均值为-10.5 dB,均匀性小于0.5 dB,自动调芯时间小于10 min.实验结果表明该方法及其系统具有实现自动高效、低损耗对接的良好性能.     

基于光楔阵列产生光学涡旋阵列的研究

在研究光楔衍射法产生单涡旋的基础上,基于长条形光楔阵列,提出了利用光束阵列衍射产生涡旋阵列的方法.该方法要求光束阵列在平行于光楔边缘方向上的光束间距等于光束直径的整数倍.利用超精密机床采用一体化加工法加工了光楔阵列元件,验证了该方法的可行性.利用空间光调制器快速灵活调整光束阵列的优点,搭建了借助空间光调制器加载达曼光栅衍射产生所需光束阵列的实验光学系统.针对光束阵列与光楔阵列的匹配问题,研究了达曼光栅掩模图基本单元对光束阵列的调控,获得了可调结构的光束阵列.实验产生了拓扑荷一致的光学涡旋阵列,与仿真结果相一致,证明所提方法的有效性.     

高质量光学涡旋阵列的实验研究

将光学涡旋与计算全息技术相结合,提出一种高质量光学涡旋阵列的产生方法.从理论上研究了光学涡旋阵列的形成和分布特征,并模拟仿真产生涡旋光束阵列.基于面向目标的共轭对称延拓傅里叶计算全息方法编码生成光学涡旋阵列的全息图,利用单个反射式空间光调制器光电再现了与理论一致的光学涡旋阵列,并通过马赫-增德尔干涉法对生成的光学涡旋阵列进行验证.产生的高质量光学涡旋阵列提供了更复杂的结构分布和更多的可控参量,且实验光路易实现.研究结果在光学微操控、光通信等领域具有潜在应用价值.     

Performance Comparison and Evaluation of Options for Arranging Data in Digital Optical Parallel Computing

Performances of options for arranging data in the input images of digital optical computing are analyzed and evaluated on the basis of the characteristic functions that the authors formularized. The methods for arranging the data are classified into two options: the spatially expanded bit-pattern arrangement (SEPA) and the bit-slice arrangement (BSA). To analyze and evaluate the performance, the options are applied to the parallel operations based on the optical array logic that is a paradigm of digital parallel computing. Processing steps, storage capacity, and scale of the operation kernel required for performing various parallel operations are analyzed as the characteristic functions. The performances of the SEPA and the BSA are evaluated by comparing these functions. The features of each option are clarified and a guideline for choosing the optimal option is proposed to efficiently perform digital parallel optical operations.     

Optical array logic network computing: Design of a prototype system

We present the architecture of novel optoelectronic parallel computing called optical array logic network computing (OAL-NC). In the OAL-NC, electronic processing elements execute local operations, whereas an optical network processor is dedicated to global operations as well as data transfer. As a prototype of the OAL-NC, a 16x16 pixel system is designed at a logical level. Issues associated with the design task are discussed.     

两层阵列二的补码复数并行算法及其光学实现

提出了直接的二的补码阵列算法，基于混合二的补码表达方式，加法，减法和阵列乘法都能全并行地实现。直接算法克服了传统算法中存在的困难，即原理上对操作数的编码位数无限制，更有效地利用了空间带宽积，卷积后的序更直接加权求和即为两数的乘积，复数运算由包含四个区域的两层阵列完成，并进行了光学实验。     

Real-Time Reconfigurable Interconnections for Parallel Optical Processing

In this letter we describe the advantages of a dynamic optical interconnection system for parallel information processing applications. The system is based on a liquid crystal television which acts as a binary phase-only spatial light modulator. We describe example algorithms where reconfigurable interconnects would be useful and present results of several interconnection topologies which have been implemented.Presented at the International Commission for Optics Topical Meeting, Kyoto, 1994.     

Parallel Image Processing Algorithms Based on Mathematical Morphology Using a Multiple-Imaging System

Morphological techniques are applied to binary image processors in a multiple-imaging optical system and the following algorithms are proposed: performing morphological basic operations, extraction of boundary lines, detection of characteristic points in small-scale patterns, elimination of salt noises with few pixels, and smoothing of boundaries. Techniques for processing a complicated binary image are demonstrated using the algorithms and a hybrid parallel computing system with a simple optical multiple-imaging system and a personal computer.     

双路可编程光学多重成像并行二值逻辑处理器

基于多重成像原理，在本文中提出一种光学并行图像逻辑处理系统。对输入图案用两个正像和两个补偿进行编码。通过光调制器控制，可以实现逻辑操作的实时可编程。而且，该系统可以通过控制一组四个偏振片的状态实现独立的双输出逻辑通道，每一个输出通道各有十六种逻辑功能。文中也给出了实验结果。     

Simplification of Space-Variant Parallel Logic Operations Using the Temporal Method

A time-domain encoding method (temporal method) for space-variant parallel logic operations, which can execute different operations in parallel, is proposed. The temporal method is based on temporal encoding of two input patterns, temporal gating of the coded pattern, and decoding by temporal addition of the gated patterns. The first feature of the proposed method is that parallel logic operations can be performed without complex pattern transformations. The second feature is that the logical output can be directly fed to succeeding systems without specific decoding. Therefore, the logic operation system can be constructed using conventional optics and existing spatial light modulators. In order to confirm these features, an optoelectronic experimental system is constructed and space-variant parallel logic operations are performed.     

基于相机阵列的光学组件缺陷在线检测方法

利用调焦方式可以实现焦距的连续变化从而对不同物距下的光学组件进行在线检测,但是调焦过程操作复杂且对调焦位移精度要求较高,景深内光学元件缺陷无法区分,难以实现真正意义上的在线检测。因此,本文提出了基于相机阵列的光学组件缺陷在线检测方法。首先建立了相机阵列的成像模型并给出了数字重聚焦表达式以及空间分辨率的表达式。接着利用MATLAB模拟相机阵列成像过程和数字重聚焦过程。最后进行实验验证,通过二维位移台带动相机对不同物距下的多个光学元件表面缺陷进行成像获得阵列相机图像,通过数字重聚焦算法得到不同物距下的光学元件表面缺陷分布信息。实验结果表明,基于相机阵列的光学组件缺陷在线检测技术能够同时对位于景深范围内的光学组件进行在线检测。该方法在光学元件缺陷在线检测方面有着一定的应用价值。     

Cellular processing for edge detection in an optical shadow casted architecture

The principles of two-dimensional cellular processing are utilized to obtain the edges of a binary image in an optical shadow casted architecture. The processing steps, based on image logic algebra are introduced where operation kernels are set by the on/off states of four light sources. The input binary image is shifted, superimposed, then logically operated and finally decoded to obtain the edges. The whole process is controlled by a microcomputer. Experimentally simulated results are presented.     

可编程光学并行模糊逻辑门

提出了可级联的光学并行模糊逻辑门系统，十六种模糊逻辑运算可通过编程偏振半波片的状态而得到实现，语言中给出了实验结果。