Abstract:本文从傅里叶变换（ＦＴ）的角度引入连续和离散自适应于波变换（ＷＴ），作为应用，在数学上比较子耳，眼功能和连续ＷＴ，以给人工神经网络（ＡＮＮ）提供多分辨率分析（ＭＲＡ）并行输入，这种七妙的预处理可以实现特征保持的数据压缩，避免过训练或地拟合对ＡＮＮ概括和抽象能力的影响。这可以用来解释嘈杂鸡尾酒会效应。一个例子是，对Ｎ个数据要求有Ｎ阶阶复杂性快速数据压缩的多波段红外图像，引入离散ＷＴ作为完备，正交和归

Abstract:A general complex neural unit integrated with synaptic connection and lateral interaction is proposed. Various kinds of connections can be generated automatically by using different routing codes contained in the input stimuli. Based on the mixed negabinary system, an inner-product algorithm is suggested to perform complex matrix-vector multiplications without carries, signs and decimal points. An incoherent optical correlator with spatial coding is developed to execute the algorithm, thus the complex neural network is implemented optically in parallel. Numerical simulations and experiments of laterally inhibited Hopfield neural network are demonstrated.

Abstract:详细介绍了二元位相型菲涅尔透镜列阵器件（ＢＰＦＬＡＳ）的设计原理、制作过程和测试方法，研制了２４×２４元８阶ＢＰＦＬＡｓ，对０．８４μｍ波长的焦距为４６．５ｍｍ，单元尺寸为１．５×１．５ｍｍ２，衍射效率达７７．３％．有关应用实验表明，该器件具有平行光分束、多路成像及傅里叶交换功能，在列阵发光器件的光学准直，光学相关器等方面有广泛的应用前景．

Abstract:In order to discuss the characters of infrared images, the eigenhand method with the high efficiency K-L method for extracting the principal components is applied to achieve computer-based automatic recognition. The basic principles of eigenhands, system realization and the corresponding experimental results and conclusions are also given in this paper.

Abstract:The system which can quickly obtain and deal with signals of electrocardiogram(ECG) and pulsilogram (PLG) is presented. Based on traditional digital signal processing, the adaptive neural network and wavelet techniques are adopted. Experimental results show that the system has great potential in obtainting, analyzing and recognizing human health information. It is easy to obtain a miniaturized version of this system.

Abstract:A real-time optical system for fuzzy associative memory (FAM) neural network was set up. The spatial area-encoding method was used to implement optical fuzzy logic.A time-division processing approach is presented, which offers an effective way to realize neural network with a large amount of neurons. Furthermore, the feasibility of programmable interconnection is also discussed. Simulated calculation and experimental results are given.

Abstract:A fast incoherent optoelectronic hybrid neural network systenl is reported for pattern recognition of multiplex 3-D objects with error tolerance. In the first network, the positive and negative interconnections and parallel processing are used. A white-light source is used to replace He-Ne laser and a TFT liquid crystal display (LCD) panel is used as a real-time input device. In this system,coherent noise is eliminated, moreover,fast recognition and smaller size are realized.

Abstract:Based on one of the improved probabilistic neural network models-FDO neural network, the paper presents an idea of taking the effect of the & neighbors of every pixel into consideration when designing the convergent domain of the network. The activation function of the network is modified, thus endowing the network with good stability and high running speed. The recognition capabilities of FDO network and its improved version are compared in the simulation experiment. The result proves that the improved network is especially suitable for the recognition of targets with Gaussian noise.

Abstract:dge-detection, feature representation, recognition and locating algorithms of line-featured objects,taking full advantage of the knowledge about line-featured objects in the whole process, are presented in this paper. The traditional data-driving method,which is usually used in the low and intermediate level image processing, is abandoned in the process of edge detection and feature representation. The process of recognition and locating integrates the edge and gray-level information to avoid different interpretation of the same image, at the same time, to solve the recognizing difficulties caused by the discontinuous edges because of high noise in the aerial images successfully. The algorithms are of computing simplicity, speedy and of high practicality.

Abstract:The increment algorithm with limited gray-levels is improved. The training set with edge enhancement is proposed to calculate the interconnection weights. The recognition results for four kinds of aircrafts with edge enhancement, which rotate in plane, indicate that this method can reduce the gray-levels of the interconnection weights further and is suitable for optical implementation.

Abstract:The infrared photoluminescence spectra at 77K of efficient visible light-emitting porous silicon samples are studied. The perfection of silicon crystal on the surface of the porous silicon is analyzed and characterized.