Pattern recognition using discriminative feature extraction

We propose a new design method, called discriminative feature extraction for practical modular pattern recognizers. A key concept of discriminative feature extraction is the design of an overall recognizer in a manner consistent with recognition error minimization. The utility of the method is demonstrated in a Japanese vowel recognition task     

Single-mode AlGaAs-GaAs lasers using lateral confinement bynative-oxide layers

We report on results of wet oxidized narrow-stripe QW laser diodes operating in single lateral and longitudinal mode around an emission wavelength of 850 nm. Devices with an active width of 4-10 μm achieved output powers of up to 240 mW in continuous-wave (CW) operation at room temperature     

Research on pulse edge extraction by using nonlinear optical fiber-loop mirror

The output characteristics of nonlinear optical fiber-loop mirror are analyzed in detail when the pump pulses with the same wavelength are input in the both directions for recovering the clock component of the signal spectrum.It is found that the double output pulses are produced in the transmission port of the nonlinear optical fiber-loop mirror.The output pulse peaks are located in time domain at the rising and falling edges of the pump pulses.It is demonstrated that the rising and falling edges of the pump pulse can be directly extracted by this method.Through numerical simulation,the effects of therelative delay of pump pulses and the dispersion of fiber on the characteristics of output pulses are studied.By spectrum analysis,it is found that the spectrum of output pulse sequence includes the clock components of the pump pulse sequence,and a new idea is provided for all-optical clock extraction.     

Gigabit-per-second cryogenic optical link using optimizedlow-temperature AlGaAs-GaAs vertical-cavity surface-emitting lasers

We describe the design of GaAs-AlGaAs vertical-cavity surface-emitting lasers (VCSELs) that are optimized for operation at very low temperatures and the experimental demonstration of a free-space optical interconnect for cryogenic electronic systems using a VCSEL. We demonstrate high-speed modulation of the optical link at a data rate of up to 2 Gb/s at 77 K, with a very low bit-error rate of <10^-13 , and thermally stable operation is achieved over a wide range of cryogenic temperatures without laser bias current compensation. Cryogenic VCSELs with excellent lasing characteristics have been achieved over the entire temperature range from 150 K to 6 K, including high output power (22 mW), high power-conversion efficiency (32%), high slope efficiency (~100%), low threshold voltage (1.75 V) and current (1.7 mA), as well as a high-modulation bandwidth (12 GHz) for a 16 μm diameter device at 80 K     

一种改进的图像边缘提取算法

针对强干扰背景下的图像在边缘提取时难以在保留边缘细节的同时抑制噪声的情况,提出一种基于二维经验模态分解(Bidimensional Empirical Mode Decomposition,BEMD)和互信息(Mutual Information,MI)的图像边缘提取算法。首先通过BEMD对图像信息进行分解,然后对分解得到的各阶固有模态分量求出能量和能量熵值,并根据互信息准则,通过依次计算相邻分量能量熵之间的互信息值来区分高频和低频信号。最后,结合小波变换模极大值和数学形态学两种方法的优点分别对高低频信号进行边缘检测,叠加融合得到图像边缘。结果表明,此方法提取出的图像边缘连续完整,并保持了边缘的细节特征。     

Retinal vessel centerline extraction using multiscale matched filters, confidence and edge measures

Motivated by the goals of improving detection of low-contrast and narrow vessels and eliminating false detections at nonvascular structures, a new technique is presented for extracting vessels in retinal images. The core of the technique is a new likelihood ratio test that combines matched-filter responses, confidence measures and vessel boundary measures. Matched filter responses are derived in scale-space to extract vessels of widely varying widths. A vessel confidence measure is defined as a projection of a vector formed from a normalized pixel neighborhood onto a normalized ideal vessel profile. Vessel boundary measures and associated confidences are computed at potential vessel boundaries. Combined, these responses form a six-dimensional measurement vector at each pixel. A training technique is used to develop a mapping of this vector to a likelihood ratio that measures the "vesselness" at each pixel. Results comparing this vesselness measure to matched filters alone and to measures based on the Hessian of intensities show substantial improvements, both qualitatively and quantitatively. The Hessian can be used in place of the matched filter to obtain similar but less-substantial improvements or to steer the matched filter by preselecting kernel orientations. Finally, the new vesselness likelihood ratio is embedded into a vessel tracing framework, resulting in an efficient and effective vessel centerline extraction algorithm.     

All-optical routing using wavelength recognizing switches

This paper presents system considerations for optical packet-switched network using a wavelength recognizing switch (WRS) device for all-optical control and routing. Networks with the WRS device are capable of truly all-optical routing; the packet header is processed in the optical domain. This unique feature allows the self-routing of optical packets in a flexible and dynamically reconfigurable way, but introduces new challenges for the network architect. Our novel architecture combines the use of the WRS with arrayed waveguide devices in a powerful addressing mode. In this paper we explore some of the system issues, including crosstalk, noise performance, cascadability. We present experimental data on a broad-area WRS device we fabricated and assess the feasibility of an integrated version of the device     

Time division multiplexing using optical switches

Design considerations are considered for optical time-division multiplexing in terms of bandwidth requirements for the modulators and switches for a particular multiplexing scheme and the corresponding drive signals are derived using a time-domain approach. It is shown that these requirements are stricter than expected from simple considerations. Different types of multiplexing schemes are analyzed and are shown to behave differently due to varying chirping properties; this chirping is also calculated. A comparison to nonmultiplexed systems is made     

Waveguide antenna switches using p-i-n diodes

A waveguide antenna switch at L-band, using mass-produced p-i-n diodes located in the aperture, is described. Low forward loss and good isolation over a bandwidth of about 30% are achieved.     

Multiwavelength highway photonic switches using wavelength-sortingelements-design

We propose multiwavelength highway photonic switch architectures for cross-connects using the wavelength routing function of the waveguide-array-grating demultiplexer. The wavelength router is used as wavelength-sorting elements. The wavelength division multiplexed (WDM) optical signals from multiple input ports are routed to group of output ports with certain combination of wavelengths. This enables multiport WDM systems to be configured using the reduced number of wavelength demultiplexing and multiplexing elements     

Single-pole-four-throw switch using high-aspect-ratio lateral switches

A single-pole-four-throw (SP4T) switch using a high-aspect ratio lateral metal-contact micromachined switch is reported. This simplified SP4T micromachined switch is developed using deep reactive ion etching fabrication technology based on silicon-on-insulator wafer. The measurement results of the SP4T switch show an insertion loss of less than 1 dB and isolation of 30 dB from DC to 6 GHz.     

Gradient-based edge detection using nonlinear edge enhancingprefilters

This correspondence examines the use of nonlinear edge enhancers as prefilters for edge detectors. The filters are able to convert smooth edges to step edges and suppress noise simultaneously. Thus, false alarms due to noise are minimized and edge gradient estimates tend to be large and localized. This leads to significantly improved edge maps.     

An edge extraction technique for noisy images

We present an algorithm for extracting edges from noisy images. Our method uses an unsupervised learning approach for local threshold computation by means of Pearson's method for mixture density identification. We tested the technique by applying it to computer-generated images corrupted with artificial noise and to an actual Thallium-201 heart image and it is shown that the technique has potential use for noisy images.     

一种用于立体匹配的边缘提取改进算法

针对立体匹配中的边缘提取.本文提出一种简单而实用的改进方法.通过Canny算子检测最初的边缘.再利用几何信息选择关键点.根据关键点所确定的标记进行边缘连接,获得完整的边缘.该方法排除了孤立边缘,虚假边缘对边缘检测的影响,并通过引入直线分段法提取有用边缘.降低了算法的复杂度.实验结果表明,改进方法的边缘提取定位精确且连续性好.轮廓全面清晰.点、线特征突出,为立体匹配打下了良好的基础.     

Combining static and dynamic features using neural networks and edge fusion for video object extraction

Semantic object representation is an important step for digital multimedia applications such as object-based coding, content-based access and manipulations. The authors propose an image sequence segmentation scheme which provides region information for the semantic object representation of those applications. The objective is to develop a hardware-friendly segmentation algorithm by combining static and dynamic features simultaneously in one scheme. In the initial stage, a multiple feature space is transformed to one-dimensional label space by using self-organising feature map (SOFM) neural networks. The next stage is an edge fusion process in which edge information is incorporated into the neural network outputs to generate more precisely located boundaries of segmentation. The proposed algorithm differs from existing methods as follows: it can segment textured images with low-dimensional features; leads to more meaningful segmentation region boundaries; and is easier to map into hardware than existing methods. Experimental results are compared with an existing segmentation method using evaluation metrics to clarify the advantages of the proposed algorithm objectively.     

A LUMINANCE EDGE EXTRACTION METHOD USING GRADIENT OF GRAY LEVEL FOR SYNTHETIC APERTURE RADAR IMAGERY

In the multistage imaging processing for SAR digital imaging and applications ofSAR imagery,extraction of luminance edge for the SAR imageis often required.It is well studiedto extract the luminance edge for ordinary images,The methods using gradient are effective andcommonly used.Because of the serious noise of coherent speckle exists in SAR images,somepeople believe that edge extraction by using gradient for SAR imagery gives poor results.Inthis paper,we have derived a rather ideal method for the extraction of luminance edge for SARimagery with the consideration of the characteristics of SAR imagery.This method uses therelative average gradient and combines detection with tracking.     

A quartz multivibrator using two regenerative switches

A new quartz multivibrator having a 10-ns transition time is described. The frequency dependence on the supply voltage is less than 10^-6per volt. The output voltage is nearly equal to the supply voltage. The circuit can be heavily loaded.     

一种提取单像素边缘的方法

在对图像进行边缘提取时,水平和垂直的边缘一般通过简单的Soble算子就能够提取出它的单像素边缘,但是对于倾斜的或其他形状的边缘用简单的Sobel算法就很难提取出它的单像素边缘。本文提出了一种基于新的Soble算子改进算法来提取倾斜一定角度的单像素边缘。首先,通过引入衰减因子对原灰度图像进行处理,先后利用8个模板的Sobel算子得到一个边缘图,然后利用统计学的方法,统计边缘图像每一行的灰度最大值,最大值所在的位置就是单像素的位置。通过实验验证,利用本文的算法能够将倾斜一定角度的边缘提取出其单像素点。     

新的基于边缘特征的图像相关匹配方法

传统的图像相关匹配方法利用求取对应像素灰度差的累加和进行图像之间的相似性度量,或者设定阈值计算具有相近灰度值的像素点的个数进行相似性度量,由于参考图和实测图之间存在着灰度差异、模糊以及局部遮挡等,使得仅依靠图像灰度信息的算法的性能受到影响。结合图像的边界特征匹配技术,提出了一种新的图像相关匹配算法。该算法首先对图像进行边界提取并二值化,然后利用所定义的基于边缘的相似距离(ESD)对处理后的二值图像进行相似性匹配运算。试验结果表明这种算法在多种图像畸变情况下的有效性。     

基于绝对关联度的图像目标边缘提取方法

针对传统边缘检测方法对光学经纬仪实际拍摄的含噪声图像边缘检测效果不理想的问题,将灰色系统理论引入图像边缘检测,通过绝对关联分析与边缘检测算子相结合,提出一种利用二阶微分算子作为参考序列,图像像素作为比较序列,通过计算灰色绝对关联度来检测图像边缘的方法。实验结果表明,该方法与现有方法相比,边缘检测效果更好,检测的边缘清晰、连续,对噪声图像边缘的检测精度较高,抗噪声能力强。