基于分形理论和神经网络的红外图像分割算法

针对实现红外图像的分割,快速准确地检测出红外序列中的人造目标的需要,结合分形特征和Ko-honen神经网络的特点,提出了一种基于分形技术的图像分割算法。该算法利用自然背景和人造目标的不同分形特征检测目标,提出了包括分形维数在内的7个红外图像特征,结合神经网络的自组织学习能力来进行图像分割。给出了算法实现的具体步骤。仿真试验结果表明,该算法能有效地实现红外图像的分割。     

基于形状描述直方图的声呐图像目标识别算法

针对水下声呐图像目标检测与识别问题,在图像规格化条件下,定义了几种不同的形状描述子,并统计计算描述子直方图,定义了几种不同的直方图匹配测度函数用以计算直方图间的相似性。分析了算法中关键参数的选取问题。仿真结果表明,基于该算法的识别率可达到98%以上,具有较强的不变性和鲁棒性。通过进一步信息融合算法可快速准确地实现水下声呐图像目标识别,具有较强的应用价值。     

多人层次单排序下权重向量的算法

旨在对ＡＨＰ中多人层次单排序下权重向量的算法作一概括。本文做了两方面的工作：其一是将单人层次单排序下,求权重向量的对数最小二乘法和最小偏差法,拓广成多人层次单排序下权重向量的算法;其二是沟通了数种算法的关系,证明了它们的等价性和相似性。     

不确定性决策中完全信息获得的事前分析

明确了规范决策分析中信息的概念,介绍了信息获得与否的价值判断法。然后主要针对不确定性决策完全信息获得的事前分析进行了讨论,给出了单状态完全信息获得的概念和一个实际分析例子;同时对完全信息与风险态度的关系进行了简单的分析。     

一种改进的基于小波变换图像边缘检测算法

本文提出了一种改进的图像边缘检测算法。它基于小波变换的方法,可以使检测处理中图像边缘相互影响所造成的边缘混叠现象减少到很小。实验结果证实了该算法的可行性。     

图像目标的可视识别与智能跟踪

本文研究图像目标的可视识别与跟踪,给出了图像目标识别印章模型和系统总体设计方案。在此基础上,采用智能跟踪算法,实现了正常跟踪、判断、预测和记忆跟踪的功能,而且跟踪方式随着环境条件的变化而变化。正常跟踪算法的图像代数描述是简明的,与算法是内在一致的。通过给出两个实时可视跟踪的实例说明在实时ＤＳＰ可视跟踪系统上算法的性能及其可行性。     

Angular glint calculation and analysis of radar targets via adaptive cross approximation algorithm

Angular glint is a significant electromagnetic （EM） scattering signature of extended radar targets. Based on the adaptive cross approximation （ACA） algorithm accelerated method of moments （MoM） and the plane incident wave assumption, the narrowband, wideband and newly developed high-resolution range profile （HRRP） based angular glint calculation formulations are derived and applied to arbitrarily shaped three-dimensional （3D） perfectly electrical y conducting （PEC） objects. In addition, the near-field angular glint is emphasized, which is of great importance for radarseeker applications. Furthermore, with the HRRP based angular glint, an approach to rigorously determine range resolution cel s which own relatively smal er angular glint is provided. Numerical results are presented with new findings to demonstrate the usefulness of the developed formulations.     

Azimuth resolution improvement for spaceborne SAR images with quasi-non-overlapped Doppler bandwidth

The azimuth resolution improvement problem is solved via a coherent combination of synthetic aperture radar （SAR） ima-ges with the quasi-non-overlapped Doppler bandwidth. Prior to the spectra combination, SAR images should be coregistered, while phase biases induced by topography, atmospheric propagation delays and baseline measurement errors should be calibrated. However, the coregistration accuracy suffers from large Doppler decorrelation caused by the quasi-non-overlapped Doppler band-width. Furthermore, the method used to estimate phase biases from interferogram of azimuth prefiltered SAR image pairs wil fail when there is no overlapped spectrum. The fringe simulation and maximum sharpness optimization are adopted to deal with the problems. Accordingly, a novel algorithm to coherently synthesize SAR images is presented. The experiment with the Terra SAR X-band （TerraSAR-X） satel ite data validates the performance of the presented method.     

Multi-label dimensionality reduction and classification with extreme learning machines

In the need of some real applications, such as text categorization and image classification, the multi-label learning gradually becomes a hot research point in recent years. Much attention has been paid to the research of multi-label classification algorithms. Considering the fact that the high dimensionality of the multi-label datasets may cause the curse of dimensionality and wil hamper the classification process, a dimensionality reduction algorithm, named multi-label kernel discriminant analysis （MLKDA）, is proposed to reduce the dimensionality of multi-label datasets. MLKDA, with the kernel trick, processes the multi-label integrally and realizes the nonlinear dimensionality reduction with the idea similar with linear discriminant analysis （LDA）. In the classification process of multi-label data, the extreme learning machine （ELM） is an efficient algorithm in the premise of good accuracy. MLKDA, combined with ELM, shows a good performance in multi-label learning experiments with several datasets. The experiments on both static data and data stream show that MLKDA outperforms multi-label dimensionality reduction via dependence maximization （MDDM） and multi-label linear discriminant analysis （MLDA） in cases of balanced datasets and stronger correlation between tags, and ELM is also a good choice for multi-label classification.     

Multi-dimensional scattering properties diagnosis system of scale aircraft model in an anechoic chamber

Aiming at the concept of ＂diagnosis＂, a simple and effective broadband radar cross section （RCS） measurement system is constructed, and some multi-dimensional scattering properties diagnosis techniques are presented based on the system. Firstly, a stepped-frequency signal is employed to achieve high range resolution, combining with a variety of signal processing tech- niques. Secondly, cross-range resolution is gained with a rotating table, and the high-resolution two-dimensional （2-D） imaging of the scale model is obtained by the microwave imaging theory. Finally, two receiving antennas with a small distance in altitude are used, and the three-dimensional （3-D） height distribution of scattering points on the scale model is extracted from the phase of images. Some typical bodies and a scale aircraft model are diagnosed in an anechoic chamber. The experimental results show that, after scaling with a metal sphere, the accurate one- dimensional （l-D） RCS pattern of the model is obtained, and it has a large dynamic range. When the bandwidth of the transmitting signal is 4 GHz, the resolution of the 2-D image can reach to 0.037 5 m. The 3-D height distribution of scattering points is given by interferometric measurement. This paper provides a feasible way to obtain high-precision scattering properties parameters of the scale aircraft model in a conventional rectangular anechoic chamber.