基于Contourlet变换和混沌粒子群的红外小目标检测方法

项目名称： 基于Contourlet变换和混沌粒子群的红外小目标检测方法

项目编号： No.60872065

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 吴一全

作者单位： 南京航空航天大学

项目金额： 27万元

中文摘要： 红外小目标检测是精确制导的难题。本项目研究了基于Contourlet变换和混沌粒子群优化的红外小目标检测新方法： （1）研究了基于Contourlet变换和NSCT变换的红外小目标检测方法。将Contourlet变换以及NSCT变换取代小波变换并更有效地应用到红外小目标检测中，取得了优于基于小波变换的检测结果； （2）研究了基于最小一乘和混沌粒子群优化的红外小目标检测方法。建立基于最小一乘准则的红外小目标自适应背景预测模型，应用混沌粒子群算法提取最优预测参数，从而更准确地进行红外背景预测； （3）研究了基于独立分量分析和混沌粒子群优化的红外小目标检测方法。将包含复杂背景和运动小目标的图像序列视作混合信号，目标视作独立分量，应用混沌粒子群优化算法实现不同准则下的快速独立分量分析以检测出运动小目标； （4）研究了基于小波支持向量机和混沌粒子群优化的红外小目标检测方法。利用混沌粒子群优化基于小波的支持向量机进行背景预测，进一步提高了检测概率。

中文关键词： 红外小目标检测；Contourlet变换；混沌粒子群优化；支持向量机；独立分量分析

英文摘要： Infrared small target detection is a difficult problem of precision guidance.Some new methods of infrared samll target detection based on Contourlet transform and chaotic particle swarm optimization are studied in this project: (1) The infrared small target detection methods based on contourlet transform or NSCT are studied.Contourlet transform or NSCT instead of wavelet transform is used more effectively in infrared samll target detection.Better detection results are obtained compared with the method based on wavelet transform. (2) The infrared small target detection methods based on the least absolute criterion and chaotic particle swarm optimization are studied.The adaptive background prediction model is established based on the least absolute criterion.The chaotic particle swarm optimization is used to search the optimal prediction parameter.The infrared background is predicted more accurately. (3) The infrared small target detection methods based on independent component analysis and chaotic particle swarm optimization are studied.The image sequences containing complex background and small moving target are considered as mixed signal.The target is regarded as one of the independent components. The small moving target is detected by fast independent componet analysis with different criteria using chaotic particle swarm optimization. (4) The infrared small target detection methods based on wavelet, support vector machine and chaotic particle swarm optimization are studied. The background is predicted by wavelet support vector machine optimized by chaotic particle swarm optimization algorithm.The detection probability is further improved.

英文关键词： infrared small target detection; contourlet transform; chaotic particle swarm optimization; support vector machine; independent component analysis