模糊红外图像目标的人工免疫模板提取方法研究

项目名称： 模糊红外图像目标的人工免疫模板提取方法研究

项目编号： No.61272358

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 付冬梅

作者单位： 北京科技大学

项目金额： 76万元

中文摘要： 现实中有大量珍贵的模糊图像，特别是公安和军事中经常利用的隐蔽目标与目标痕迹的模糊红外图像。如何从模糊红外图像中提取清晰明确的目标，是模糊图像理论与应用中亟待解决的实际问题和关键问题。传统模板几乎都由微(差)分方程、偏微分方程或小波方程等的系数构成。为充分利用图像本身提供的各类信息，达到最佳提取效果，本项目拟以统计学习理论和张量子空间学习为理论依托，以先天性免疫与适应性免疫协同识别与作用为借鉴，研究设计新型智能模板和模板网络，使其与模糊红外图像的目标、背景及模糊边缘的特征参数相关。研究此类免疫模板和免疫模板网络的构造，计算规则和工作算法。解决从目标与背景模糊、边界不清的红外图像中提取清晰明确目标的理论与应用难题。

中文关键词： 模糊红外图像；目标提取；人工免疫模板；流形理论；张量分析

英文摘要： In reality there are many precious blurred images, especially blurred infrared images which comprise of hidden objectives and target traces are often used in public security and military. How to extract specific target from the blurred infrared images is a key scientific problem needed to be solved instantly, which comes from the reality and the blurred image theory. The traditional templates mostly constituted by the coefficients of the differential equation or the difference equation or partial differential equation or wavelet equation and so on. To get the best extracted objects from blurred infrared images, all kinds of information from blurred infrared images should be employed. Therefore, this project intends to rely on statistical learning theory and tensor subspace learning theory and to learn from the synergetic recognition and effect between innate immunity and adaptive immunity for getting information as much as possible. The new immune templates and template network, which relate to the characteristic parameters of the target and background and unclear edge in blurred infrared images, will be studied and designed in this project. The structure and calculation rules and work algorithm of the immune templates(network) will be researched in this subject. The purpose of this project is to solve the key p

英文关键词： Blurred infrared images；Object extraction；Artificial immune；Manifold theory；Tensor analysis