机载不定视觉下的图像小目标定位算法仿真  

Simulation on Image Small Target Location Algorithm under Airborne Indefinite Vision

作　　者：赵大伟[1] 陈刚[1] 

机构地区：[1]东北师范大学人文学院,吉林长春130117

出　　处：《计算机仿真》2013年第10期108-111,共4页Computer Simulation

摘　　要：研究红外图像导航定位优化问题,机载视觉设备在图像采集过程中,很难小范围控制飞机的航向,采集的图像中小目标区域受到外界航行漂移或滑动的干扰,很难形成大差异的识别特征。传统的目标识别方法多是采用静态大差异特征进行识别定位,一旦图像识别特征差异小,会造成目标区域减小,在寻优过程中算法陷入不收敛的境地,造成定位失败。提出一种改进遗传算法的机载不定视觉下的小目标定位算法。通过提取红外图像的区域特征建立一个搜索区间,运用基因位来改善搜索中特征模糊带来的小差异弊端,增强像素位置寻优性能,每次基因位的固定都能改变对边沿小差异像素的寻优空间,对寻优位置进行细化,消除特征小差异带来的定位干扰,仿真结果证明,改进方法在对传感图像进行定位时,有效地解决了小目标的特征"一致"问题,提高了定位的准确性。This paper presented a small target location algorithm based on improved genetic algorithm under air- borne indefinite vision. By extracting the regional characteristics of infrared images, a search interval was created. Then by using the gene loci to improve the disadvantage of little difference caused by fuzzy features in searching process, the optimization performance of pixe] location was enhanced. Each fixed locus can change the optimization space of the small different pixels on the edge, make an optimization location refinement, and eliminate the positio- ning interference caused by the small difference of features. The simulation results show that the improved method for sensing image positioning can effectively solve the ＂consistent＂ characteristics problem of small target, and improve the positioning accuracy.

关 键 词：红外图像 定位 改进遗传算法 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]