基于红外图像帧关联的自动阈值分割方法  

Automatic Threshold Segmentation Method Based on Infrared Image Frame Association

作　　者：苗晓孔 王春平[1] 付强[1] 

机构地区：[1]军械工程学院电子与光学工程系,河北石家庄050003

出　　处：《红外》2016年第10期41-47,共7页Infrared

基　　金：国家自然科学基金项目(61141009)

摘　　要：针对目前单帧图像阈值分割中分割易受突变影响、目标背景分割不明显以及分割效果较差等问题,提出了一种基于红外图像帧关联的自动阈值分割方法。该方法利用自动阈值分割法简单分割单帧图像,然后根据图像帧关联信息对图像进行分组处理,再对每帧图像进行权重分配,最终确定每帧图像的分割阈值,以提高分割的抗干扰性,改善分割效果。通过理论分析和实验仿真验证了该算法的有效性和可行性,并将其与其他算法进行了对比实验。实验结果表明,本文提出的分割算法的抗干扰性较强,能够将目标图像从背景中清晰地分割出来,具有更好的分割效果和更强的应用性。In the current single-frame image threshold segmentation,the segmentation is susceptible to mutations,the background segmentation is not obvious and the segmentation effectiveness is poor.To solve these problems,an automatic threshold segmentation method based on infrared image frame association is proposed.Firstly,the automatic threshold segmentation method is used to segment a singleframe image simply.Secondly,the image is processed group by group according to the frame-related information.Then,the weights are assigned to each frame.Finally,the segmentation threshold of each frame is determined to improve the anti-interference and segmentation effectiveness.The validity and feasibility of the method are verified through theoretical analysis and experimental simulation.Compared with other algorithms,the proposed method is stronger in anti-interference ability.It can segment the target image clearly from the background and has better segmentation effectiveness and applicability.

关 键 词：红外图像 阈值分割 帧关联 目标检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]