远距离动态图像中关键目标无损挖掘算法设计  

Design of Destructive Mining Algorithm for Key Targets in Remote Dynamic Image

作　　者：卢思安[1] 刘江平[1] LU Si-an;LIU Jiang-ping(College of Computer and Information Engineering,Inner Mongolia Agricultural University,Hohhot Inner Mongolia 010018,China)

机构地区：[1]内蒙古农业大学计算机与信息工程学院,内蒙古呼和浩特010018

出　　处：《计算机仿真》2023年第12期256-260,共5页Computer Simulation

基　　金：内蒙古自治区自然科学基金项目(2022MS06026);内蒙古科技厅关键技术攻关项目(2020GG0169);内蒙古农业大学青年教师科研能力提升专项(BR 220116)。

摘　　要：由于远距离动态图像受到光照、大气干扰等因素的影响,同时关键目标与周围环境融合度高,导致难以准确地识别,使得图像关键目标提取完整度低。因此,提出一种远距离动态图像关键目标无损挖掘算法。基于三边滤波的HSV空间多尺度算法对远距离动态图像增强处理,采用直方图相似度匹配方法对动态图像的关键目标鬼影消除。根据动静态背景划分方法,对动态图像的关键目标无损挖掘。实验结果表明,所提方法的远距离动态图像关键目标无损挖掘精度高、完整性在98%以上、整体应用效果好。Due to the influence of factors such as lighting and atmospheric interference on remote dynamic ima⁃ges,as well as the high degree of fusion between key targets and the environment,it is difficult to accurately identify the key target,resulting in low integrity of target extraction.Therefore,this paper put forward a non-destructive min⁃ing algorithm for key targets in remote dynamic images.Firstly,the HSV spatial multiscale algorithm based on trilat⁃eral filtering was adopted to enhance the remote dynamic image,and then the histogram similarity matching method was used to eliminate the ghost of the key target in the dynamic image.Finally,non-destructive mining of key targets in dynamic images was achieved by using the dynamic and static background segmentation method.Experimental re⁃sults show that the proposed method has high accuracy and good overall application effect in the non-destructive min⁃ing of key targets in remote dynamic images,with an integrity of more than 98 percent.

关 键 词：远距离动态图像 关键目标无损挖掘 动静态背景划分方法 基于直方图相似度匹配方法 三边滤波 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]