基于深度和子树约束最小树形图的高编码效率图像删除  

High⁃coding⁃efficiency image deletion based on depth⁃and subtree⁃constrained minimum spanning tree

作　　者：沙丽娜 吴炜[2] SHA Lina;WU Wei(School of Information Engineering,Yangling Vocational&Technical College,Xianyang 712100,China;State Key Laboratory of Integrated Services Networks,Xidian University,Xi’an 710071,China)

机构地区：[1]杨凌职业技术学院信息工程分院,陕西咸阳712100 [2]西安电子科技大学空天地一体化综合业务网全国重点实验室,陕西西安710071

出　　处：《现代电子技术》2024年第9期59-65,共7页Modern Electronics Technique

基　　金：杨凌职业技术学院校内基金项目(ZK22⁃44、BG2023⁃005、JG2022003);国家自然科学基金面上项目(61471277);111计划(B08038)。

摘　　要：图像删除是指从一个压缩图像集中去除一个或多个图像,生成一个新的压缩图像集。针对当前图像删除方法存在搜寻的预测参考图像不佳,导致编码效率不足的问题,提出一种基于深度和子树约束最小树形图的高编码效率图像删除方法。该方法充分考虑所有剩余图像之间的相关性,确定最佳的预测参考。首先,提出一种新的图像分类方法,将所有图像分成需编码图像、无需编码图像和待删除图像等三类;其次,设计一种新的深度和子树约束最小树形图法,深入探究需编码图像之间以及需编码图像和无需编码图像之间的关系,构建新压缩图像集的最小树形图;最后,根据得到的最小树形图对需编码图像进行压缩,生成新的压缩图像集,实现图像删除。实验结果表明,与现有先进方法相比,所提方法取得了更高的编码效率,同时却有着相近的计算复杂度。Image deletion aims to remove one or several compressed images from a compressed image set,generating a new compressed image set.Some deficiencies exist in the existing image deletion algorithms,such as unsatisfied coding efficiency due to sub⁃optimal prediction reference images.To address the issue,a high⁃coding⁃efficiency image deletion algorithm based on depth⁃and subtree⁃constrained minimum spanning tree(DSCMST)is proposed.In the method,the correlations among all the remaining images are fully considered to determine the most appropriate prediction references.A new image categorization method is advanced,in which all the images are classified into three kinds,named images needed to be encoded,images unneeded to be encoded,and to⁃be⁃deleted images.A new DSCMST method is designed to thoroughly explore the relationships among images to be encoded and the relationships among images to be encoded and images not to be encoded,so as to establish the DSCMST of the new compressed image set.According to the obtained DSCMST,the image to be encoded is compressed to generate a new compressed image set to accomplish the image deletion.Experimental results show that the proposed algorithm achieves higher coding efficiency while having basically equivalent complexity in comparison with the existing advanced methods.

关 键 词：图像删除 压缩图像集 深度和子树约束最小树形图 编码效率 计算复杂度 预测参考图像 

分 类 号：TN919.82-34[电子电信—通信与信息系统]