医学图像哈希中二值化方法的比较分析
图像与视频处理
2026-01-09 v2 计算机视觉与模式识别
信息检索
摘要
本研究评估了四种二值化方法在利用深度特征嵌入在ODIR数据集上的表现。实验结果表明,SDH方法取得了最佳性能,以仅32位代码的mAP@100为0.9184,超越了LSH、ITQ和KSH。与先前研究相比,我们的方法表现出高度竞争力:Fang等人报告了0.7528(Fundus-iSee,48位)和0.8856(ASOCT-Cataract,48位),而Wijesinghe等人实现了94.01(KVASIR,256位)。尽管使用了显著更少的位数,我们的SDH-based框架的检索准确率接近最先进的水平。这些发现表明,在所测试的方法中,SDH是最有效的方法,为医学图像检索和设备库存管理提供了实际的精度、存储和效率之间的平衡。
引用
@article{arxiv.2601.02564,
title = {Comparative Analysis of Binarization Methods For Medical Image Hashing On Odir Dataset},
author = {Nedim Muzoglu},
journal= {arXiv preprint arXiv:2601.02564},
year = {2026}
}
备注
After publication of the conference version, we identified fundamental methodological and evaluation issues that affect the validity of the reported results. These issues are intrinsic to the current work and cannot be addressed through a simple revision. Therefore, we request full withdrawal of this submission rather than replacement