当前位置: X-MOL 学术Multimed. Tools Appl. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
An efficient approach for sub-image separation from large-scale multi-panel images using dynamic programming
Multimedia Tools and Applications ( IF 3.0 ) Pub Date : 2020-10-07 , DOI: 10.1007/s11042-020-09950-y
Mushtaq Ali , Muhammad Zubair Asghar , Amanullah Baloch

Multi-panel images are increasingly used in research and medical domains for describing complicated situations like results’ comparison in paper; or case depiction of a patient by combining all his medical images into a consolidated image. However, the content based image retrieval (CBIR) systems face the issue of performance decline in terms of poor retrieval accuracy because the individual sub-images of the multi-panel images cannot be accessed during the searching process. Representing multi-panel images in the form of sub-images is a necessary step for improving the retrieval accuracy of CBIR systems. The state-of-the-art multi-panel image segmentation approaches use recursive approach for sub-image separation, which detects the location of the sub-lines of a line in the multi-panel image appearing in its sub-images repeatedly. This characteristic of the available approaches makes the CBIR incapable to provide the intended results to the end users in real time. In this work, a line detection-based method using dynamic programming is proposed for sub-image separation, which detects the position of every line in the multi-panel image only once, instead of several times as in the case of state-of-art approaches. We evaluated the proposed method on a subset of the imageCLEFmed 2013 dataset, containing 1050 images belonging to different groups. The experimental results depict the effectiveness of the proposed method in term of generating the results quickly without losing the accuracy as compare to the state-of-the-art approaches.



中文翻译:

使用动态规划从大型多面板图像中分离子图像的有效方法

在研究和医学领域中,越来越多地使用多面板图像来描述复杂的情况,例如纸上的结果比较。通过将其所有医学图像合并为合并图像来对患者进行案例描述。但是,基于内容的图像检索(CBIR)系统由于检索精度差而面临性能下降的问题,因为在搜索过程中无法访问多面板图像的各个子图像。以子图像的形式表示多面板图像是提高CBIR系统检索精度的必要步骤。最新的多面板图像分割方法使用递归方法进行子图像分离,该方法可检测多面板图像中重复出现在其子图像中的线的子线的位置。可用方法的这一特性使CBIR无法实时向最终用户提供预期结果。在这项工作中,提出了一种使用动态编程的基于行检测的方法进行子图像分离,该方法只检测一次多行图像在多面板图像中的位置,而不是像检测状态那样多次检测位置。艺术方法。我们在imageCLEFmed 2013数据集的子集上评估了该方法,该子集包含1050个属于不同组的图像。实验结果表明,与现有方法相比,该方法在快速生成结果的同时不损失准确性的有效性。提出了一种使用动态编程的基于行检测的方法进行子图像分离,该方法只检测一次多行图像在多面板图像中的位置,而不是像现有方法那样检测几次。我们在imageCLEFmed 2013数据集的子集上评估了该方法,该子集包含1050个属于不同组的图像。实验结果表明,与现有方法相比,该方法在快速生成结果的同时不损失准确性的有效性。提出了一种使用动态编程的基于行检测的方法进行子图像分离,该方法只检测一次多行图像在多面板图像中的位置,而不是像现有方法那样检测几次。我们在imageCLEFmed 2013数据集的子集上评估了该方法,该数据集包含1050个属于不同组的图像。实验结果表明,与现有方法相比,该方法在快速生成结果的同时不损失准确性的有效性。我们在imageCLEFmed 2013数据集的子集上评估了该方法,该数据集包含1050个属于不同组的图像。实验结果表明,与现有方法相比,该方法在快速生成结果的同时不损失准确性的有效性。我们在imageCLEFmed 2013数据集的子集上评估了该方法,该数据集包含1050个属于不同组的图像。实验结果表明,与最新方法相比,该方法在快速生成结果的同时不损失准确性的有效性。

更新日期:2020-10-08
down
wechat
bug