Superpixel-Segmentation Based on Energy Minimization and Convolution with the Geodesic Distance Kernel

IF 0.4 4区计算机科学 Q4 ENGINEERING, ELECTRICAL & ELECTRONIC

Journal of Communications Technology and Electronics Pub Date : 2024-09-11 DOI:10.1134/s1064226924700189

V. N. Karnaukhov, V. I. Kober, M. G. Mozerov, L. V. Zimina

{"title":"Superpixel-Segmentation Based on Energy Minimization and Convolution with the Geodesic Distance Kernel","authors":"V. N. Karnaukhov, V. I. Kober, M. G. Mozerov, L. V. Zimina","doi":"10.1134/s1064226924700189","DOIUrl":null,"url":null,"abstract":"<p><b>Abstract</b>—The energy minimization or maximum a posteriori probability (MAP) method is the basis for solving many computer vision problems, including the segmentation problem. However, it is assumed that the number of regions during segmentation is quite small. At the same time, in the problem of superpixel segmentation or otherwise excessive segmentation, the number of such areas exceeds 1000, which makes the computational optimization problem by the MAP method practically impossible. In this paper, we propose a solution that reduces segmentation with any number of areas to the problem of marking only nine labels. In addition, convolution with the geodesic distance kernel is used to enhance the robustness of optimization. This makes it possible to obtain single-linked superpixels at the output of the algorithm, unlike many other methods that require additional adjustments. The effectiveness of the proposed method is compared and measured by the precision-recall criteria, as well as by visual illustration.</p>","PeriodicalId":50229,"journal":{"name":"Journal of Communications Technology and Electronics","volume":null,"pages":null},"PeriodicalIF":0.4000,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Communications Technology and Electronics","FirstCategoryId":"94","ListUrlMain":"https://doi.org/10.1134/s1064226924700189","RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}

引用次数: 0

Abstract—The energy minimization or maximum a posteriori probability (MAP) method is the basis for solving many computer vision problems, including the segmentation problem. However, it is assumed that the number of regions during segmentation is quite small. At the same time, in the problem of superpixel segmentation or otherwise excessive segmentation, the number of such areas exceeds 1000, which makes the computational optimization problem by the MAP method practically impossible. In this paper, we propose a solution that reduces segmentation with any number of areas to the problem of marking only nine labels. In addition, convolution with the geodesic distance kernel is used to enhance the robustness of optimization. This makes it possible to obtain single-linked superpixels at the output of the algorithm, unlike many other methods that require additional adjustments. The effectiveness of the proposed method is compared and measured by the precision-recall criteria, as well as by visual illustration.

Abstract Image

查看原文本刊更多论文

基于能量最小化和卷积大地距离核的超像素分割技术

摘要能量最小化或最大后验概率（MAP）法是解决包括分割问题在内的许多计算机视觉问题的基础。然而，在分割过程中，假设区域的数量相当少。同时，在超像素分割或其他过度分割问题中，这类区域的数量超过 1000 个，这使得用 MAP 方法计算优化问题实际上是不可能的。在本文中，我们提出了一种解决方案，可将任意数量区域的分割简化为只标记 9 个标签的问题。此外，我们还利用大地距离核的卷积来增强优化的鲁棒性。这使得在算法输出时获得单链超像素成为可能，而不像许多其他方法需要额外的调整。通过精确度-调用标准以及直观的图示，对所提出方法的有效性进行了比较和测量。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Journal of Communications Technology and Electronics 工程技术-电信学

CiteScore

1.00

自引率

20.00%

发文量

170

审稿时长

10.5 months

期刊介绍： Journal of Communications Technology and Electronics is a journal that publishes articles on a broad spectrum of theoretical, fundamental, and applied issues of radio engineering, communication, and electron physics. It publishes original articles from the leading scientific and research centers. The journal covers all essential branches of electromagnetics, wave propagation theory, signal processing, transmission lines, telecommunications, physics of semiconductors, and physical processes in electron devices, as well as applications in biology, medicine, microelectronics, nanoelectronics, electron and ion emission, etc.