Improving SLIC superpixel by color difference-based region merging
基于色差的区域合并改进SLIC超像素
Kefaya Sabaneh 1Muath Sabha1
作者信息
- 1. Faculty of Engineering and Information Technology, Arab American University, Jenin, Palestine
- 折叠
摘要
基于超像素的图像分割已被广泛应用于图像分割中简化后续图像处理任务。因为确定群集的数量是主观的和不同类型的图像，分割算法可以提供过分割或欠分割超像素。本文提出了一种图像分割一种利用区域合并提高SLIC超像素的方法。它的目的是改进不定义精确的超数的分割精度。色差在合并过程中，采用超像素之间的均匀性准则。使用Berkeley数据集和不同的量化性能指标来评估所提算法模型的性能。结果来自概率兰德指数(PRI)，边界查全率和欠分割误差证明了该算法的有效性具有减少聚类数量的可比分段。
Abstract
Superpixel-based segmentation has been widely used as a primary prepossessing step tosimplify the subsequent image processing tasks. Since determining the number of clusters issubjective and varies based on the type of image, the segmentation algorithm may provideover-segmented or under-segmented superpixels. This paper proposes an image segmentationmethod to improve the SLIC superpixel by region merging. It aims to improve thesegmentation accuracy without defining a precise number of superls. The color differencebetween superpixels is employed as a homogeneity criterion for the merging process. TheBerkeley dataset is used with different quantitative performance metrics to evaluate the proposedmodel’s performance. Results obtained from probabilistic rand index (PRI), boundaryrecall, and under-segmentation error proved the ability of the proposed algorithm to providecomparable segmentation with a reduced number of clusters.
Key words
Image segmentation/Image processing/SLIC/Region merging/Superpixel引用本文复制引用
出版年
2024
NETL
NSTL
Springer Nature