Line-Based Region Growing Image Segmentation for Mobile Device Applications
Line-Based Region Growing Image Segmentation for Mobile Device Applications
复制标题
DOI：
10.11371/iieej.41.7
复制
发表时间：
2012-01
期刊：
影响因子：
--
通讯作者：
Bo Yu;L. Diago;M. Savchenko;I. Hagiwara
中科院分区：
文献类型：
--
作者：
Bo Yu;L. Diago;M. Savchenko;I. Hagiwara
< 摘要> 最近，智能手机中开发了多种利用图像处理技术进行物体检测的应用程序。图像分割是图像处理技术的主要组成部分，是指将数字图像分割成多个片段的过程。有许多不同的图像分割算法在桌面平台上显示出良好的性能。在本文中，我们提出了一种新的基于种子的区域生长分割算法，该算法由基于行的数据结构支持，可以减少分割过程中的存储空间和计算时间，并消除以前算法的缺点。为了处理现实世界图像区域边界的不确定性，通过色差评估的自动阈值选择和边缘生长侵蚀操作的组合来集成区域生长和边缘提取的结果，从而提供更准确的图像分割。利用当前手机提供的图像处理技术，本文提出了一种新的移动应用程序，用于测量建筑物墙壁以进行修复。建筑面积测量应用的实现和实验结果证明了该算法的效率。
< Summary> Recently, several applications that use image processing technology for object detection have been developed in smart phones. As a main part of image processing technology, image segmentation refers to the process of partitioning a digital image into multiple segments. There are many different image segmentation algorithms that show the good performance on desktop platforms. In this paper, we propose a new seed-based region-growing segmentation algorithm supported by a line-based data structure that allows decreasing the storage space and computation time during the segmentation process and eliminating the drawbacks of the previous algorithms. To deal with uncertainty in region boundaries of real world images, the results of region growing and edge extraction are integrated by a combination of automatic threshold selection for color difference evaluation and edge grow-erode operations that provide more accurate image segmentation. Taking advantage of image processing technologies offered by current cellular phones, this paper presents a new mobile application for measuring building walls for restoration. The implementation and experimental results of the building area measuring application demonstrate the efficiency of the proposed algorithm.