没有合适的资源?快使用搜索试试~ 我知道了~
首页自动拟合物体轮廓的LIFS新方法
自动拟合物体轮廓的LIFS新方法
需积分: 15 0 下载量 42 浏览量
更新于2024-08-07
收藏 1.12MB PDF 举报
"使用 LIFS 对提取对象进行边界拟合" 这篇学术论文《使用 LIFS 对提取对象进行边界拟合》由Takashi Ida、Yoko Sambonsugi和Toshiaki Watanabe撰写,他们来自日本川崎的东芝公司研发中心。文章主要介绍了一种新的自动对物体粗略形状进行轮廓拟合的方法。 传统的图像处理方法往往在对象边界检测和精确轮廓获取上面临挑战,尤其是在处理手工绘制的粗略轮廓时。本文提出的方法首先利用分形编码算法提取原始图像的自相似性。这种自相似性是图像特征的重要组成部分,特别是在压缩灰度图像时,分形编码能够有效地捕捉图像的复杂结构。 接下来,论文引入了局部迭代函数系统(Local Iterated Function Systems, LIFS)的概念。LIFS 是一种迭代过程,可以用于修改表示对象区域的 alpha 映射。通过LIFS变换,可以检测到图像的自相似性参数,并通过对变换的多次迭代,使得alpha映射的轮廓逐渐逼近实际的物体边界。这种方法不仅能够处理平滑曲线,还能准确拟合尖锐的角落,从而实现高精度的轮廓提取。 实验结果表明,该方法在提供手动绘制的粗略轮廓后,能够生成高度精确的物体边界。关键词包括:分形、自相似性、对象提取、分割和图像处理。 这篇论文提出了一个创新的图像处理技术,它结合了分形理论和LIFS变换,为从图像中准确提取物体边界提供了一种有效途径。这对于图像分析、物体识别以及计算机视觉领域具有重要的理论和应用价值。通过这种方法,即使面对手工绘制的不精确轮廓,也能生成接近真实物体边界的精确模型。
资源详情
资源推荐
Boundary Fitting of Extracted Objects Using LIFS
Takashi Ida, Yoko Sambonsugi, and Toshiaki Watanabe
Research and Development Center, Toshiba Corporation, Kawasaki, Japan 210-8582
SUMMARY
This paper proposes a new method for fitting the
rough shape automatically to the contour when the rough
shape of an object is provided. The self-similarity of the
original image is extracted first in the proposed method,
using an encoding algorithm of the fractal coding for
grayscale image compression. Next, the alpha map that
indicates the object region is modified by a transformation
called Local Iterated Function Systems (LIFS). The LIFS
parameters are detected by the extracted self-similarity of
the image. Iterations of the transformation made the alpha
map contour converge to the object contour, and not only
the smooth curves but also the sharp corners were fitted
correctly in the experiments. A highly accurate contour is
available by the proposed method, simply by providing a
rough shape of the contour drawn by hand. © 2000 Scripta
Technica, Syst Comp Jpn, 31(8): 19, 2000
Key words: Fractal; self-similarity; object extrac-
tion; segmentation; image coding.
1. Introduction
The technique of object contour extraction has been
investigated and developed since it is a significant compo-
nent of image recognition, understanding, search, editing,
and so on. An international standard of coding arbitrarily
shaped images is included in MPEG-4, which was publish-
ed in 1999 [1]. From now on, the need for contour extraction
is expected to increase in the image communication and
image storage fields. Our proposal is aimed at applications
that demand highly accurate contours, such as separating
an object image from the background and synthesizing it
with another background image.
When we input the contour using a mouse device to
select the object region in an image displayed on a PC
monitor, a slight difference arises between the contour
required and the input line. In another case, that of fitting
an input rough shape to the contour using the Snakes
method [2], which extracts the contour as a smooth line, the
sharp corner top of the contour is often missed and an
extraction error remains. As a result, a high-quality synthe-
sized image cannot be obtained even though the extraction
error is only a few pixels. The objective of this paper is to
correct the rough shape automatically along the object
contour including sharp corners, thereby reducing the time
to input the contour line manually. The proposed method
can also reduce the extraction error if a slight error remains
due to the use of another method.
In the proposed method, an algorithm of fractal cod-
ing, which is one of the image compression methods, is
used with modification. Fractal coding using the Iterated
Function Systems (IFS) was applied to self-similar binary
images first [3]. IFS produces totally self-similar images:
Each part of the image is equal to a contracted image of the
whole image. Local IFS (LIFS) is an extension of IFS for
producing local self-similar images: Each part is equal to a
contracted image of another part of the image. LIFS had
been applied for contour image compression [4], and fractal
coding in which LIFS is applied to grayscale images was
proposed by Barnsley and Jacquin [57]. Subsequently,
fractal coding for grayscale images has been improved in
terms of compression efficiency and computational com-
plexity [8, 9].
The authors have also proposed some methods to
improve compression efficiency [1012] and proposed
© 2000 Scripta Technica
Systems and Computers in Japan, Vol. 31, No. 8, 2000
Translated from Denshi Joho Tsushin Gakkai Ronbunshi, Vol. J82-D-II, No. 8, August 1999, pp. 12821289
1
下载后可阅读完整内容,剩余8页未读,立即下载
weixin_38638312
- 粉丝: 6
- 资源: 957
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- 十种常见电感线圈电感量计算公式详解
- 军用车辆:CAN总线的集成与优势
- CAN总线在汽车智能换档系统中的作用与实现
- CAN总线数据超载问题及解决策略
- 汽车车身系统CAN总线设计与应用
- SAP企业需求深度剖析:财务会计与供应链的关键流程与改进策略
- CAN总线在发动机电控系统中的通信设计实践
- Spring与iBATIS整合:快速开发与比较分析
- CAN总线驱动的整车管理系统硬件设计详解
- CAN总线通讯智能节点设计与实现
- DSP实现电动汽车CAN总线通讯技术
- CAN协议网关设计:自动位速率检测与互连
- Xcode免证书调试iPad程序开发指南
- 分布式数据库查询优化算法探讨
- Win7安装VC++6.0完全指南:解决兼容性与Office冲突
- MFC实现学生信息管理系统:登录与数据库操作
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功