Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/68675
Title: Method for extracting area target of image based on fuzzytopology
Other Titles: 基于模糊拓扑的图像区域目标提取方法
Authors: Shi, WZ 
Liao, JF
Huang, CQ 
Issue Date: 28-Nov-2012
Publisher: 中华人民共和国国家知识产权局
Source: 中国专利 ZL 200810187167.8 How to cite?
Abstract: The invention relates to a method for extracting an area target of an image based on fuzzytopology, which comprises the following steps: an original image is treated in advance; an optimal threshold value T is dawn into; an object to be extracted is defined as a fuzzytopology object; the optimal threshold T is used for inducing two threshold values T1 and T2; regional division is carried out on the object to be extracted; and finally, the object is extracted out. In the method, the two threshold values are dawn into by a fuzzytopology method for determining the interior, boundary and externalof the object to be extracted, thus improving extracting quality to a certain degree; fuzzytopology connexity is carried out on boundary pixels so as to determine whether belong to the extracting range; a part duality mole (LBP) functor is dawn in in the analyzing process so as to increase judgment precision; in addition, the method for extracting the area target of the image based on fuzzytopology can significantly the extraction precision of the object, the boundary pixels of which is more difficult to be judged, thus obtaining better extraction results.
本发明涉及一种基于模糊拓扑的图像区域目标提取方法,包括如下步骤:对原始图像进行预处理;引入最优阈值T,将待提取目标定义为模糊拓扑目标;以最优阈值T诱导两个阈值T1与T2;对待提取目标进行区域划分;并最后将目标提取出来。本发明所述方法以模糊拓扑方法引入两个阈值以用于确定待提取目标的内部、边界以及外部,从而在一定程度上提高了提取质量,并且通过对边界像素进行模糊拓扑连通性分析以确定其是否属于提取范围,并在分析过程中引入局部二元模式(LBP)算符以提高判断精度。此外,本发明所述基于模糊拓扑的图像区域目标提取方法能够显著提高那些具有较高边界像素判断难度的目标提取精度,从而得到较优的提取结果。
URI: http://hdl.handle.net/10397/68675
Rights: 专利权人: The Hong Kong Polytechnic University.
Appears in Collections:Patent

Files in This Item:
File Description SizeFormat 
ZL200810187167.8.pdf383.76 kBAdobe PDFView/Open
Show full item record

Page view(s)

1
Last Week
0
Last month
Checked on Oct 15, 2017

Google ScholarTM

Check



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.