Show simple item record

dc.contributor.author曲延云
dc.contributor.author郑南宁
dc.contributor.author李翠华
dc.contributor.author袁泽剑
dc.contributor.author叶聪颖
dc.date.accessioned2017-11-14T03:19:21Z
dc.date.available2017-11-14T03:19:21Z
dc.date.issued2007-01-30
dc.identifier.citation计算机研究与发展,2007,(01):143-149
dc.identifier.issn1000-1239
dc.identifier.otherJFYZ200701019
dc.identifier.urihttps://dspace.xmu.edu.cn/handle/2288/156623
dc.description.abstract提出了一种针对自然图像中显著性建筑物的检测方法.首先,采用自底向上的注意力机制,对图像进行Haar小波分解,对得到的HL,LH分量进行平方求和,得到增强图像,然后对该增强图像在垂直方向上进行侧投影,基于得到的投影曲线进行多层阈值分割,找到显著性建筑物候选区域.进而,利用Sobel算子进行水平边缘与垂直边缘的检测,并统计较长的水平边缘与垂直边缘的数目,组成特征矢量.最后利用线性支持向量机对特征进行分类.实验证明了所提算法的有效性.
dc.description.abstractThis paper focuses on detecting salient buildings in a scenery image. A method based on bottom-up attention mechanism is proposed to detect salient buildings. Firstly, Haar wavelet decomposition is used to obtain the enhanced image which is the sum of the square of LH sub-image and HL sub-image. Secondly, the enhanced image is projected in the vertical direction to obtain the projection profile, and building candidates are separated from the background based on multi-level thresholding. Thirdly, the structure statistic features of buildings are extracted based on Sobel operator. The feature vector is formed by the number of long horizontal edges and that of vertical edges. Finally, linear support vector machines are used to classify buildings and the others. The proposed approach has been experimented on many real-world images with promising results.
dc.description.sponsorship国家自然科学基金项目(60635050,60405004);; 国家自然科学创新研究群体基金项目(60021302)
dc.language.isozh_CN
dc.subject建筑物检测
dc.subject自底向上的注意力机制
dc.subjectHaar小波分解
dc.subject支持向量机
dc.subjectbuilding detection
dc.subjectbottom-up attention mechanism
dc.subjectHaar wavelet decomposition
dc.subjectSVM
dc.title基于支持向量机的显著性建筑物检测
dc.title.alternativeSalient Building Detection Based on SVM
dc.typeArticle


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record