宣东东,汪军.基于扩散的视觉显著目标检测[J].南华大学学报(自然科学版),2018,32(2):74~80, 86.[XUAN Dong-dong,WANG Jun.Visually Salient Target Detection Based on Diffusion[J].Journal of University of South China(Science and Technology),2018,32(2):74~80, 86.]
基于扩散的视觉显著目标检测
Visually Salient Target Detection Based on Diffusion
投稿时间:2018-03-05  
DOI:
中文关键词:  扩散矩阵  种子向量  拉普拉斯矩阵  视觉显著目标检测
英文关键词:diffusion matrix  seed vector  laplace matrix  visually salient target detection
基金项目:安徽省高校省级自然科学研究重点项目(KJ2015A311);安徽省重点研究与开发计划科技预警项目(1604d0802002)
作者单位E-mail
宣东东 安徽工程大学 计算机与信息学院, 安徽 芜湖 241000 274349401@qq.com 
汪军 安徽工程大学 计算机与信息学院, 安徽 芜湖 241000  
摘要点击次数: 681
全文下载次数: 480
中文摘要:
      针对图像背景颜色和视觉显著检测目标信息颜色相近时,显著检测结果的精度和效率不高的问题,提出了一种基于扩散的视觉显著目标检测模型.首先通过对对称矩阵的构造并丢弃常数特征向量和辨别力进行了重新组合扩散矩阵,然后对扩散矩阵中的种子向量进行重新构造,最后将重新组合的扩散矩阵和重新构造的种子向量进行扩散,得出了基于扩散的视觉显著目标检测算法.实验结果表明:该算法较之前的算法有一定的改进,在背景颜色和视觉目标信息颜色相近的情况下,该算法显著目标检测效果有了明显的提升.
英文摘要:
      When the image background and visual saliency detection’s target information are similar in color,the accuracy and efficiency of the saliency detection’s result are not high.Aiming at this problem,a diffusion-based visual saliency target detection model is proposed.Firstly,it recombined the diffusion matrix by constructing the symmetric matrix and discarding the constant eigenvector and discriminating force and the constructed symmetric matrix.Secondly,it reconstructs the seed vectors in the diffusion matrix.Finally,a diffusion-based visual saliency target detection algorithm is obtained by diffusing the recombined diffusion matrix and the reconstructed seed vector.The experimental results show that the proposed algorithm has some improvements over the previous one.In the case that background color is similar with the visual target information color,the proposed algorithm can significantly improve the detection effect.
查看全文  查看/发表评论  下载PDF阅读器
关闭