• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Artificial Intelligence IC-AI'2000 Vol.3, Jun 26-29, 2000, Las Vegas, Nevada, USA >Saliency Map Model based On Human Scan Path
【24h】

Saliency Map Model based On Human Scan Path

机译：基于人员扫描路径的显着性图模型

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

We propose a new saliency map model, which is based on the measurement of human visual scan path. Even though volitional component of attention can be superimposed onto the bottom-up task, the proposed model has been implemented by considering only the bottom-up process to get a salient map in a gray image. We use the intensity and orientation information as the bases to represent the saliency map. Each basis feature map is globally multiplied by an optimal weighting value. The final input to the saliency map is then the point - wise sum of all weighted feature maps. We use the real scan path data obtained from the Eye-Track system as a teaching signal of a neural network for generating the attended locations. Computer experimental results show that the proposed model successfully generates the selective attention locations, of which results are similar to the attended locations of human scan path.
机译：我们提出了一种新的显着性图模型，该模型基于对人类视觉扫描路径的测量。即使可以将注意力的自愿成分叠加到自下而上的任务上，也仅通过考虑自下而上的过程以获取灰色图像中的显着图来实现所提出的模型。我们以强度和方向信息为基础来表示显着性图。将每个基本特征图全局乘以最佳加权值。显着性图的最终输入是所有加权特征图的逐点求和。我们使用从Eye-Track系统获得的真实扫描路径数据作为神经网络的示教信号来生成关注位置。计算机实验结果表明，所提出的模型成功生成了选择性注意位置，其结果与人体扫描路径的关注位置相似。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号