作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (21): 182-184. doi: 10.3969/j.issn.1000-3428.2011.21.062

• 人工智能及识别技术 • 上一篇    下一篇

一种改进的自然场景特征提取方法

刘 宏,普杰信   

  1. (河南科技大学电子信息工程学院,河南 洛阳 471003)
  • 收稿日期:2011-04-22 出版日期:2011-11-05 发布日期:2011-11-05
  • 作者简介:刘 宏(1981-),男,硕士研究生,主研方向:模式识别;普杰信,教授
  • 基金资助:
    河南省基础与前沿技术研究计划基金资助项目(102300 410113);河南省重点科技攻关计划基金资助项目(092102210293)

Improved Feature Extraction Method for Natural Scene

LIU Hong, PU Jie-xin   

  1. (College of Electronic and Information Engineering, Henan University of Science & Technology, Luoyang 471003, China)
  • Received:2011-04-22 Online:2011-11-05 Published:2011-11-05

摘要: 基于场景全局语义特征描述符gist的自然场景分类方法在特征提取过程中计算量较大、识别精度较低。为此,提出一种改进的特征提取方法,将3尺度的gist特征与梯度方向直方图特征相结合对场景进行描述,并利用支持向量机实现分类。实验结果表明,改进的方法加快了特征提取速度,提高了分类正确率。

关键词: 梯度方向直方图, gist特征, 自然场景, 特征提取, 自然场景分类

Abstract: Feature extraction method for natural scene has some shortcomings such as large computing costs and low identification accuracy. So this paper proposes an improved extraction method. It combines three-scale gist feature with Histograms of Oriented Gradients(HOG) to describe the scene, and uses support vector machine to realize classification. Experimental result shows that the feature computation is accelerated and the classification accuracy of natural scene image is improved by using the improved method.

Key words: Histograms of Oriented Gradients(HOG), gist feature, natural scene, feature extraction, natural scene classification

中图分类号: