摘要
图像的低层视觉特征(颜色、纹理和形状等)中包含着大量人类可感知的情感语义信息。利用纹理特征,提出一种新的索引方法-FRD(Fuzzy Recognize Degree,模糊认识度)聚类法,用来描述与情感相关联的语义图像。FRD聚类法使用三个感性的纹理特征:方向性、对比度和粗糙度,生成FRD值。从高层的情感概念进行图像检索。实验图像采用自然景观图片,验证了新方法的良好性能。
The low-level visual features（color,texture and shape） in image contain lots of emotional semantics information with human perception.In this paper a new indexing scheme is proposed,called FRD（Fuzzy Recognize Degree） clustering using the texture feature.This method has allowed us to retrieve images for emotion-assisted semantic analysis based on high-level affective concepts.For this purpose,a set of perceptual relevant features is introduced： directionality,contrast and coarseness.Experiment has been conducted to a collection of natural images,and the result shows the high performance of the proposed technique.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第33期171-174,232,共5页
Computer Engineering and Applications
