摘要
对于很大一部分待检索图像,在分割后的区域中,不仅面积较大的区域对整幅图像和人的视觉有意义,那些细腻而零散的小区域同样会对视觉产生不可忽略的影响。该文提出将图像的小区域考虑到图像检索中去,以更完整、更准确地描述图像的特征,并使用均值和方差的方法提取小区域的整体分布特征,再和其他大区域的图像特征相结合的方法进行图像检索。实验表明,该方法与只考虑大区域图像特征的检索方法相比,提高了检索的精度。
For a lot of images to be retrieved, after cut step, not only the big areas contribute to the whole image and human vision, but also the small ones do the same effect. This paper proposes to take these small areas into account in order to describe the whole image more completely and more accurately. Abstract the distribute characteristics of the small areas using mean value and variance methods, and then combines them with the big areas' characteristics, and start retrieving last. The experiment shows that this retrieval method's performance does better than that considering the big areas' characteristics only.
出处
《计算机工程》
CAS
CSCD
北大核心
2009年第7期200-202,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60573032)
武警部队军事应用研究课题基金资助项目
关键词
基于内容的图像检索
图像分割
小区域
疏密度
content based image retrieval
image division
small areas
compactness