摘要
运用数字图像处理技术、人脸模式识别技术与传统的数据库检索技术,融合基于文本的图像检索和基于内容的图像检索两者技术优势为一体,避开了在图像之间进行的繁琐匹配处理过程;以200人为待测样本进行检索实验,正确检索出来的为155人,完全匹配的为125人,系统的识别率为75.55%,平均检索时间小于0.1 s;实验证明,这种方法具有很强的鲁棒性,使用此方法所建立的语义人脸图像检索系统,具有快速、高效、实用的特点。
This paper uses digital image traditional database retrieval technology and technology and content-based image retrieval processing technology, human face model recognition technology, the integration of the advantages of text-based image retrieval technology to avoid tedious matching process between images. By taking 200 people as testing examples to conduct retrieval experiment, 155 people were correctly retrieved, 125 people were exactly matched ,the recognition rate of the system is 75.55 percent and average retrieval time is less than 0. 1 s. The experiment verifies that this method has strong robustness, and the semantic human face image retrieval system based on this method has the features such as rapidness, high-efficiency and practicability.
出处
《重庆工商大学学报（自然科学版）》
2014年第2期56-60,65,共6页
Journal of Chongqing Technology and Business University:Natural Science Edition
关键词
人脸图像检索
形状特征参数
语义描述
不等距搜索
human face image retrieval
shape feature parameter
semantic description
unequal distance search
