摘要
为提升我国现有刑侦图像检索算法准确率,提出一种基于纹理与形状特征融合的检索算法。采用双树复小波结合四方向六参数灰度共生矩阵的方法提取24个特征值描述图像纹理,计算7个Hu不变矩作为形状特征,对纹理特征与形状特征进行融合,采用街区距离作为相似度量。对刑侦图像库进行检索实验,实验结果表明了所提算法的有效性。
Aiming at the problem that the accuracy of the existing forensic image retrieval algorithm needs to be improved,a retrieval algorithm based on fusion of texture and shape features was proposed.Two methods were used simultaneously.The dual tree complex wavelet and six parameters in each of four directions of gray level co-occurrence matrix were used to extract 24 chara-cteristic values to describe textures for an image.7 Hu invariant moments were calculated as the shape features.The texture features and shape features were fused.The city block distance was chosen as similarity measure.Experimental results show that the proposed retrieval algorithm is effective on the forensic image database.
作者
兰蓉
郭思忱
贾世英
LAN Rong;GUO Si-chen;JIA Shi-ying(School of Communication and Information Engineering,Xi’an University of Posts and Telecommunications,Xi’an 710121,China)
出处
《计算机工程与设计》
北大核心
2018年第4期1106-1110,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(61571361)
陕西省教育厅科学研究计划基金项目(16JK1709)
陕西省科技计划基金项目(2014KJXX-72)
关键词
纹理特征
形状特征
相似性度量
图像检索
刑侦图像
texture feature
shape feature
similarity measure
image retrieval
forensic image
