摘要
在安检一线X射线透视系统检查中,快速有效地识别被检物质有着重要的意义。本文针对实际伪双能图像的特性,由于检测物质复杂,分类识别较难,为此结合OTSU算法和自适应区域生长算法,提出一种快速有效的新算法来提高物质分类识别准确率。在正确区分图像前景和背景的基础上,充分考虑到相邻区域的相似性,对同类物质进行统一识别。实验结果表明,新算法能够降低噪声的干扰,快速有效地实现对有机物、无机物和混合物的分类识别,能够满足安检设备的实际需求。
It is of great significance for X-ray inspecting system to sort materials inspected effectively during on-site security work.A new method,combining OTSU algorithm with region growing method,was developed to raise material sorting ability of pseudo dual-energy X-ray imaging.On the basis of distinguishing foreground and background correctly,the method uses the similar information of adjacent regions sufficiently and identifies the materials of the same category uniformly instead of traditional methods of single-point identification for picture elements.The experiment shows that the new method decreases the noise-interference effectively and sorts the materials inspected into the organic,the inorganic and the mixture,meanwhile,the fast and simple algorithm also meets the requirements of commercial equipments.
出处
《计算机仿真》
CSCD
北大核心
2012年第10期288-292,共5页
Computer Simulation
关键词
伪双能
区域分割
物质分类
Pseudo dual-energy
Region-dependent segmentation
Material sorting
作者简介
郑健(1959.7-),男(汉族),山东省蒙阴人,研究员,主要研究领域为光电子技术在安检设备中的应用。
俞俊鑫(1983.2-),男(汉族),浙江绍兴人,硕士,研究实习员,主要研究领域为信号处理及信息处理。
王强(1988.9-),男(汉族),山东临沂人,硕士研究生,主要研究领域为图像处理与模式识别。